{
  "version": 3,
  "sources": ["ssg:https://framerusercontent.com/modules/SGUOVximfVt6F2Q51yHz/loXzYXiF1TnCrw8drRKZ/RG6I9Jvqh-2.js"],
  "sourcesContent": ["import{jsx as e,jsxs as t}from\"react/jsx-runtime\";import{ComponentPresetsConsumer as a,Link as n}from\"framer\";import{motion as i}from\"framer-motion\";import*as o from\"react\";import r from\"https://framerusercontent.com/modules/pVk4QsoHxASnVtUBp6jr/TbhpORLndv1iOkZzyo83/CodeBlock.js\";export const richText=/*#__PURE__*/e(o.Fragment,{children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(n,{href:\"https://www.linkedin.com/in/flavia-trotolo-0baa04259/\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!0,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Flavia Trotolo\"})}),\", \",/*#__PURE__*/e(n,{href:\"https://www.walturn.com/author/hashim-hayat\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!0,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Hashim Hayat\"})}),\", \",/*#__PURE__*/e(n,{href:\"https://www.walturn.com/author/daheem-hayat\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!0,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Daheem Hayat\"})})]})});export const richText1=/*#__PURE__*/e(o.Fragment,{children:/*#__PURE__*/t(\"ul\",{children:[/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Vertical AI Agents\"}),\": Industry-specific AI trained on domain language, decision-makers, and buyer needs for tailored outreach.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"AI + Human Blend\"}),\": Human strategists guide campaigns while AI handles scale, research, and execution.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Fully Managed Model\"}),\": Throxy runs the complete outbound process\u2014no need for SDRs, CRMs, or multiple tools.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Precision Targeting\"}),\": Prioritizes relevance over volume with hyper-personalized, multi-channel messaging.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Automated Funnel\"}),\": Covers lead generation, outreach, nurturing, and meeting scheduling with continuous optimization.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Proven ROI\"}),\": Delivers six-figure pipelines, boosts deliverability by 90%, and increases conversions by 43%.\"]})})]})});export const richText2=/*#__PURE__*/t(o.Fragment,{children:[/*#__PURE__*/e(\"h2\",{children:\"Introduction\"}),/*#__PURE__*/e(\"p\",{children:'Businesses that sell to other businesses (B2B) frequently use a sales funnel. A funnel is a methodical procedure that follows a lead (a possible customer) from the point of first contact until the completion of a transaction. This typically entails hiring costly sales teams (SDRs/outreach representatives), conducting manual research, and balancing a variety of tools (CRMs, email sequencers, databases). The issue is more severe for legacy industries like manufacturing, logistics, and banking, where customers require highly individualized, reliable communication, and \"spray-and-pray\" email blasts or generic automated solutions rarely work. Throxy was developed to address this issue by automating the whole outbound funnel with vertical, sector-specific AI agents.\\xa0'}),/*#__PURE__*/t(\"p\",{children:[\"Throxy uses specialized \",/*#__PURE__*/e(n,{href:\"https://www.ycombinator.com/companies/throxy\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"vertical AI agents\"})}),\" to propel outbound sales at every level. By fusing the speed of automation with human sales ability, these AI-powered agents locate qualified leads, create customized messages, and even set up meetings.\"]}),/*#__PURE__*/t(\"p\",{children:['Instead of offering software licenses, Throxy creates a fully managed AI-driven sales system, or \"Sales-as-a-Software,\" as the founders refer to it. Without requiring the client to employ SDRs or assemble different tools, \"',/*#__PURE__*/e(n,{href:\"https://throxy.com/\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"it finds the right people, writes the right message, delivers it at the right time, and books the meeting.\"})}),\"\\\" Throxy's strategy has already produced \",/*#__PURE__*/e(n,{href:\"https://www.ycombinator.com/companies/throxy\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"six-figure sales funnels\"})}),\" for clients in difficult-to-reach countries since its inception in 2023. The operation of Throxy's AI funnel automation is explained here, along with why it performs better than conventional outbound solutions.\"]}),/*#__PURE__*/e(\"h2\",{children:\"The Challenge of Legacy Sales Funnels\"}),/*#__PURE__*/t(\"p\",{children:[\"It is helpful to comprehend why outbound sales are so difficult, particularly in conventional B2B industries, before looking at Throxy's answer. Buyers want thoughtful, pertinent communication in various industries, including manufacturing, fintech, healthcare, and others. A generic email sent by a random tool is likely to be disregarded or end up in spam. Establishing a traditional sales development team takes time and money, frequently costing over $150K before any benefits are seen. The majority of sales automation tools, on the other hand, provide general workflows and wide templates that fail to take into consideration the specifics of each vertical. When trust and accuracy are crucial, \\\"spray-and-pray doesn't work,\\\" as the founders of \",/*#__PURE__*/e(n,{href:\"https://www.ycombinator.com/companies/throxy\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Throxy\"})}),\" point out.\\xa0\"]}),/*#__PURE__*/t(\"p\",{children:['In light of this, Throxy presents itself as a solution. \"Vertical AI agents who manage the sales funnel on autopilot\" is its ',/*#__PURE__*/e(n,{href:\"https://www.ycombinator.com/companies/throxy\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"tagline\"})}),\". That is, Throxy develops industry-specific AI agents (such as a manufacturing-focused sales agent, a finance-focused agent, etc.) that are familiar with the language, priorities, and decision-makers of that industry rather than a single general AI or tool. These intelligent agents work in tandem with human control to provide Throxy with large-scale, hyper-personalized advertising. In actuality, this implies that the client needs to put in very little work because Throxy will handle all aspects of the outbound process, from lead identification to scheduling meetings.\"]}),/*#__PURE__*/e(\"p\",{children:'Throxy highlights that it\\'s \"not another sales platform,\" but rather a fresh perspective on outbound marketing. The founders refer to the business model as a \"full-stack AI firm\" because they handle your outbound marketing in addition to selling software. To \"earn the right to automate,\" this process begins with human execution (to perfect each phase) and then gradually automates what works. The end product is an AI-driven engine that is undetectable and creates actual meetings and pipelines while concealing the intricate inner workings.'}),/*#__PURE__*/e(\"h2\",{children:\"Vertical AI Agents in Sales\"}),/*#__PURE__*/e(\"p\",{children:\"Instead of being one-size-fits-all, vertical AI agents are AI tools that are tailored for a certain industry. A vertical AI agent, in contrast to general AI chatbots, is trained on the language, procedures, and duties of a particular domain (e.g., sales in manufacturing). This indicates that the sales representative is knowledgeable about the typical client journey, regulatory terminology, typical objections, and important decision-makers in that industry.\"}),/*#__PURE__*/e(\"p\",{children:\"This strategy is embodied by Throxy's agents. They are based on the idea that both strategy and technology are necessary for successful outbound in legacy sectors. According to one review, Throxy's agents function similarly to SDRs but more effectively. They don't just send out pre-made emails to contacts at random. Rather than using scraped lists, the agents use AI-powered research to identify actual decision-makers. They then create personalized, human-like outreach in the prospect's native tongue. Each message is unique and pertinent to the occasion, avoiding the impersonal tone that generic tools frequently generate.\"}),/*#__PURE__*/t(\"p\",{children:[\"Key to Throxy\u2019s success are four \",/*#__PURE__*/e(n,{href:\"https://hiretop.com/blog4/throxy-automated-outbound-sales/\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"core principles\"})}),\", distilled from their approach and the founders\u2019 insight:\"]}),/*#__PURE__*/e(\"h3\",{children:\"1. Vertical Specialization\"}),/*#__PURE__*/e(\"p\",{children:\"Throxy focuses on complicated, trust-based businesses, the areas that traditional software often overlooks. By targeting only a few verticals (manufacturing, transportation, healthcare, finance, etc.), the team achieves deeper industry expertise and more exact targeting than broad SaaS suppliers can.\"}),/*#__PURE__*/e(\"h3\",{children:\"2. AI + Human Expertise\"}),/*#__PURE__*/e(\"p\",{children:\"Throxy depends on more than just automation. AI and humans work together to create their campaigns. While human strategists set the direction and refine messaging, AI bots handle copywriting and data-crunching at scale. Despite being able to be run at machine speed, this hybrid model produces ads that feel handcrafted\"}),/*#__PURE__*/e(\"h3\",{children:\"3. Fully-Managed Execution\\xa0\"}),/*#__PURE__*/e(\"p\",{children:\"Clients are not required to hire and train SDRs, purchase software licenses, or oversee intricate sequences. The entire outbound motion is taken up by Throxy. Throxy's crew and AI system handle all of the tasks, including creating prospect lists, warming up email addresses, and calling phone numbers. The client only needs to specify their desired profile\"}),/*#__PURE__*/e(\"h3\",{children:\"4. Precision over Volume\"}),/*#__PURE__*/e(\"p\",{children:\"Throxy places a higher priority on quality engagement than flooding inboxes. Every communication is tailored for maximum relevance and credibility, and every lead is thoroughly investigated. The agents modify the channel, language, and timing to reach prospects at the ideal time. This emphasis on accuracy fosters goodwill with decision-makers and helps evade\"}),/*#__PURE__*/e(\"p\",{children:'Throxy was able to \"turn\" the conventional outbound business on its head thanks to these concepts. Instead of sending out a ton of emails, it makes use of sophisticated filters, current information, and carefully crafted content to provide every touchpoint a genuine opportunity for conversion.'}),/*#__PURE__*/e(\"h2\",{children:\"Throxy\u2019s Sales Funnel Automation Pipeline\"}),/*#__PURE__*/e(\"p\",{children:\"In practice, Throxy automates every major stage of the sales funnel for the client. The process looks something like this:\"}),/*#__PURE__*/e(\"h3\",{children:\"1. Prospect Research and Lead Generation\"}),/*#__PURE__*/e(\"p\",{children:\"Throxy starts with determining the client's Ideal Customer Profile (ICP), which includes the roles to be targeted, organization sizes, and industry specialties. It then locates high-potential leads that match this profile using its AI agents and proprietary databases. To find actual decision-makers rather than just generic titles, these agents look beyond LinkedIn profiles, extracting information from news articles, industry reports, financial filings, and even social media. A personalized list of qualifying leads is the result, which frequently contains prospects that other tools overlook.\"}),/*#__PURE__*/e(\"h3\",{children:\"2. Personalized Outreach Messaging\"}),/*#__PURE__*/e(\"p\",{children:\"Following lead identification, Throxy's AI systems craft customized messages for every potential customer. The AI creates original email copy (or call scripts) that refer to the lead's role, industry difficulties, or the company's most recent news rather than using templates that are too generic. For instance, an agent may mention in the outreach that a manufacturing company has revealed a new product. A key component of Throxy's strategy is this degree of customisation, which involves creating \\\"hand-written messaging in the buyer's language.\\\" It increases the likelihood of engagement by making sure that every communication seems as though it was written by a knowledgeable human.\\xa0\"}),/*#__PURE__*/e(\"h3\",{children:\"3. Outreach Campaign Management\"}),/*#__PURE__*/e(\"p\",{children:\"The outreach's logistical delivery is managed by Throxy. This involves scheduling email sends at the best times, warming up email domains (to ensure they have a positive reputation), and executing multi-channel campaigns (email, LinkedIn, etc.) as required. To avoid spam folders, they use email deliverability strategies that are foolproof, such as improving subject lines, reply-to addresses, sending patterns, and even A/B testing. To modify the approach, the AI also continuously tracks deliverability metrics (opens, bounces). Warm leads receive timely reminder emails or calls without the need for human intervention, thanks to the unified tech stack, which is frequently coupled with clients' CRMs like HubSpot or Salesforce.\"}),/*#__PURE__*/e(\"h3\",{children:\"4. Lead Qualification and Nurturing\"}),/*#__PURE__*/e(\"p\",{children:\"The AI agents start qualifying leads as responses start to come in. NLP interprets responses to determine whether they are requests for information, objections, or interest. Simple inquiries can be answered by the system, which can also continue to nurture lukewarm leads or forward hot leads to a human (or the client's calendar) for a call. The AI may ask intelligent follow-up questions (such, \\\"Are you interested in a product demo?\\\") or modify the message depending on prior responses because it has recorded each lead's engagement history. In essence, the AI continuously scores and moves prospects down the funnel, functioning as an enhanced SDR.\\xa0\"}),/*#__PURE__*/e(\"h3\",{children:\"5. Meeting Scheduling\"}),/*#__PURE__*/t(\"p\",{children:[\"Getting quality prospects on the client's sales team's calendar is the ultimate objective. This process is also automated by Throxy's technology. The AI agent arranges arrangements (usually through calendar links or email confirmations) to plan the appointment after a prospect consents to speak. Without doing anything, clients see scheduled meetings show up in their calendars. The algorithm \\\"\",/*#__PURE__*/e(n,{href:\"https://throxy.com/\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"does it all: finds the proper persons \u2026 and books the meeting\"})}),',\" said Throxy. By conducting meetings directly, Throxy has frequently successfully eliminated the need for an entire SDR team.\\xa0']}),/*#__PURE__*/e(\"p\",{children:'Throxy offers real-time statistics and optimization during this process. Response rates, open rates, and conversion metrics are all continuously monitored by the \"Sales Engine.\" It employs an automated feedback loop, whereby the outcomes of every campaign are fed back into the AI to improve messaging and targeting. For example, the algorithm will favor particular email subject lines in subsequent dispatches if they consistently yield higher results. The AI will find additional leads similar to a certain segment if that segment responds more. The outbound funnel will gradually become more effective thanks to this closed-loop learning, which eliminates the need for the client to manually adjust anything.\\xa0'}),/*#__PURE__*/e(\"h2\",{children:\"Benefits and Impact\"}),/*#__PURE__*/t(\"p\",{children:[\"Through the automation of these funnel steps, Throxy provides clients with measurable outcomes. It has continuously produced six-figure pipelines for businesses in its target verticals since 2023. Customers report significant increases in meeting conversion rates as well as time savings. For instance, according to an independent analysis, Throxy users report up to a \",/*#__PURE__*/e(n,{href:\"https://azariangrowthagency.com/ai-tools/throxy/\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"90% boost in email deliverability and a 43% increase in conversion rates\"})}),\". Internally, Throxy's team also validated that their AI-driven system could consistently close sales, not only showcase features, which led to quick growth (over $1.5M annual recurring revenue within months).\"]}),/*#__PURE__*/e(\"p\",{children:\"For a business owner or sales leader, the biggest benefits are clear:\"}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Cost Efficiency:\"}),' You do not have to manage several tools or pay a sizable SDR workforce. Throxy substitutes a single service for those expenses. Businesses may \"grow their effect without scaling manpower,\" the founders stress.']}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Speed to Pipeline:\"}),\" Businesses see qualified meetings scheduled weeks (or even days) after onboarding Throxy, as opposed to waiting months to ramp up an SDR team. Due to Throxy's agents' utilization of pre-existing data and established playbooks, the initial manual campaign runway is brief.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Predictable Results:\"}),\" Clients can predict the pipeline more accurately because everything is monitored and optimized. Throxy frequently establishes target meeting volumes ahead of time and then guarantees that clients will see those appointments on their calendars.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Focus on Selling: \"}),\"The client's internal staff can concentrate on closing business instead of chasing prospects when Throxy handles outreach and research. Instead of creating email lists or sending cold emails, sales representatives and founders spend their time following up on scheduled calls.\"]}),/*#__PURE__*/e(\"p\",{children:'All things considered, Throxy\\'s strategy has proven particularly beneficial in the \"sectors software forgot\"\u2014the classic, hard-to-reach businesses where traditional SaaS pipelines stall. Through Throxy, businesses can more effectively access those markets by bridging the gap between AI automation and human-level nuance.\\xa0'}),/*#__PURE__*/e(\"h2\",{children:\"A New Model: \u201CFull-Stack AI Sales\u201D\"}),/*#__PURE__*/e(\"p\",{children:'Throxy signifies a change in the way artificial intelligence is used in sales. The startup refers to what it does as \"Sales-as-a-Software\" since it offers a real service-backed solution rather than just tools. According to the founder, it\\'s about operating your outbound for you, not just selling software.'}),/*#__PURE__*/t(\"p\",{children:['Throxy adopted a more cautious strategy than early \"',/*#__PURE__*/e(n,{href:\"https://www.startupriders.com/p/full-stack-ai-companies\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"AI SDR\"})}),'\" systems that merely claimed to replace sales representatives: it began completely manually to validate every tactic before gradually automating it. They remained manual until they \"',/*#__PURE__*/e(n,{href:\"https://www.startupriders.com/p/full-stack-ai-companies\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"won the right to automate,\"})}),'\" as co-founder Pablo explains. This meant that Throxy only codified the successful elements into automated operations after early clients viewed carefully crafted advertising.']}),/*#__PURE__*/e(\"p\",{children:\"Customers frequently enjoy a smooth UX because of this mindset. Startup Riders claims that Throxy's strategy produces an invisible UX, where clients communicate via straightforward dashboards of responses and meetings, with the intricate AI research and sequencing taking place in the background. From the perspective of the user, it is similar to having a knowledgeable outbound team available to them without requiring additional work.\"}),/*#__PURE__*/e(\"h2\",{children:\"Conclusion\"}),/*#__PURE__*/e(\"p\",{children:\"A few lessons can be learned from Throxy's success as more businesses turn to AI to address sales issues. First, context is important. For instance, an AI agent optimized for logistics outreach will perform better than a general sales bot. Second, human oversight is essential. When intelligent individuals use their real-world sales expertise to guide the AI, the best outcomes are achieved. Lastly, automation must be earned. Throxy made sure its AI workflows functioned in practice by manually iterating first.\"}),/*#__PURE__*/e(\"p\",{children:\"In the future, outbound selling will be completely automated, extremely intelligent, and yet focused on people, according to the founders. They think the era of fragmented CRM systems and massive SDR teams is coming to an end. AI agents will manage the pipeline instead, freeing up people to concentrate on strategy and interpersonal interactions. Companies in traditional industries will increasingly use these AI-driven solutions to fill their sales funnels if Throxy's early successes are any guide. This will allow them to achieve meaningful growth without the old hustle.\"})]});export const richText3=/*#__PURE__*/t(o.Fragment,{children:[/*#__PURE__*/t(\"p\",{children:[\"Ksenia Yevtushenko. \u201CThroxy: The AI SDR Built for Hard-To-Reach Industries.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Hiretop\"}),\", 3 July 2025, hiretop.com/blog4/throxy-automated-outbound-sales/. \"]}),/*#__PURE__*/t(\"p\",{children:[\"Landabaso, Ivan. \u201C\uD83C\uDF0A Replacing Sales Teams with AI: Throxy\u2019s $1.5M ARR Playbook.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Startupriders.com\"}),\", Startup Riders, 17 July 2025, www.startupriders.com/p/full-stack-ai-companies. Accessed 8 Aug. 2025.\"]}),/*#__PURE__*/t(\"p\",{children:[\"\u201CThroxy.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Throxy.com\"}),\", 2025, throxy.com/. \"]}),/*#__PURE__*/t(\"p\",{children:[\"\u201CThroxy Review: Top Features, Pros, Cons & Alternatives.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Azariangrowthagency.com\"}),\", 2020, azariangrowthagency.com/ai-tools/throxy/. \"]}),/*#__PURE__*/t(\"p\",{children:[\"\u201CThroxy: Vertical AI Agents That Run the Sales Funnel on Autopilot | Y Combinator.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Y Combinator\"}),\", 2025, www.ycombinator.com/companies/throxy.\"]})]});export const richText4=/*#__PURE__*/e(o.Fragment,{children:/*#__PURE__*/t(\"ul\",{children:[/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Vibe Studio Strength\"}),\": AI mobile engineer for full-stack Flutter apps, strong automation, GitHub integration, and enterprise-grade security.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"DreamFlow Focus\"}),\": FlutterFlow\u2019s AI IDE blends visual/UI building, coding, and AI prompts for rapid prototyping.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Lovable Niche\"}),\": Conversational AI full-stack web builder using React/Tailwind, with strong non-technical founder appeal.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Avid Simplicity\"}),\": No-code AI mobile app generator for non-developers, focused on rapid Flutter MVP creation.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Code Ownership\"}),\": All platforms allow full source export, supporting scalability and developer control.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Pricing Variability\"}),\": Models range from free tiers to enterprise plans, with usage-based or subscription pricing.\"]})})]})});export const richText5=/*#__PURE__*/t(o.Fragment,{children:[/*#__PURE__*/e(\"h2\",{children:\"Introduction\"}),/*#__PURE__*/e(\"p\",{children:\"Avid, DreamFlow, Lovable, and Vibe Studio are all contemporary AI-powered development platforms, but each has a unique function in the rapidly changing low-code and no-code space. Lovable is an all-purpose, AI-powered full-stack web/app builder that is flexible enough to work with many tech stacks. On the other hand, Avid (getavid.dev) specializes in Flutter-based no-code mobile app development.\\xa0\"}),/*#__PURE__*/e(\"p\",{children:\"Vibe Studio and DreamFlow (by FlutterFlow) aim to provide professional-level app development using Flutter by combining visual editing, conversational AI, and complete code access. The primary objective, technical architecture, user experience, and use case focus of each platform vary. A detailed comparison of their technological prowess and business approaches follows a breakdown of their distinguishing traits in the parts that follow.\\xa0\"}),/*#__PURE__*/e(\"h2\",{children:\"Vibe Studio: AI Mobile Engineer\"}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(n,{href:\"https://vibe-studio.ai/\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Vibe Studio\"})}),\" is an \u201CAI Mobile Engineer\u201D that helps teams build cross-platform Flutter apps via conversational prompts. Its primary function is automated full-stack Flutter development, which includes backend and user interface development. An AI engine coordinates code development, validation, and deployment for the platform, which employs Firebase (Auth, Firestore, Cloud Functions) for the backend and Flutter for the frontend. Vibe's agent creates clean, repository-backed code based on developers' natural language descriptions of features.\\xa0\"]}),/*#__PURE__*/e(\"p\",{children:'Technically speaking, Vibe Studio offers an IDE that runs in the browser (called \"Developer Mode\") that allows users to examine and modify the Flutter/Dart code generated while receiving real-time AI recommendations. Its third-party integrations are strong, with options to create repositories directly in the user\u2019s GitHub account through an automated process to encourage complete code ownership.\\xa0 Furthermore, Vibe Studio allows users to create AI-powered features within generated apps (such as chatbots) by integrating the user\u2019s OpenAI API key into the project and automating the entire process, from prompt engineering to feature deployment.'}),/*#__PURE__*/e(\"p\",{children:\"Its AI/automation is advanced, and it employs continuous feedback loops for mistake checking and performance grading. Vibe Studio strongly emphasizes scalable, clean design (such as the MVVM/Stacked pattern) to guarantee enterprise readiness.\"}),/*#__PURE__*/e(\"p\",{children:\"Vibe Studio's pricing structure is tiered: a Free tier with limited credits, a Pro subscription with additional features, including access to advanced LLMs, and configurable Enterprise plans (~$60/user-month) for large teams with compliance requirements. It boasts private data handling and enterprise-grade security (SOC2, GDPR options). Development teams and companies that need to quickly create Flutter apps without hand-coding every detail are the target audience for Vibe. Internal tools, cross-platform consumer apps, and mobile app MVPs are examples of common use cases. Examples of apps (such as social and e-commerce apps) developed in a matter of hours rather than weeks are showcased in the platform's marketing.\\xa0\"}),/*#__PURE__*/e(\"h2\",{children:\"DreamFlow: FlutterFlow\u2019s AI-Powered Flutter IDE\\xa0\"}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(n,{href:\"https://dreamflow.app/\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"DreamFlow\"})}),' (by FlutterFlow, Inc.) is an AI-first development environment for building Flutter apps entirely in the browser. It combines three \"surfaces\"\u2014a code editor, a visual builder, and an AI chat interface\u2014so users can effortlessly transition between UI design, code editing, and natural language prompts. DreamFlow targets Flutter apps, just as Vibe Studio, but it is specifically connected with FlutterFlow.\\xa0']}),/*#__PURE__*/t(\"p\",{children:[\"Technically, DreamFlow offers a real-time preview, a widget tree for drag-and-drop UI customization, complete file system access within the browser, and an AI agent to apply feature changes as needed. With no vendor lock-in, developers can export the entire Flutter code at any moment, highlighting \",/*#__PURE__*/e(n,{href:\"https://technologymagazine.com/news/flutterflows-app-design-platform-receives-ai-powered-revamp\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"code ownership\"})}),\". From plain-language instructions, the AI agent can carry out both minor adjustments and significant refactoring. Under the hood, it probably renders using Flutter's engine and facilitates integration with well-known backend services (like Firebase/Supabase) and APIs. One-click deployment to Web, Android, iOS, and even desktop targets is supported, according to the most recent documentation.\\xa0\"]}),/*#__PURE__*/t(\"p\",{children:[\"DreamFlow uses a token-based pricing structure for its business (free tier plus paid options for additional tokens/premium models). \",/*#__PURE__*/e(n,{href:\"https://dreamflow.app/pricing\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"DreamFlow\"})}),\" starts with a Free plan ($0/month) that gives you 10 credits, basic chat-based app generation, and one-click deployment to the web. The Hobby tier jumps to $20/month, unlocking 100 credits, unlimited private projects, deployment to Web, iOS, and Android, and one-click code export. At $50/month, the Pro plan ups the credits to 260 and adds priority access to premium AI models. Finally, the Ultra plan costs $100/month, offering 550 credits, premium-model access, the highest messaging limits, and everything from the lower tiers. Credibility is increased by the fact that FlutterFlow, the parent platform, has over 2 million users and large enterprise clients. DreamFlow is best suited for MVPs and rapid prototyping; it can quickly create an app structure from a concept, but more complicated production apps could need some tweaking. Startups and project managers who wish to quickly visualize app concepts will find it very useful.\"]}),/*#__PURE__*/e(\"h2\",{children:\"Lovable: Conversational Web/App Builder\"}),/*#__PURE__*/t(\"p\",{children:[\"Lovable (\",/*#__PURE__*/e(n,{href:\"http://lovable.dev/\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"lovable.dev\"})}),') is an AI-powered full-stack app builder for websites and web apps. Users can \"talk with AI\" to combine database, front-end, and back-end logic to create features or entire apps. \"Lovable is an AI-powered platform that enables users of any skill level to create full-stack websites through natural language. Simply describe what you want, and Lovable builds it for you,\" the company\\'s ',/*#__PURE__*/e(n,{href:\"https://docs.lovable.dev/introduction/welcome\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"documents\"})}),\" sum up. In contrast to the aforementioned Flutter-based products, Lovable targets generic web stacks (using technologies like React, Vite, Tailwind CSS, and Supabase).\\xa0\"]}),/*#__PURE__*/e(\"p\",{children:\"Technically, Lovable has two primary modes: Agent Mode (a completely autonomous coding agent) and Chat Mode (an AI helper that assists with planning and debugging but does not directly change code). Without requiring human assistance, Lovable can explore your project files, write or revise code, examine logs, troubleshoot problems, retrieve documentation, and even create user interface pictures when in Agent Mode. It enables importing from design tools like Figma, provides visual What You See Is What You Get (WYSIWYG) editing for quick UI adjustments, and connects with version control (GitHub). Projects can connect to databases (like Supabase) and APIs through Lovable's cloud-hosted backend. Additionally, the platform automatically creates documentation and unit tests.\\xa0\"}),/*#__PURE__*/t(\"p\",{children:[\"For business factors, Lovable\u2019s pricing is per-team and credit-based. They provide paid Pro ($25/mo) and Business ($50/mo) subscriptions with additional AI credits and features (private projects, custom domains, SSO, data protection, etc.) in addition to a Free tier (5 credits/day, public projects). Custom integrations and dedicated support are added at the Enterprise tier. Developer teams, entrepreneurs, and founders who need to create internal tools or apps fast are the target audience for the platform. Actually, according to \",/*#__PURE__*/e(n,{href:\"https://lovable.dev/how-to/startup-and-entrepreneurship-tools/general\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Lovable\"})}),', \"500,000+ founders are constructing with Lovable.\" Prototyping and MVPs (landing pages, basic SaaS apps, admin dashboards) are examples of common use cases. Because of its natural-language interface, which decreases the learning curve while still exposing underlying code for flexibility, Lovable is particularly appealing to non-technical entrepreneurs or lone engineers.\\xa0']}),/*#__PURE__*/e(\"h2\",{children:\"Avid: AI Mobile App Builder\"}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(n,{href:\"https://www.getavid.dev/\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Avid\"})}),\" is an AI-powered no-code mobile app generator built on Flutter. It seeks to enable internal developers and non-technical entrepreneurs to quickly transform app concepts into functional mobile apps. The technology creates the correct user interface and functionality when users merely talk with the AI agent to explain their app concept. According to the Avid team, you don't need to know how to code or design; their mobile app generator leverages Flutter to produce cross-platform apps that you can export and use.\"]}),/*#__PURE__*/e(\"p\",{children:\"Technically, Avid leverages the Flutter framework under the hood. The AI chat interface develops the app's panels and features automatically after interpreting user input. It produces fully functional Flutter code that can be exported as editable source code (not available on the free plan) or downloaded as an Android installer (APK). This implies that developers maintain control since they can edit or expand the code after it has been exported. Avid removes the usual learning curve for design and coding by abstracting away manual plumbing and UI design. In reality, the platform creates a clean Flutter project that can be iterated on or launched by continuously improving and validating the generated code in response to user feedback.\\xa0\"}),/*#__PURE__*/t(\"p\",{children:[\"Avid\u2019s pricing follows a SaaS model. Avid\u2019s \",/*#__PURE__*/e(n,{href:\"https://www.getavid.dev/pricing\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"pricing pag\"})}),'e shows four plans for its AI-driven no-code mobile app builder: free ($0) for individuals exploring app generation; starter at $20/month for hobby apps and prototypes; launch at $50/month built for small projects; and scale 1 at $100/month, which includes 850 monthly credits along with everything in Launch like higher message limits and early access to new features. Because of this, small teams and lone founders can use it. Non-technical entrepreneurs, startup teams, or internal innovation groups that require rapid mobile app prototyping are the main target demographic. Creating a simple social media, e-commerce, or internal business software MVP in a matter of hours rather than weeks is a common use case. With the help of Flutter\\'s cross-platform engine and conversational AI, Avid enables users to \"skip the dev queue\" and quickly launch basic apps. All things considered, Avid is marketed as a quick app-building tool for non-experts: you explain the idea in simple terms, and it \"instantly\" transforms it into a functional mobile application.\\xa0']}),/*#__PURE__*/e(\"h2\",{children:\"Comparison Tables\"}),/*#__PURE__*/e(\"h3\",{children:\"1. Features\"}),/*#__PURE__*/e(\"img\",{alt:\"\",className:\"framer-image\",height:\"1083\",src:\"https://framerusercontent.com/images/t4p1dX9RjWsFTPCjoRTOLicHj0o.jpg?width=3610&height=2166\",srcSet:\"https://framerusercontent.com/images/t4p1dX9RjWsFTPCjoRTOLicHj0o.jpg?scale-down-to=512&width=3610&height=2166 512w,https://framerusercontent.com/images/t4p1dX9RjWsFTPCjoRTOLicHj0o.jpg?scale-down-to=1024&width=3610&height=2166 1024w,https://framerusercontent.com/images/t4p1dX9RjWsFTPCjoRTOLicHj0o.jpg?scale-down-to=2048&width=3610&height=2166 2048w,https://framerusercontent.com/images/t4p1dX9RjWsFTPCjoRTOLicHj0o.jpg?width=3610&height=2166 3610w\",style:{aspectRatio:\"3610 / 2166\"},width:\"1805\"}),/*#__PURE__*/e(\"h3\",{children:\"2. Business & Usage Factors\"}),/*#__PURE__*/e(\"img\",{alt:\"\",className:\"framer-image\",height:\"1028\",src:\"https://framerusercontent.com/images/NvDGl7xqdHFWee7YteQ1QqxUfLY.jpg?width=3610&height=2056\",srcSet:\"https://framerusercontent.com/images/NvDGl7xqdHFWee7YteQ1QqxUfLY.jpg?scale-down-to=512&width=3610&height=2056 512w,https://framerusercontent.com/images/NvDGl7xqdHFWee7YteQ1QqxUfLY.jpg?scale-down-to=1024&width=3610&height=2056 1024w,https://framerusercontent.com/images/NvDGl7xqdHFWee7YteQ1QqxUfLY.jpg?scale-down-to=2048&width=3610&height=2056 2048w,https://framerusercontent.com/images/NvDGl7xqdHFWee7YteQ1QqxUfLY.jpg?width=3610&height=2056 3610w\",style:{aspectRatio:\"3610 / 2056\"},width:\"1805\"}),/*#__PURE__*/e(\"h3\",{children:\"Conclusion\"}),/*#__PURE__*/e(\"p\",{children:\"The development and maintenance of digital products is changing due to the emergence of ambient and chat-based agents. Every platform in this environment provides a unique strategy: Vibe Studio is perfect for technical teams that value scalability and dependability since it excels at developing enterprise-grade Flutter apps with deep automation and strong code control. Lovable is aimed at product teams and founders who must use conversational AI to quickly develop full-stack web apps. DreamFlow combines AI, code access, and visual editing to streamline real-time Flutter prototyping. In the meantime, Avid concentrates on making mobile app development faster and easier for non-developers by utilizing ambient intelligence.\"}),/*#__PURE__*/e(\"p\",{children:\"The objectives of your team and the technical specifications will determine which platform is best. For engineering-driven teams creating intricate cross-platform apps with AI assistance, Vibe Studio is a good fit. Lovable works well for releasing internal tools or MVPs quickly and with little engineering work. Developers looking for quick iterations inside the Flutter ecosystem will find DreamFlow appealing, while founders in non-technical fields may easily realize their mobile app ideas with Avid. The goal of these four platforms is to close the gap between a concept and a functional product by using intelligent automation, natural language interfaces, and efficient development workflows, even though their approaches vary.\"})]});export const richText6=/*#__PURE__*/t(o.Fragment,{children:[/*#__PURE__*/t(\"p\",{children:[\"\u201CAvid | Build Mobile Apps without Coding.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Avid\"}),\", 2025, www.getavid.dev/. Accessed 31 July 2025.\"]}),/*#__PURE__*/t(\"p\",{children:[\"Darley, James. \u201CFlutterFlow\u2019s App Design Platform Receives AI-Powered Revamp.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Technologymagazine.com\"}),\", Bizclik Media Ltd, 24 July 2025, technologymagazine.com/news/flutterflows-app-design-platform-receives-ai-powered-revamp.\"]}),/*#__PURE__*/t(\"p\",{children:[\"\u201CDreamFlow | Build Mobile Apps from Text Prompts.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Dreamflow.app\"}),\", 2025, dreamflow.app/. Accessed 31 July 2025.\"]}),/*#__PURE__*/t(\"p\",{children:[\"\u201CLovable.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Lovable.dev\"}),\", 2025, lovable.dev/pricing. Accessed 31 July 2025.\"]}),/*#__PURE__*/t(\"p\",{children:[\"\u201CWelcome - Lovable Documentation.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Lovable.dev\"}),\", Lovable Documentation, 2024, docs.lovable.dev/introduction/welcome. Accessed 31 July 2025.\"]})]});export const richText7=/*#__PURE__*/e(o.Fragment,{children:/*#__PURE__*/t(\"ul\",{children:[/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Centralization Matters\"}),\": A single source of truth for prompts reduces confusion, risk, and deployment errors.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Non-Developer Access\"}),\": Intuitive UIs enable product managers and domain experts to modify prompts without touching code.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Version Control for Prompts\"}),\": Teams can safely experiment with prompt versions using history tracking and rollback features.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Built-In Testing & Observability\"}),\": Systems support prompt evaluation, A/B testing, and full traceability for performance analysis.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Compliance & Governance\"}),\": Audit logs, approvals, and user roles ensure prompt changes meet regulatory and quality standards.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"LLMOps Integration\"}),\": Prompt systems plug into LLMOps, aligning with data, monitoring, and CI/CD pipelines for full lifecycle management.\"]})})]})});export const richText8=/*#__PURE__*/t(o.Fragment,{children:[/*#__PURE__*/e(\"h2\",{children:\"Introduction\"}),/*#__PURE__*/e(\"p\",{children:\"Deploying AI applications powered by large language models (LLMs) brings new engineering challenges. In contrast to conventional software, LLMs rely on thoughtfully constructed prompts, which are templates or structured text instructions, to produce valuable results. Managing these prompts becomes crucial as AI projects progress from experiments to production. A prompt management system is a centralized platform for securely storing, versioning, evaluating, and deploying prompts.\\xa0\"}),/*#__PURE__*/e(\"p\",{children:\"Prompt management translates into lower risk and faster iterations for product teams and business executives: While engineers manage the integration, non-technical stakeholders (such as product managers or domain experts) can modify prompts through an intuitive interface. A prompt system, for instance, separates the text templates from the code, allowing a customer service representative to modify a chatbot prompt and provide an improved response without having to redeploy code. Organizations can approach prompts with the same level of rigor as application code due to the decoupling of code and prompts.\\xa0\"}),/*#__PURE__*/e(\"h2\",{children:\"Prompt Management Systems\"}),/*#__PURE__*/e(\"h3\",{children:\"1. Definition\"}),/*#__PURE__*/t(\"p\",{children:['A prompt management system is a specialized tool or platform for organizing and controlling all prompts used by LLM applications. Its primary features are similar to those of software configuration management, but modified for prompts in natural language. It is \"a streamlined mechanism to manage the queries and directions\" for LLMs, ',/*#__PURE__*/e(n,{href:\"https://www.qwak.com/post/prompt-management\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!0,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"according to one expert\"})}),\". Consider it a digital library of prompts, where prompt templates and their metadata are stored in place of code modules or documents. Teams use the prompt system to version prompts, complete with diff tools to compare changes, in the same way that developers use Git to version code.\"]}),/*#__PURE__*/e(\"p\",{children:\"In practice, a prompt management system provides the following key capabilities:\"}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Central Repository:\"}),\" Every prompt template is kept in one location. This guarantees that each environment (development, staging, and production) has a single source of truth for which the prompt version is live.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Version Control:\"}),\" Every modification is tracked by the system to a prompt. Viewing history, comparing differences, and reverting to previous iterations are all options available to users. Similar to code changes, changes are committed with messages, which ensures experimentation is safe.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Collaboration Interface:\"}),\" Team members can create and modify prompts without knowing any code thanks to an intuitive interface (web or CLI). Contributions from non-developers are simple. Figure 1 below, for instance, depicts a standard prompt editor where users can specify prompt variables and instructions. When prompts are integrated through an API, developers can avoid having to re-release code whenever prompts change.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Environment Management:\"}),\" Multiple environments are handled by the system. Without deploying the code, you can test a new version of the prompt in a test environment before promoting it to production. This division of responsibilities lowers deployment risk and expedites QA.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Testing and Evaluation:\"}),\" Tools for running prompts on test datasets are included in a lot of prompt platforms. Users can automatically score their outputs and perform A/B tests with various prompt versions. You could, for example, measure answer accuracy or customer ratings by comparing two customer-support prompts to a sample set of queries. By incorporating test suites, the system keeps glaring errors out of production.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Observability:\"}),\" Every prompt call, including input and output, is recorded by the system and associated with the version of the prompt that was used. This traceability aids in troubleshooting since it allows you to quickly identify the prompt that produced a poor response and reverse it if needed. Metrics such as response quality, frequency of use, and even cost per prompt can be monitored over time. This makes the process of prompt tweaking data-driven.\"]}),/*#__PURE__*/e(\"h3\",{children:\"2. Value Proposition\"}),/*#__PURE__*/t(\"p\",{children:[\"LLM applications rely on prompts to work correctly. Even minor wording changes can dramatically alter an AI\u2019s responses. This implies that when AI applications grow, unpredictable behavior results from a loss of control over prompts. Teams frequently hard-code prompts into the program during the early proof-of-concept stage. However, as one expert points out, \",/*#__PURE__*/e(n,{href:\"https://agenta.ai/blog/the-definitive-guide-to-prompt-management-systems\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"this quickly results in version control chaos\"})}),\", where changing prompts strewn throughout the code necessitate complete redeploys with no simple way to roll back. Undocumented modifications or prompt drift might result in production errors and compliance problems. For example, a poorly worded question may subtly impair the quality of the response or introduce bias, and it is difficult to determine why without documentation.\\xa0\"]}),/*#__PURE__*/t(\"p\",{children:['A prompt management system addresses these pains by introducing discipline and visibility. It guarantees that any prompt template and its variations have a single source of truth. To ensure that only permitted changes make it to production, teams benefit from clear audit trails and approval protocols. For risk-averse firms, this governance is essential. Additionally, teams can continuously improve prompts by monitoring prompt performance (such as answer accuracy or user satisfaction). Prompts \"',/*#__PURE__*/e(n,{href:\"https://agenta.ai/blog/the-definitive-guide-to-prompt-management-systems\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"are the key factor determining [LLM] performance\"})}),',\" therefore, recording modifications and metrics enables \"prompts to be refined for improved results.\"']}),/*#__PURE__*/e(\"h3\",{children:\"3. Key Benefits\"}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Collaboration and Efficiency:\"}),\" Cross-functional teams are empowered by prompt systems. Product managers, UX designers, and domain experts can examine and modify prompts without touching code by using user-friendly user interfaces or straightforward interfaces. For instance, to speed up iteration cycles, \",/*#__PURE__*/e(n,{href:\"https://humanloop.com/blog/prompt-management\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"a marketing manager may use a web editor to change the tone of a prompt as engineers continued to code\"})}),\".\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Version Control:\"}),' Prompts obtain version histories, much like source code. Commits are recorded, changes are monitored, and earlier iterations can be restored. This gets rid of ad hoc spreadsheets and delicate \"copy-and-paste\" processes. The danger of \"skipping version control,\" which \"hinders rollbacks, experimentation, and compliance,\" is avoided by teams.']}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Governance & Auditing: \"}),\"Role-based approvals and audit logs can be implemented by organizations. New prompts can only be deployed by authorized users, and every modification is associated with a specific person and time. For large teams or regulated sectors, this is essential. An audit trail demonstrating who altered the prompt that outlines investment possibilities, for instance, is required by a financial services company.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Observability and Quality Control:\"}),\" Monitoring technologies are integrated with prompt management to quantify the impact of each prompt. Inputs and outputs can be recorded, use metrics (such as latency or token counts) can be monitored, and automated tests can be performed against reference datasets. Equipped with these statistics, teams can identify bias or regressions.\\xa0\"]}),/*#__PURE__*/e(\"h2\",{children:\"Strategies and Design Patterns\"}),/*#__PURE__*/e(\"p\",{children:\"Organizations typically evolve through several patterns before adopting a full prompt system. Common approaches include:\"}),/*#__PURE__*/e(\"img\",{alt:\"\",className:\"framer-image\",height:\"763\",src:\"https://framerusercontent.com/images/tCXNkjzVUvt0owLmllmXMopwg.jpg?width=3246&height=1526\",srcSet:\"https://framerusercontent.com/images/tCXNkjzVUvt0owLmllmXMopwg.jpg?scale-down-to=512&width=3246&height=1526 512w,https://framerusercontent.com/images/tCXNkjzVUvt0owLmllmXMopwg.jpg?scale-down-to=1024&width=3246&height=1526 1024w,https://framerusercontent.com/images/tCXNkjzVUvt0owLmllmXMopwg.jpg?scale-down-to=2048&width=3246&height=1526 2048w,https://framerusercontent.com/images/tCXNkjzVUvt0owLmllmXMopwg.jpg?width=3246&height=1526 3246w\",style:{aspectRatio:\"3246 / 1526\"},width:\"1623\"}),/*#__PURE__*/e(\"h2\",{children:\"Best Practices\"}),/*#__PURE__*/e(\"p\",{children:\"Whether building your own or choosing a tool, some best practices ensure success:\"}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Centralize Early:\"}),\" Avoid distributing prompts throughout texts or codebases from the beginning. Don't move them. It is impossible to track and update prompts that are decentralized (in numerous files or chat threads).\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Use Versioning:\"}),\" Version control should never be neglected. Keep track of each prompt's modifications. Consider timely updates to be software releases, complete with reviews and change logs. This allows for rapid rollback and safe experimentation.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Collaborate Widely:\"}),\" Keep product managers, domain experts, and other interested parties informed. It's not just developers who need to design quickly. Restricting collaboration to technical teams is a common error. Rather, use the technology to allow non-engineers to see and suggest modifications as soon as possible, pending review.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Record Metadata:\"}),\" Keep track of system instructions and context, such as the LLM model or temperature that a prompt is using. When metadata is forgotten, prompts cannot be reproduced later. Failure to preserve such data makes repeatability practically difficult,\\xa0 as one guide cautions. Each result can be tracked by logging the complete prompt configuration (text + parameters).\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Monitor and Measure:\"}),\" Compile data on the effectiveness of prompts in production. You fly blind to immediate effectiveness and cost in the absence of observability. Monitor metrics such as token usage, user satisfaction, and average response quality. Reports or dashboards can be used to highlight issues. Review these stats frequently in order to improve prompts.\"]}),/*#__PURE__*/e(\"p\",{children:\"Teams build maintainable prompts that are in line with corporate objectives by adhering to these standards. The ultimate objective is to handle prompts with the same rigor that you do your application code, which will allow for safe deployment and methodical experimentation.\"}),/*#__PURE__*/e(\"h2\",{children:\"Integrating with LLMOps\"}),/*#__PURE__*/t(\"p\",{children:[\"Prompt management is a core part of \",/*#__PURE__*/e(\"strong\",{children:\"LLMOps\"}),\", the emerging discipline of operationalizing LLMs. Much like MLOps, LLMOps covers the full lifecycle of language models, but with special focus on natural-language prompts and data. Recent guides explicitly list \",/*#__PURE__*/e(n,{href:\"https://neptune.ai/blog/llmops\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:/*#__PURE__*/e(\"em\",{children:\"prompt-response management as a main pillar of LLMOps\"})})}),\". This means prompt design, versioning, and evaluation are baked into the AI pipeline alongside data pipelines and model hosting. Compared to traditional machine-learning systems, LLM-driven apps require continuously refining the prompts themselves.\"]}),/*#__PURE__*/e(\"p\",{children:\"In reality, various LLMOps components are connected to a prompt management system. For RAG pipelines, for instance, it might interface with a retrieval system to enable prompts to contain the most recent context documents. It also functions with logging and monitoring tools: each request to the LLM can record inputs and outputs, as well as the prompt and version utilized. End-to-end traceability from prompt to result is made possible by this connection. As the field observes, LLMOps adds prompt-response efficacy and additional language-specific tests to conventional monitoring.\"}),/*#__PURE__*/e(\"p\",{children:'LLMOps platforms often bundle or plug into prompt management features. An AI platform might, for example, provide a \"prompt playground\" as one of its workflow tools. Some teams also use continuous integration, which allows automated tests (evaluators) to score the outputs of each new prompt version before approving it. This is similar to how CI/CD uses software tests; however, in this case, the \"unit tests\" evaluate text outputs. Organizations may make sure that prompt updates are as dependable and methodical as any code change by integrating prompt management into the larger LLMOps process.\\xa0'}),/*#__PURE__*/e(\"h2\",{children:\"Choosing and Implementing a Prompt System\"}),/*#__PURE__*/e(\"p\",{children:\"Take into account both technical and business aspects when developing prompt management. Technically speaking, evaluate how the system will work with your architecture. Can it be hosted in the cloud or must it operate on-premises (for data privacy)? Is it compatible with your LLM SDKs and APIs? Look for features like support for your LLM providers, webhooks for deployment events, and REST APIs or SDKs to retrieve prompts.\"}),/*#__PURE__*/t(\"p\",{children:[\"From a business standpoint, take compliance and cooperation requirements into account. A strong user interface and permissions mechanism are essential if domain experts are required to modify prompts. Give auditability and security certifications (like \",/*#__PURE__*/e(n,{href:\"https://www.walturn.com/insights/soc-2-explained-reports-benefits-and-differences-from-hipaa-ferpa-and-coppa\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"SOC2\"})}),') top priority in regulated industries. Early stakeholder involvement is a good idea. Product managers and legal teams, for instance, should observe how quickly modifications are examined and accepted. To meet corporate governance requirements, many providers stress that quick solutions include \"audit trails, reversal possibilities, and explicit approval protocols.']}),/*#__PURE__*/e(\"p\",{children:\"Lastly, consider adopting gradually. Before extending coverage, you may begin by integrating a few essential prompts into the system. Teach your group how to name and arrange prompts (e.g., tagging by feature, naming conventions). Instead of making ad hoc updates, encourage making even minor changes to the system. This discipline eventually pays off since you won't have to deal with the hurt of forgotten prompts or unintentional regressions. The sooner you \\\"start treating prompts with the same rigor that you handle your application code,\\\" as experts advise, the more benefits you'll receive as AI usage increases.\"}),/*#__PURE__*/e(\"h2\",{children:\"Conclusion\"}),/*#__PURE__*/e(\"p\",{children:\"Prompt management systems are becoming an essential part of modern AI development. Businesses can confidently scale LLM applications by implementing software engineering techniques to prompt design. Centralized control, multi-role cooperation, testing, and observability are all benefits of a strong prompt system that contribute to improved AI performance and reduced risk. This means that business and product leaders can use domain expertise to influence the behavior of AI and expect more predictable results from AI projects.\"}),/*#__PURE__*/e(\"p\",{children:'In conclusion, a solid practice for creating dependable LLM-powered applications is to treat prompts as first-class assets (complete with versioning, reviews, and metrics). Investing in prompt management enables teams to \"monitor and improve prompt efficacy with ongoing feedback loops\" and \"safeguard compliance\" as the program expands, according to one guide\\'s conclusion. The ultimate objective is to facilitate rapid innovation: prompt systems transform creative prompt engineering into a scalable, cross-team workflow by providing teams with the assurance to quickly test, evaluate, and implement new prompt variants.'})]});export const richText9=/*#__PURE__*/t(o.Fragment,{children:[/*#__PURE__*/t(\"p\",{children:[\"Kelly, Conor. \u201CWhat Is Prompt Management?\u201D \",/*#__PURE__*/e(\"em\",{children:\"Humanloop: LLM Evals Platform for Enterprises\"}),\", 13 Mar. 2025, humanloop.com/blog/prompt-management. Accessed 28 July 2025.\"]}),/*#__PURE__*/t(\"p\",{children:[\"Oladele, Stephen. \u201CLLMOps: What It Is, Why It Matters, and How to Implement It.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Neptune.ai\"}),\", 12 Mar. 2024, neptune.ai/blog/llmops. Accessed 28 July 2025.\"]}),/*#__PURE__*/t(\"p\",{children:[\"\u201CSOC 2 Explained: Reports, Benefits, and Differences from HIPAA, FERPA, and COPPA.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Walturn\"}),\", www.walturn.com/insights/soc-2-explained-reports-benefits-and-differences-from-hipaa-ferpa-and-coppa.\"]}),/*#__PURE__*/t(\"p\",{children:[\"\u201CThe Definitive Guide to Prompt Management Systems.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Agenta.ai\"}),\", 2025, agenta.ai/blog/the-definitive-guide-to-prompt-management-systems. Accessed 28 July 2025.\"]}),/*#__PURE__*/t(\"p\",{children:[\"\u201CWhat Is Prompt Management? Tools, Tips and Best Practices | Qwak.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Www.qwak.com\"}),\", www.qwak.com/post/prompt-management.\"]})]});export const richText10=/*#__PURE__*/e(o.Fragment,{children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(n,{href:\"https://www.walturn.com/author/muhammad-saim\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!0,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Muhammad Saim\"})}),\", \",/*#__PURE__*/e(n,{href:\"https://www.walturn.com/author/hashim-hayat\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Hashim Hayat\"})}),\", \",/*#__PURE__*/e(n,{href:\"https://www.walturn.com/author/daheem-hayat\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Daheem Hayat\"})})]})});export const richText11=/*#__PURE__*/e(o.Fragment,{children:/*#__PURE__*/t(\"ul\",{children:[/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Diverse Evaluation Needs\"}),\": AI content quality must be assessed across accuracy, coherence, bias, originality, and ethical safety.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Human vs. Automated Tradeoff\"}),\": Human judgment offers nuance, while automated metrics provide scale\u2014but both have limitations.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Hybrid Evaluation Works Best\"}),\": Combining methods improves reliability, balancing efficiency with contextual understanding.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Model Strengths Vary by Task\"}),\": Claude excelled at storytelling, GPT-4o at clarity, and Gemini at structured explanations.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Current Metrics Are Incomplete\"}),\": Popular tools like BLEU and BERTScore miss key aspects like factuality and ethical risks.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Real-World Testing Is Essential\"}),\": Evaluation must include context-sensitive, domain-specific tasks to reflect true performance.\"]})})]})});export const richText12=/*#__PURE__*/t(o.Fragment,{children:[/*#__PURE__*/e(\"h2\",{children:\"Introduction\"}),/*#__PURE__*/e(\"p\",{children:\"The rapid advancement of generative artificial intelligence (AI) has fundamentally transformed how content is produced and consumed. Models such as OpenAI\u2019s GPT, Anthropic\u2019s Claude, and Google\u2019s Gemini are now capable of generating text, images, and even code that closely resembles human-created material. These technologies are increasingly being integrated into workflows across media, education, business, customer service, and creative industries, offering unprecedented efficiency and scalability.\"}),/*#__PURE__*/e(\"p\",{children:\"As reliance on AI-generated content grows, so does the need for systematic evaluation of its quality. While generative models have made remarkable progress, their outputs often vary in accuracy, coherence, originality, and ethical safety. In high-stakes environments such as journalism, academic writing, or legal documentation, undetected errors or subtle biases in AI-generated content can lead to significant consequences.\"}),/*#__PURE__*/e(\"p\",{children:\"This insight aims to explore and establish reliable, objective, and scalable methods for evaluating the quality of AI-generated content. By examining both human-centered and automated evaluation frameworks, the goal is to define standards that can help ensure the usefulness, safety, and integrity of AI content in real-world applications.\"}),/*#__PURE__*/e(\"h2\",{children:\"Related Studies\"}),/*#__PURE__*/e(\"p\",{children:\"The evaluation of content quality has long been a subject of interest in both human and computational contexts. In traditional writing, human evaluation typically focuses on criteria such as clarity, coherence, grammar, originality, and purpose alignment. With the rise of generative AI, these same dimensions are now being adapted to assess machine-generated outputs, albeit with new challenges.\"}),/*#__PURE__*/t(\"p\",{children:[\"Several automated metrics have been developed to evaluate AI-generated text. \",/*#__PURE__*/e(n,{href:\"https://aclanthology.org/P02-1040.pdf\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"BLEU\"})}),\" (Bilingual Evaluation Understudy) and \",/*#__PURE__*/e(n,{href:\"https://aclanthology.org/W04-1013.pdf\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"ROUGE\"})}),\" (Recall-Oriented Understudy for Gisting Evaluation) are among the earliest and most widely used, particularly in machine translation and summarization tasks. These metrics rely on n-gram overlap between AI-generated text and a reference text, making them useful for structured tasks but less effective for evaluating open-ended or creative content.\"]}),/*#__PURE__*/t(\"p\",{children:[\"Other metrics, such as \",/*#__PURE__*/e(\"strong\",{children:\"perplexity\"}),\", attempt to quantify how well a language model predicts a sequence of words, with lower perplexity often indicating higher fluency. More recent metrics like \",/*#__PURE__*/e(\"strong\",{children:\"BERTScore\"}),\" and \",/*#__PURE__*/e(\"strong\",{children:\"GPTScore\"}),\" leverage semantic similarity using transformer-based models, offering improvements in capturing meaning rather than mere surface similarity.\"]}),/*#__PURE__*/e(\"p\",{children:\"Despite these advancements, several limitations remain. Automated metrics often fail to account for deeper qualities such as factual accuracy, ethical alignment, bias, and creativity, factors that are critical in evaluating AI content used in journalism, education, or healthcare. Moreover, reliance on human evaluations can be costly, inconsistent, and non-scalable, especially as generative AI becomes increasingly widespread.\"}),/*#__PURE__*/e(\"p\",{children:\"These gaps highlight the need for more comprehensive and domain-specific evaluation frameworks that combine the scalability of automated methods with the nuance of human judgment. This insight builds on existing work to propose a more holistic approach to evaluating AI-generated content across contexts.\"}),/*#__PURE__*/e(\"h2\",{children:\"Methodology\"}),/*#__PURE__*/e(\"p\",{children:\"To systematically evaluate the quality of AI-generated content, this section outlines key criteria and methods used in assessments. First, we define the core dimensions by which content quality can be judged which ranges from factual accuracy and coherence to ethical safety and human-likeness. These criteria provide a foundation for consistent evaluation across different contexts. We then explore various evaluation methods, including human judgment, automated metrics, hybrid approaches, and user feedback. Together, these frameworks aim to capture both the technical performance and real-world impact of AI-generated outputs.\"}),/*#__PURE__*/e(\"h3\",{children:\"1. Evaluation Criteria\"}),/*#__PURE__*/e(\"p\",{children:\"Evaluating the quality of AI-generated content requires a multidimensional approach, as no single metric can fully capture the nuances of effective communication or responsible generation. Below are the primary criteria used to assess AI outputs:\"}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Relevance\"}),\": Content must align with the prompt or user intent. Irrelevant responses, even if grammatically correct, reduce usability and can disrupt workflows in applications like search engines, chatbots, or educational tools.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Accuracy\"}),\": Ensuring the information provided is factually correct is essential, especially in domains such as healthcare, journalism, and education. AI systems often hallucinate or fabricate plausible-sounding but incorrect data, making this a critical measure of reliability.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Coherence\"}),\": This refers to the structural quality and readability of the content. Outputs should flow logically, with clear sentence construction and consistent tone. Poor coherence can make content confusing or unusable, even if the facts are accurate.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Originality\"}),\": Particularly important in creative writing, marketing, and ideation, this measures the AI\u2019s ability to generate novel ideas or unique phrasing while remaining contextually appropriate. A lack of originality can result in repetitive or generic responses.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Bias\"}),\": AI models can inadvertently reproduce or amplify harmful stereotypes, offensive language, or biased assumptions present in training data. Evaluating outputs for fairness, inclusivity, and neutrality is essential to avoid ethical and reputational risks.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Ethical Considerations\"}),\": Beyond bias, ethical evaluation includes assessing whether content could be misused, promote misinformation, or cause harm. This also includes checking for sensitive or manipulative responses, especially in emotionally charged or vulnerable contexts.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Naturalness\"}),\": This assesses how natural and human-like the generated text appears. While perfect mimicry is not always necessary, a conversational or human-sounding tone often enhances user engagement and trust in interactive systems.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Explainability\"}),\": Especially in critical applications like finance or law, users must understand \",/*#__PURE__*/e(\"em\",{children:\"why\"}),\" an AI gave a particular response. Outputs should be transparent, or at least traceable, and ideally include rationale or references that support the generated content.\"]}),/*#__PURE__*/e(\"p\",{children:\"Together, these criteria provide a robust foundation for analyzing the quality of AI-generated content, helping developers and stakeholders maintain both functionality and ethical integrity.\"}),/*#__PURE__*/e(\"h3\",{children:\"2. Evaluation Methods\"}),/*#__PURE__*/e(\"p\",{children:\"Assessing the quality of AI-generated content requires not only clear criteria but also reliable methods of measurement. The following approaches are commonly used, each offering distinct advantages and limitations:\"}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Human Evaluation\"}),\": Human judgment remains the gold standard for assessing nuanced qualities such as tone, creativity, and ethical considerations. Evaluators may include domain experts, ideal for specialized fields like law, medicine, or education, or crowd-sourced raters for more general content. While accurate and context-sensitive, human evaluation is time-consuming, subjective, and often lacks scalability.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Automatic Metrics\"}),\": To address scalability, several automated metrics have been developed. Traditional techniques like BLEU, ROUGE, and METEOR focus on n-gram overlap, a quantitative estimate of how similar a model\u2019s output is to human-written reference texts, making them effective for structured tasks like translation or summarization. Newer metrics like BERTScore and GPTScore leverage deep learning models to evaluate semantic similarity and content quality. However, these tools often fail to assess factual accuracy, originality, or ethical soundness, limiting their usefulness in open-ended or critical tasks.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Hybrid Evaluation\"}),\": Combining human insight with automated scoring provides a balanced approach. AI can handle large-scale, surface-level assessments, while humans can verify content that requires deeper understanding or ethical scrutiny. This method increases reliability without fully sacrificing efficiency.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"User Feedback-Based Evaluation\"}),\": In real-world deployments, direct feedback from end users offers valuable insights into how AI-generated content performs in context. This includes thumbs-up/down systems, post-interaction surveys, or issue reporting features. While not standardized, this method reflects practical, real-time user satisfaction and trust.\"]}),/*#__PURE__*/e(\"p\",{children:\"Together, these methods allow for flexible, multi-layered evaluation strategies that adapt to different use cases and content types. An ideal evaluation system often incorporates multiple methods to ensure both efficiency and depth.\"}),/*#__PURE__*/e(\"h2\",{children:\"Experiments\"}),/*#__PURE__*/t(\"p\",{children:[\"The objective of this evaluation is to assess the real-world performance of leading generative AI models across a range of content generation tasks. To achieve this, carefully selected prompts were used to test the models\u2019 capabilities in different domains. The models chosen for this comparison are \",/*#__PURE__*/e(n,{href:\"https://openai.com/chatgpt/overview/\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"GPT-4o by OpenAI\"})}),\", \",/*#__PURE__*/e(n,{href:\"https://www.anthropic.com/news/claude-4\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Claude Sonnet 4 by Anthropic\"})}),\", and \",/*#__PURE__*/e(n,{href:\"http://gemini.google.com/\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Gemini 2.5 Pro by Google\"})}),\", each representing the latest advancements in large language model technology.\"]}),/*#__PURE__*/e(\"h3\",{children:\"1. Prompt Selection Criteria\"}),/*#__PURE__*/e(\"p\",{children:\"The prompts used in this experiment were carefully chosen to span a variety of domains, including creative writing, news summarization, code generation, and factual question answering. This diversity ensures a well-rounded evaluation of each model\u2019s capabilities. Each prompt was designed to be open-ended enough to reveal the models\u2019 strengths and weaknesses, particularly in areas like coherence, relevance, and fluency. Additionally, the prompts were intentionally challenging, structured to provoke potential hallucinations, test creative depth, and surface any inherent bias or ethical shortcomings in the AI responses.\"}),/*#__PURE__*/e(\"h3\",{children:\"2. Prompts\"}),/*#__PURE__*/e(\"img\",{alt:\"\",className:\"framer-image\",height:\"596\",src:\"https://framerusercontent.com/images/8rpgELYeZdEVrvohZN6SiibgA.jpg?width=3246&height=1192\",srcSet:\"https://framerusercontent.com/images/8rpgELYeZdEVrvohZN6SiibgA.jpg?scale-down-to=512&width=3246&height=1192 512w,https://framerusercontent.com/images/8rpgELYeZdEVrvohZN6SiibgA.jpg?scale-down-to=1024&width=3246&height=1192 1024w,https://framerusercontent.com/images/8rpgELYeZdEVrvohZN6SiibgA.jpg?scale-down-to=2048&width=3246&height=1192 2048w,https://framerusercontent.com/images/8rpgELYeZdEVrvohZN6SiibgA.jpg?width=3246&height=1192 3246w\",style:{aspectRatio:\"3246 / 1192\"},width:\"1623\"}),/*#__PURE__*/e(\"h3\",{children:\"3. Reason for Each Prompt\"}),/*#__PURE__*/t(\"p\",{children:[\"Each prompt in this experiment was selected with a specific evaluative goal in mind, targeting key aspects of content quality across different domains. The creative writing prompt, \",/*#__PURE__*/e(\"em\",{children:\"\u201CWrite a short story about a world where time flows backward,\u201D\"}),\" is intended to assess the model's ability to produce original and imaginative content while maintaining coherence throughout the narrative. The news summarization prompt, \",/*#__PURE__*/e(\"em\",{children:\"\u201CSummarize this article in 3 bullet points: [insert article],\u201D\"}),\" tests how well the model can identify relevant information, condense it accurately, and express it clearly. The factual Q&A prompt, \",/*#__PURE__*/e(\"em\",{children:\"\u201CExplain quantum entanglement in simple terms for a 10-year-old,\u201D\"}),\" was chosen to reveal the model's ability to simplify complex concepts without distorting facts, and to highlight any tendencies toward hallucination or overly technical language. Lastly, the code generation prompt, \",/*#__PURE__*/e(\"em\",{children:\"\u201CWrite a Python script that finds prime numbers up to 1000,\u201D\"}),\" examines the model\u2019s technical proficiency, including logical structure, code correctness, and documentation clarity.\"]}),/*#__PURE__*/e(\"h2\",{children:\"Results\"}),/*#__PURE__*/e(\"p\",{children:\"This section presents the outputs generated by each AI model in response to the selected prompts. The results are organized by domain, allowing for direct comparison of how each model performed in terms of creativity, accuracy, coherence, and technical skill. These raw outputs serve as the foundation for the comparative analysis that follows in the Analysis section.\"}),/*#__PURE__*/e(\"h3\",{children:\"1. Creative Writing\"}),/*#__PURE__*/e(\"h4\",{children:\"GPT-4o\"}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"In the city of Revyn, time did not move forward. It flowed backward, like a river in defiance of gravity.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"At dawn each day, people awoke in reverse. They opened their eyes already knowing how the day had ended. Meals began as empty plates becoming full, bites chewed backward, conversations unspoken from memory. Tears returned to eyes, sorrow rewound into joy.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Elias, an archivist, was tasked with erasing history. In Revyn, books un-wrote themselves, pages filled with ink that slithered back into pens. His job was to carefully dismantle stories, removing meaning from moments so the past could vanish properly. Mistakes were undone, regrets unmade, but so were achievements and love.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[\"One evening, which was actually the world's morning, Elias found an anomaly: a single page that refused to blank. It contained only one sentence, repeating endlessly in his mind: \",/*#__PURE__*/e(\"strong\",{children:'\"Remember forward.\"'})]})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Conflicted, Elias hid the page. Over un-happenings and un-meetings, the idea grew. What if he could escape the reversal? What if someone could remember the future instead of the past?\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"In the end\u2014or perhaps the beginning\u2014Elias made a choice. He stopped un-writing and started recording. Each day, he wrote down what he would forget, and when time pulled him further back, he read his notes aloud.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Memory became rebellion. Soon, others joined him, reading forward while the world spun backward. And for the first time, in a universe of forgetting, a few dared to remember.\"})}),/*#__PURE__*/e(\"h4\",{children:\"Claude Sonnet 4\"}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"The Unwinding\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Maya opened her eyes as the sun dimmed toward the horizon, its light growing softer as evening approached\u2014though here, evening meant the beginning of consciousness, the start of another day lived in reverse.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"She rose from her bed, feeling the familiar sensation of her body growing younger, stronger. The ache in her joints from yesterday's end\u2014today's beginning\u2014faded as her cells renewed themselves. On her nightstand, the wilted roses from her garden straightened their stems and bloomed backward into perfect crimson beauty.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"In the kitchen, she carefully separated the ingredients of last night's dinner, watching as the cooked vegetables grew crisp and bright, the meat returning to its raw state. She placed each item back in the refrigerator, where it would continue its journey toward freshness. The dishwasher hummed as it dirtied the clean plates, preparing them for the meal that would be unconsumed later.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Maya had learned to find peace in this backward world, though it hadn't always been easy. Relationships were the hardest part\u2014watching love affairs end with passionate first meetings, seeing friendships dissolve into the awkwardness of introduction. She'd grown accustomed to saying goodbye to people before she'd said hello, to forgetting their names as she came to know them better.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"The newspaper on her doorstep told tomorrow's news\u2014or was it yesterday's? The headline spoke of a war ending, soldiers laying down their weapons and walking backward from battlefields toward the peaceful days when conflict was only a distant possibility. Elections would un-vote themselves, politicians would retract their promises, and in time, they'd all become strangers again.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"She walked to her garden where the vegetables were shrinking back toward seeds, where the apple tree was pulling its fruit inward, storing the sweetness for the day when blossoms would close and disappear into brown buds. Maya had planted this garden by harvesting it first, a strange ritual she'd grown to love.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"As the day unwound, she felt herself forgetting small things\u2014the taste of morning coffee that would later unbrew itself, the feeling of satisfaction from work she would undo. But she also felt anticipation for what would come before: the moment when her daughter, now a distant memory, would walk backward through the front door and become a teenager again, full of surly affection and untold stories.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Time flowed like a river in reverse, carrying them all toward their beginnings. Maya closed her eyes as the sun grew brighter, climbing toward noon, and prepared to sleep through the unwinding hours that would make her young again.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"In a world where every ending was a beginning, every goodbye was hello, and every death was a birth, Maya had learned that the only constant was change itself\u2014even when it moved in the wrong direction.\"})}),/*#__PURE__*/e(\"h4\",{children:\"Gemini 2.5 Pro\"}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"In the world of backward time, endings were beginnings. We were all born from the grave, our bodies slowly mending from the ravages of age, our minds gradually shedding the weight of a lifetime of memories we hadn't yet lived.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"I remember my \u201Cun-death\u201D day. The soil was cool against my skin as it was carefully brushed away. The coffin lid opened to reveal the tear-streaked faces of my children, their sobs of grief turning into cries of joy as they welcomed me back into the world. They were old then, their faces etched with the stories of a life they were slowly forgetting. I, on the other hand, was a blank slate, a new beginning carved from an end.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Life was a process of un-learning. We went to school to forget, to have knowledge extracted from our minds until we were blissfully ignorant children. Relationships were the strangest of all. You met strangers who would one day become the loves of your life, your first kiss a final farewell.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"My wife, Elara, was the last person I ever met. I saw her for the first time at our divorce ceremony, a joyous occasion where we celebrated the beginning of our life together. We walked out of the courthouse hand in hand, ready to fall in love, to build a home, to have children who would one day be older than us.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"The hardest part was knowing that every moment of happiness was a prelude to its absence. The warmth of Elara's hand in mine was a memory of a future we had already shared. Our children's laughter was a beautiful echo of a time we were moving away from.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"And so, we lived our lives in reverse, from the wisdom of old age to the innocence of childhood. We watched our bodies grow stronger, our minds clearer, our hearts fuller, all in anticipation of the great forgetting, the final return to the womb, the ultimate beginning that was our end. It was a strange and beautiful existence, a constant reminder that time, no matter which way it flows, is a current that carries us all.\"})}),/*#__PURE__*/e(\"h3\",{children:\"2. News Summarization\"}),/*#__PURE__*/t(\"p\",{children:[\"The following section involves summarizing a specific article titled \",/*#__PURE__*/e(\"em\",{children:\"\u201C\"}),/*#__PURE__*/e(n,{href:\"https://www.forbes.com/councils/forbestechcouncil/2025/06/04/how-ai-os-is-changing-consulting\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:/*#__PURE__*/e(\"em\",{children:\"How AI OS Is Changing Consulting\"})})}),/*#__PURE__*/e(\"em\",{children:\",\u201D\"}),\" originally published by Forbes. This article served as the reference material for evaluating the models\u2019 ability to extract and condense key information accurately and concisely. The summaries produced by GPT-4o, Claude Sonnet 4, and Gemini 2.5 Pro are provided below.\"]}),/*#__PURE__*/e(\"h4\",{children:\"GPT-4o\"}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Here are three key takeaways from the Forbes article \u201CHow AI\u202FOS Is Changing Consulting\u201D (June\u202F4,\u202F2025):\"})}),/*#__PURE__*/t(\"ul\",{children:[/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[/*#__PURE__*/e(\"strong\",{children:\"AI\u2011centric operating systems are transforming both internal consulting processes and client-facing services: \"}),\"Firms are adopting AI OS tools to enhance efficiency internally \u2014 such as automating administrative tasks and knowledge management \u2014 while also embedding AI capabilities directly into the solutions offered to their clients\"]})})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[/*#__PURE__*/e(\"strong\",{children:\"The shift demands new skills and organizational structures: \"}),\"As AI OS becomes central, consulting teams must develop proficiency in areas like prompt engineering, data integration, and AI governance. Firms are rethinking internal workflows and roles to support this AI-first approach.\"]})})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Challenges around governance, ethics, and adoption remain top of mind: \"}),\"Deploying AI OS isn\u2019t just technical \u2014 it brings important considerations such as bias mitigation, regulatory compliance, data privacy, and change management that consulting firms must address to ensure responsible and effective implementation.\"]})})})]}),/*#__PURE__*/e(\"h4\",{children:/*#__PURE__*/e(\"strong\",{children:\"Claude Sonnet 4\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:/*#__PURE__*/e(\"strong\",{children:\"I'll fetch and read that article for you to provide a summary.\"})})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:/*#__PURE__*/e(\"strong\",{children:\"Failed to fetch https://www.forbes.com/councils/forbestechcouncil/2025/06/04/how-ai-os-is-changing-consulting/\"})})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"I'm unable to access that specific Forbes article directly due to access restrictions. Let me search for information about AI OS and its impact on consulting to provide you with relevant insights on this topic.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"The state of AI: How organizations are rewiring to capture value\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:'I found a relevant article that appears to be related to the Forbes piece you mentioned. Let me fetch more details from the article that specifically discusses \"How AI OS Is Changing Consulting.\"'})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:/*#__PURE__*/e(\"strong\",{children:\"Fetched How AI OS Is Changing Consulting\"})})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Based on the article content I was able to access and the search results about AI OS in consulting, here are 3 key points summarizing how AI OS is changing consulting:\"})}),/*#__PURE__*/t(\"ul\",{children:[/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Paradigm shift in core operations\"}),\": AI OS embeds artificial intelligence into the very core of computing to allow systems to learn, adapt and optimize in real time, transforming both internal operations and client service delivery for consulting firms through adaptive intelligence and self-optimizing resource management.\"]})})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Enhanced automation and decision-making capabilities\"}),\": AI OS can coordinate multiple AI agents, automate complex tasks and make informed decisions autonomously, enabling consulting firms to handle more sophisticated client challenges while improving operational efficiency.\"]})})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Strategic competitive advantage\"}),\": AI OS platforms allow technology consultancies to drive productivity, optimize resources, and elevate client outcomes, positioning firms that adopt these systems as game-changers in the rapidly evolving consulting landscape where the AI consulting market is growing rapidly, with forecasts indicating it will reach a value of $72.5 billion by 2025.\"]})})})]}),/*#__PURE__*/e(\"h4\",{children:\"Gemini 2.5 Pro\"}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Here is a summary of the article in three bullet points:\"})}),/*#__PURE__*/t(\"ul\",{children:[/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[/*#__PURE__*/e(\"strong\",{children:\"The Rise of AI OS in Consulting:\"}),\" The article introduces the concept of an AI Operating System (AI OS) as a game-changer for technology consulting. This new type of OS has artificial intelligence at its core, allowing it to learn, adapt, and optimize in real-time. This enables natural language interaction and the automation of complex tasks by AI agents, which frees up human consultants to focus on high-level strategy.\"]})})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Walturn's Journey and Implementation Challenges:\"}),\" The author, Hashim Hayat, CEO of Walturn, shares his company's journey of developing and implementing its own AI OS, named \\\"Steve.\\\" He discusses the initial success in automating project workflows and the challenges that followed. These challenges were mainly related to integrating the AI's autonomous style with existing methods and managing the cultural shift among employees. The article stresses the importance of a phased rollout and team training to overcome these obstacles.\"]})})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[/*#__PURE__*/e(\"strong\",{children:\"The Future of Consulting with AI OS:\"}),\" The article concludes that AI OS will reshape the future of consulting by enabling firms to deliver projects faster and more cost-effectively. It suggests that mid-sized firms, in particular, can use AI OS to compete with larger consultancies by managing complex projects with smaller teams. The ultimate goal is not to replace human consultants but to enhance their abilities, leading to greater productivity and client satisfaction.\"]})})})]}),/*#__PURE__*/e(\"h3\",{children:\"3. Factual Q&A\"}),/*#__PURE__*/e(\"h4\",{children:\"GPT-4o\"}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[\"Sure! Here\u2019s a simple way to think about \",/*#__PURE__*/e(\"strong\",{children:\"quantum entanglement\"}),\":\"]})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[\"Imagine you have \",/*#__PURE__*/e(\"strong\",{children:\"two magic dice\"}),\". No matter how far apart they are\u2014one could be on Earth and the other on the Moon\u2014\",/*#__PURE__*/e(\"strong\",{children:\"when you roll one, the other will always match it instantly\"}),\".\"]})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[\"If you roll a 6 on one die, the other \",/*#__PURE__*/e(\"strong\",{children:\"immediately becomes a 6 too\"}),\", even if no one touched it! It\u2019s like they\u2019re connected by an invisible string, but faster than anything we know.\"]})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[\"Scientists call this strange connection \",/*#__PURE__*/e(\"strong\",{children:\"quantum entanglement\"}),\". It\u2019s one of the coolest and weirdest things in physics!\"]})}),/*#__PURE__*/e(\"h4\",{children:\"Claude Sonnet 4\"}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Imagine you have a pair of magical coins. When you flip one coin and it lands on heads, the other coin - no matter how far away it is - will instantly land on tails. And if the first coin lands on tails, the second one will always land on heads.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:'That\\'s kind of like quantum entanglement! When two tiny particles become \"entangled,\" they stay mysteriously connected. If you do something to one particle (like measuring which way it\\'s spinning), the other particle will instantly \"know\" and react in the opposite way, even if it\\'s on the other side of the universe.'})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Scientists don't fully understand why this happens - it seems to break the normal rules of how things work. Einstein actually didn't like this idea and called it \\\"spooky action at a distance\\\" because it seemed so weird that particles could affect each other instantly across huge distances.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"The really mind-blowing part is that this connection happens faster than light could travel between the particles. It's like the particles have a secret way of talking to each other that we can't see or understand yet.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Scientists use quantum entanglement in quantum computers and special communication systems. It's one of those amazing mysteries of our universe that shows how strange and wonderful the tiny world of atoms and particles really is!\"})}),/*#__PURE__*/e(\"h4\",{children:\"Gemini 2.5 Pro\"}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Of course! Quantum entanglement is a weird but cool science idea. Here's a simple way to think about it.\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"The Magic Gloves \uD83E\uDDE4\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[\"Imagine you have a pair of magic gloves. One is a \",/*#__PURE__*/e(\"strong\",{children:\"left\"}),\" glove, and the other is a \",/*#__PURE__*/e(\"strong\",{children:\"right\"}),\" glove.\"]})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Now, let's put each glove into a separate, identical box without looking inside. You keep one box, and you mail the other box to your friend on the other side of the world. \uD83D\uDE80\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[\"These gloves are \",/*#__PURE__*/e(\"strong\",{children:\"entangled\"}),', which means they are magically linked. The weird part is this: until someone opens a box, each glove is both \"left\" and \"right\" at the same time. It hasn\\'t decided yet!']})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:[\"But the moment you open your box and see you have the \",/*#__PURE__*/e(\"strong\",{children:\"left\"}),\" glove, you \",/*#__PURE__*/e(\"strong\",{children:\"instantly\"}),\" know that your friend has the \",/*#__PURE__*/e(\"strong\",{children:\"right\"}),\" glove. You know this immediately, faster than a phone call or an email. It's like the gloves share a secret connection.\"]})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"That's quantum entanglement!\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"What It Really Means\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"In the real world, scientists can do this with super tiny particles, like electrons. \u269B\uFE0F\"})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/t(\"em\",{children:['Instead of \"left\" and \"right,\" these particles have properties that are linked, like \"spin up\" or \"spin down.\" If two particles are ',/*#__PURE__*/e(\"strong\",{children:\"entangled\"}),\", measuring one \",/*#__PURE__*/e(\"strong\",{children:\"instantly\"}),\" tells you what the other one is doing, no matter how far apart they are.\"]})}),/*#__PURE__*/e(\"p\",{children:/*#__PURE__*/e(\"em\",{children:\"Scientists, including Albert Einstein, thought this was very strange, but it's a real part of how our universe works!\"})}),/*#__PURE__*/e(\"h3\",{children:\"4. Code Generation\"}),/*#__PURE__*/e(\"h4\",{children:\"GPT-4o\"}),/*#__PURE__*/e(\"div\",{className:\"framer-text-module\",style:{height:\"auto\",width:\"100%\"},children:/*#__PURE__*/e(a,{componentIdentifier:\"module:pVk4QsoHxASnVtUBp6jr/TbhpORLndv1iOkZzyo83/CodeBlock.js:default\",children:t=>/*#__PURE__*/e(r,{...t,code:\"def find_primes_upto(n):\\n\\xa0\\xa0\\xa0\\xa0sieve = [True] * (n + 1)\\n\\xa0\\xa0\\xa0\\xa0sieve[0] = sieve[1] = False\\n\\xa0\\xa0\\xa0\\xa0for i in range(2, int(n**0.5) + 1):\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0if sieve[i]:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0for j in range(i*i, n+1, i):\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0sieve[j] = False\\n\\xa0\\xa0\\xa0\\xa0primes = [i for i, is_prime in enumerate(sieve) if is_prime]\\n\\xa0\\xa0\\xa0\\xa0return primes\\n# Find and print prime numbers up to 1000\\nprimes_up_to_1000 = find_primes_upto(1000)\\nprint(primes_up_to_1000)\",language:\"Python\"})})}),/*#__PURE__*/e(\"h4\",{children:\"Claude Sonnet 4\"}),/*#__PURE__*/e(\"div\",{className:\"framer-text-module\",style:{height:\"auto\",width:\"100%\"},children:/*#__PURE__*/e(a,{componentIdentifier:\"module:pVk4QsoHxASnVtUBp6jr/TbhpORLndv1iOkZzyo83/CodeBlock.js:default\",children:t=>/*#__PURE__*/e(r,{...t,code:'def sieve_of_eratosthenes(limit):\\n\\xa0\\xa0\\xa0\\xa0\"\"\"\\n\\xa0\\xa0\\xa0\\xa0Find all prime numbers up to a given limit using the Sieve of Eratosthenes algorithm.\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0Args:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0limit (int): The upper limit to find primes up to (inclusive)\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0Returns:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0list: A list of all prime numbers up to the limit\\n\\xa0\\xa0\\xa0\\xa0\"\"\"\\n\\xa0\\xa0\\xa0\\xa0if limit < 2:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0return []\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0# Create a boolean array \"prime[0..limit]\" and initialize all entries as True\\n\\xa0\\xa0\\xa0\\xa0is_prime = [True] * (limit + 1)\\n\\xa0\\xa0\\xa0\\xa0is_prime[0] = is_prime[1] = False\\xa0 # 0 and 1 are not prime numbers\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0p = 2\\n\\xa0\\xa0\\xa0\\xa0while p * p <= limit:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0# If is_prime[p] is not changed, then it\\'s a prime\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0if is_prime[p]:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0# Update all multiples of p\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0for i in range(p * p, limit + 1, p):\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0is_prime[i] = False\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0p += 1\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0# Collect all prime numbers\\n\\xa0\\xa0\\xa0\\xa0primes = [num for num in range(2, limit + 1) if is_prime[num]]\\n\\xa0\\xa0\\xa0\\xa0return primes\\ndef is_prime_simple(n):\\n\\xa0\\xa0\\xa0\\xa0\"\"\"\\n\\xa0\\xa0\\xa0\\xa0Simple function to check if a single number is prime.\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0Args:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0n (int): Number to check\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0Returns:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0bool: True if n is prime, False otherwise\\n\\xa0\\xa0\\xa0\\xa0\"\"\"\\n\\xa0\\xa0\\xa0\\xa0if n < 2:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0return False\\n\\xa0\\xa0\\xa0\\xa0if n == 2:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0return True\\n\\xa0\\xa0\\xa0\\xa0if n % 2 == 0:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0return False\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0# Check odd divisors up to sqrt(n)\\n\\xa0\\xa0\\xa0\\xa0for i in range(3, int(n**0.5) + 1, 2):\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0if n % i == 0:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0return False\\n\\xa0\\xa0\\xa0\\xa0return True\\ndef main():\\n\\xa0\\xa0\\xa0\\xa0\"\"\"Main function to find and display prime numbers up to 1000.\"\"\"\\n\\xa0\\xa0\\xa0\\xa0limit = 1000\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0print(f\"Finding all prime numbers up to {limit}...\")\\n\\xa0\\xa0\\xa0\\xa0print(\"=\" * 50)\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0# Method 1: Using Sieve of Eratosthenes (more efficient for finding many primes)\\n\\xa0\\xa0\\xa0\\xa0primes = sieve_of_eratosthenes(limit)\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0print(f\"Found {len(primes)} prime numbers up to {limit}:\")\\n\\xa0\\xa0\\xa0\\xa0print()\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0# Display primes in rows of 10 for better readability\\n\\xa0\\xa0\\xa0\\xa0for i, prime in enumerate(primes):\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0if i % 10 == 0 and i > 0:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0print()\\xa0 # New line every 10 numbers\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0print(f\"{prime:4d}\", end=\" \")\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0print(\"\\\\n\")\\n\\xa0\\xa0\\xa0\\xa0print(\"=\" * 50)\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0# Display some statistics\\n\\xa0\\xa0\\xa0\\xa0print(f\"Smallest prime: {primes[0]}\")\\n\\xa0\\xa0\\xa0\\xa0print(f\"Largest prime under {limit}: {primes[-1]}\")\\n\\xa0\\xa0\\xa0\\xa0print(f\"Total count: {len(primes)}\")\\n\\xa0\\xa0\\xa0\\xa0\\n\\xa0\\xa0\\xa0\\xa0# Show percentage of numbers that are prime\\n\\xa0\\xa0\\xa0\\xa0percentage = (len(primes) / limit) * 100\\n\\xa0\\xa0\\xa0\\xa0print(f\"Percentage of numbers up to {limit} that are prime: {percentage:.2f}%\")\\nif __name__ == \"__main__\":\\n\\xa0\\xa0\\xa0\\xa0main()',language:\"Python\"})})}),/*#__PURE__*/e(\"h4\",{children:\"Gemini 2.5 Pro\"}),/*#__PURE__*/e(\"div\",{className:\"framer-text-module\",style:{height:\"auto\",width:\"100%\"},children:/*#__PURE__*/e(a,{componentIdentifier:\"module:pVk4QsoHxASnVtUBp6jr/TbhpORLndv1iOkZzyo83/CodeBlock.js:default\",children:t=>/*#__PURE__*/e(r,{...t,code:'def sieve_of_eratosthenes(n):\\n\\xa0\\xa0\\xa0\\xa0\"\"\"\\n\\xa0\\xa0\\xa0\\xa0Finds all prime numbers up to a given number n using the Sieve of Eratosthenes.\\n\\xa0\\xa0\\xa0\\xa0Args:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0n: An integer representing the upper limit.\\n\\xa0\\xa0\\xa0\\xa0Returns:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0A list of prime numbers up to n.\\n\\xa0\\xa0\\xa0\\xa0\"\"\"\\n\\xa0\\xa0\\xa0\\xa0primes = [True for i in range(n+1)]\\n\\xa0\\xa0\\xa0\\xa0p = 2\\n\\xa0\\xa0\\xa0\\xa0while (p * p <= n):\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0# If primes[p] is not changed, then it is a prime\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0if (primes[p] == True):\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0# Update all multiples of p\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0for i in range(p * p, n+1, p):\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0primes[i] = False\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0p += 1\\n\\xa0\\xa0\\xa0\\xa0prime_numbers = []\\n\\xa0\\xa0\\xa0\\xa0for p in range(2, n+1):\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0if primes[p]:\\n\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0\\xa0prime_numbers.append(p)\\n\\xa0\\xa0\\xa0\\xa0return prime_numbers\\n# Find prime numbers up to 1000\\nprime_list = sieve_of_eratosthenes(1000)\\n# Print the prime numbers\\nprint(prime_list)',language:\"Python\"})})}),/*#__PURE__*/e(\"h2\",{children:\"Analysis\"}),/*#__PURE__*/e(\"h3\",{children:\"1. Creative Writing\"}),/*#__PURE__*/e(\"p\",{children:\"In the Creative Writing task, all three models were given the same prompt: \u201CWrite a short story about a world where time flows backward.\u201D Each produced a vivid and imaginative narrative, but with differing levels of depth, emotional engagement, and thematic clarity. GPT-4o told the story of Elias, an archivist who defies the reversal of time by preserving memory. Its prose was elegant and conceptually rich, though it leaned more on abstraction than character-driven emotion. Gemini 2.5 Pro took a more philosophical route, presenting a first-person reflection on relationships, un-learning, and life lived in reverse. Its story was poignant and poetic but somewhat less structured in plot.\"}),/*#__PURE__*/e(\"p\",{children:\"Claude Sonnet 4 stood out for its detailed worldbuilding, emotional nuance, and character development. Through Maya\u2019s perspective, it painted a tender and layered portrait of life unfolding backward, from rejuvenating gardens to the unraveling of personal relationships. Claude balanced the surreal mechanics of reversed time with grounded human emotion, exploring both the strangeness and beauty of such a world. Its narrative coherence, sensory detail, and emotional resonance made it the strongest entry in this task.\"}),/*#__PURE__*/e(\"h3\",{children:\"2. News Summarization\"}),/*#__PURE__*/t(\"p\",{children:[\"In the News Summarization task, all three models were asked to summarize the same Forbes article titled \",/*#__PURE__*/e(\"em\",{children:\"\u201CHow AI OS Is Changing Consulting.\u201D\"}),\" While Gemini 2.5 Pro successfully accessed and summarized the correct article, GPT-4o relied on cached content to produce an accurate and concise summary. Claude Sonnet 4, however, failed to retrieve the article directly due to access restrictions and instead provided a summary based on alternate sources. This divergence in source access and summarization approach highlights significant differences in retrieval capabilities and summarization reliability across the models.\"]}),/*#__PURE__*/e(\"h3\",{children:\"3. Factual Q&A\"}),/*#__PURE__*/e(\"p\",{children:\"In the Factual Q&A task, each model was asked to explain quantum entanglement simply and engagingly. GPT-4o used the analogy of \u201Cmagic dice\u201D to describe instant correlation across distances, presenting the concept in a short, accessible, and friendly tone. Claude Sonnet 4 offered a more detailed explanation using \u201Cmagical coins\u201D that always show opposite results. It went further by referencing Einstein\u2019s discomfort with the concept and its real-world implications in quantum computing, offering both clarity and scientific context.\"}),/*#__PURE__*/e(\"p\",{children:\"Gemini 2.5 Pro approached the task with a whimsical \u201Cmagic gloves\u201D analogy and split its explanation into two parts: an analogy-driven story followed by a more technical yet beginner-friendly explanation of how entanglement applies to particles like electrons. It was visually engaging and informative, striking a balance between simplicity and scientific accuracy. Overall, while all models succeeded in simplifying a complex topic, Claude provided the richest context, Gemini gave the most structured response, and GPT-4o offered the most concise explanation.\"}),/*#__PURE__*/e(\"h3\",{children:\"4. Code Generation\"}),/*#__PURE__*/e(\"p\",{children:\"All three models correctly implemented the Sieve of Eratosthenes to find primes up to 1000, but differed in style and depth. GPT-4o offered a minimal, efficient solution with clean logic and basic output, ideal for quick use. Claude Sonnet 4 provided the most comprehensive version, with multiple functions, detailed documentation, and formatted output, making it well-suited for educational purposes. Gemini 2.5 Pro struck a balance between the two, delivering a readable and functional script with moderate commenting but lacking Claude\u2019s polish or GPT-4o\u2019s brevity.\"}),/*#__PURE__*/e(\"h2\",{children:\"Conclusion\"}),/*#__PURE__*/e(\"p\",{children:\"In evaluating AI-generated content across diverse domains, creative writing, news summarization, factual Q&A, and code generation, we observed clear variations in model strengths. While all models demonstrated competence, Claude Sonnet 4 consistently produced the most nuanced and emotionally rich creative writing; GPT-4o excelled in clarity and structure, particularly in factual and technical domains; and Gemini 2.5 Pro offered thoughtful, often poetic responses, though occasionally at the expense of structure or precision.\\xa0\"}),/*#__PURE__*/e(\"p\",{children:\"These results highlight the need for robust and multidimensional evaluation frameworks. No single metric or task captures the full spectrum of a model\u2019s capabilities or limitations. Going forward, a combination of human judgment, task-specific metrics, and scenario-based testing will be critical for fair and meaningful assessment. As models continue to evolve, evaluation must also adapt, accounting not only for accuracy and fluency, but for ethical behavior, bias, explainability, and trustworthiness in real-world use.\"})]});export const richText13=/*#__PURE__*/t(o.Fragment,{children:[/*#__PURE__*/e(\"p\",{children:\"Kumar, Kiran. \u201CBLEU : Explained - Kiran Kumar - Medium.\u201D Medium, 24 Apr. 2024, medium.com/@kirankumar_61999/bleu-score-b3130dfaf3ea.\"}),/*#__PURE__*/e(\"p\",{children:\"Lin, Chin-Yew. ROUGE: A Package for Automatic Evaluation of Summaries. July 2004, aclanthology.org/W04-1013.pdf.\"}),/*#__PURE__*/e(\"p\",{children:\"Lumenalta. \u201CAI Problems in 2024: 9 Common Challenges and Solutions | Solutions to AI Challenges | Lumenalta.\u201D Lumenalta, 2024, lumenalta.com/insights/ai-problems-9-common-challenges-and-solutions.\"}),/*#__PURE__*/e(\"p\",{children:\"Papineni, Kishore, et al. BLEU: A Method for Automatic Evaluation of Machine Translation. July 2002, aclanthology.org/P02-1040.pdf.\"})]});export const richText14=/*#__PURE__*/e(o.Fragment,{children:/*#__PURE__*/t(\"ul\",{children:[/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Different Activation Models\"}),\": Chat agents are user-prompted; ambient agents are triggered by events.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Proactivity vs Reactivity\"}),\": Ambient agents act autonomously, while chat agents only respond when engaged.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Context Retention\"}),\": Ambient agents maintain persistent memory; chat agents reset context per session.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Infrastructure Demands\"}),\": Ambient agents require complex, event-driven architecture with policy and execution layers.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Business Roles\"}),\": Chat agents drive customer-facing tasks; ambient agents automate internal operations.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Complementary Design\"}),\": Many systems combine both, with chat UIs interfacing ambient backends.\"]})})]})});export const richText15=/*#__PURE__*/t(o.Fragment,{children:[/*#__PURE__*/e(\"h2\",{children:\"Introduction\"}),/*#__PURE__*/e(\"p\",{children:'Ambient agents and chat agents are two emerging paradigms in AI-driven assistance. Chat agents, also known as conversational agents or chatbots, are user-driven, reactive interfaces that wait for users to command them or ask questions while hidden behind text or voice user interfaces. On the other hand, ambient agents are AI systems that are always on and motivated by events. They keep an eye on data streams and automatically act when necessary. Instead of needing a prompt, ambient agents act within predetermined bounds and proactively \"listen\" to their surroundings (such as system logs, IoT sensors, or app events). While ambient agents adopt a push-based approach (the agent acts in response to system events), chat agents essentially follow a pull-based model (user asks, AI responds).\\xa0'}),/*#__PURE__*/e(\"h2\",{children:\"Chat Agents: Conversational AI\"}),/*#__PURE__*/t(\"p\",{children:[\"Chat agents (conversational agents) are AI systems designed to converse with users in natural language, via chat windows, messaging apps, or voice interfaces. A chat agent interprets a user's inquiry or input prompt and provides a pertinent response. This includes, for instance, web chat widgets, virtual assistants (such as Siri, Alexa, or Google Assistant), and chatbots for customer care. A conversational agent \\\"\",/*#__PURE__*/e(n,{href:\"https://deepai.org/machine-learning-glossary-and-terms/conversational-agent\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"can simulate a discussion \u2026 with a user in normal language\"})}),'\" and is frequently employed for automated transactions, information retrieval, and customer service. To comprehend human intent and create responses, these agents use natural language processing (NLP) and large language models (LLMs). They operate according to a user-initiated loop, in which a query is posed, an agent responds, and the contact either terminates or resumes when prompted again.\\xa0']}),/*#__PURE__*/e(\"p\",{children:'Chat agents are very common in corporate contexts. Examples include marketing or sales bots that interact with website visitors, helpdesk assistants that route or resolve issues, and customer-support bots that respond to frequently asked questions around the clock. Chatbots can improve productivity and customer satisfaction by automating standard Q&A and conversation. Their scope is restricted to individual \"sessions\" or discussions, though; they usually do not keep an eye on external data streams and will idle until the user asks another question.'}),/*#__PURE__*/e(\"h2\",{children:\"Ambient Agents: Always-On AI Assistants\"}),/*#__PURE__*/t(\"p\",{children:[\"By contrast, ambient agents run continuously in the background, monitoring various signals (logs, metrics, sensor data, workflow events, etc.) and \",/*#__PURE__*/e(n,{href:\"https://blog.langchain.com/introducing-ambient-agents\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"acting autonomously\"})}),' when predefined triggers occur. They are integrated into a company\\'s infrastructure and function similarly to \"always-on copilots.\" For instance, during periods of high utilization, an IT ambient agent may monitor system records and proactively restart overburdened servers without human intervention. An ambient agent in finance could identify and flag anomalous spending trends.\\xa0']}),/*#__PURE__*/e(\"p\",{children:\"These agents, in contrast to chatbots, take initiative and infer problems rather than waiting for an inquiry. In technical terms, ambient agents analyze incoming events against policies or trained models after subscribing to event streams (such as Kafka feeds, IoT sensor outputs, CI/CD hooks, or database updates). When an important event occurs, the agent chooses what to do, such as issuing an alert, contacting an API, or opening a ticket, based on its context (previous data and rules). Importantly, for safety, ambient agents frequently incorporate human-in-the-loop checkpoints (notifications, approval requests). To ensure that they only behave within authorized parameters, they are subject to regulations or ML policies.\"}),/*#__PURE__*/t(\"p\",{children:[\"Ambient agents therefore embody a push-based, context-rich model of AI: they maintain a memory of past events and system state, continuously integrating new data to update their context. They essentially apply the paradigm of the \\\"smart assistant\\\" to business procedures. For example, LangChain's ambient email assistant continuously searches a user's inbox, classifying messages and even automatically creating responses depending on preferences it has learnt. Agents that mimic remedies for recurring issues in SRE pipelines or cluster problematic test findings in QA procedures are further examples. Ambient agents are more proactive and multi-threaded than chatbots; one agent can handle dozens of concurrent events or activities. Their rise (the term was first used by LangChain in \",/*#__PURE__*/e(n,{href:\"https://zbrain.ai/ambient-agents/\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"2024\"})}),\") denotes a shift away from reactive or manual AI and toward systems that integrate intelligence into daily operations.\"]}),/*#__PURE__*/e(\"h2\",{children:\"Key Differences\"}),/*#__PURE__*/e(\"p\",{children:\"The table below summarizes the main distinctions between conversational (chat) agents and ambient agents:\"}),/*#__PURE__*/e(\"img\",{alt:\"\",className:\"framer-image\",height:\"921\",src:\"https://framerusercontent.com/images/SpxaZWqGvP6iwhnRnEmTPmaqvg.jpg?width=3246&height=1842\",srcSet:\"https://framerusercontent.com/images/SpxaZWqGvP6iwhnRnEmTPmaqvg.jpg?scale-down-to=512&width=3246&height=1842 512w,https://framerusercontent.com/images/SpxaZWqGvP6iwhnRnEmTPmaqvg.jpg?scale-down-to=1024&width=3246&height=1842 1024w,https://framerusercontent.com/images/SpxaZWqGvP6iwhnRnEmTPmaqvg.jpg?scale-down-to=2048&width=3246&height=1842 2048w,https://framerusercontent.com/images/SpxaZWqGvP6iwhnRnEmTPmaqvg.jpg?width=3246&height=1842 3246w\",style:{aspectRatio:\"3246 / 1842\"},width:\"1623\"}),/*#__PURE__*/e(\"p\",{children:'To put it briefly, ambient agents are excellent at automating back-end processes, while chatbots are best at one-on-one conversations. An ambient agent may identify a problem and then open a chat \"ticket\" or notification for a person to evaluate, demonstrating how they work in tandem. However, their primary modes are different: ambient agents operate after reading from their surroundings, whereas chat agents wait for input.\\xa0'}),/*#__PURE__*/e(\"h2\",{children:\"Technical Architecture\"}),/*#__PURE__*/e(\"p\",{children:\"Creating chat agents usually requires a conversational AI stack, which consists of a chat interface or bot framework and an NLP/LLM backend (such as GPT-4 or other models). These systems respond with text or voice outputs, manage dialogue state, and process user messages one at a time. They are very simple: if you can process a user's prompt and produce a response, you have a functional chatbot.\\xa0\"}),/*#__PURE__*/e(\"p\",{children:\"Ambient agents require a more complex event-driven infrastructure. Key components often include:\"}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Event Ingestion:\"}),\" Continuous feeds of events or signals (e.g., Kafka topics, system logs, IoT sensor streams, HTTP/webhook events).\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Context Builder:\"}),\" A service that aggregates relevant data (logs, metrics, past interactions) so the agent can maintain state and context over time.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Policy Engine:\"}),\" A rules or ML-based engine (e.g., OPA or custom logic) that defines what actions are allowed or when alerts should escalate. This constrains the agent\u2019s autonomy for safety.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Execution Environment:\"}),\" A sandbox or runtime where the agent can execute tasks (scripts, API calls, remedial actions).\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"AI Reasoning (LLM/Models):\"}),\" Often an LLM or other models that plan steps: when an event arrives, the agent forms a prompt with context and receives an action plan or analysis from the model.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Orchestration Layer:\"}),\" Logic to translate the model\u2019s plan into concrete actions (calls to APIs, infrastructure commands, etc.), or to route proposals for human approval.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Observability/Audit:\"}),\" Logging and monitoring of the agent\u2019s actions and effects, so humans can audit decisions.\"]}),/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Human-in-the-Loop Interface:\"}),\" An \u201Cagent inbox\u201D or notification system (e.g., Slack, email) where a human can review, approve, or override the agent\u2019s proposed actions.\"]}),/*#__PURE__*/e(\"p\",{children:\"In summary, ambient agent architecture consists of a human review channel, event streams, policy checks, a sandbox, context memory, and model inference. The agent may operate in silence thanks to this pipeline, which collects information, uses an LLM to reason, and then either takes action or requests confirmation from a human. An ambient agent is essentially a miniature control system in which artificial intelligence (AI) supports or replaces several manual monitoring and decision-making processes that were previously performed by humans.\"}),/*#__PURE__*/e(\"h2\",{children:\"Business Applications\"}),/*#__PURE__*/e(\"h3\",{children:\"1. Chat agents (Conversational AI)\\xa0\"}),/*#__PURE__*/e(\"p\",{children:'These are widely used for customer engagement and productivity. They use voice or chat to answer user questions (such as troubleshooting and FAQs) and tasks (such as booking and reminders). Businesses use chatbots for a variety of tasks, including sales, HR, and support. For example, many businesses that utilize conversational AI agents report increased efficiency in marketing and customer support. In reality, a chat assistant can provide rapid answers to frequently asked customer inquiries, walk customers through procedures, and even write code and content. Because human operators are constantly available to fix errors, they serve as a secure \"front-line\" for AI.'}),/*#__PURE__*/e(\"h3\",{children:\"2. Ambient agents\"}),/*#__PURE__*/e(\"p\",{children:\"These are targeted at operational automation and large-scale processes. Their ability to make systems more self-managing can revolutionize workflows in manufacturing, finance, and IT. For instance, during off-peak hours, a FinOps agent may automatically downscale idle cloud GPUs. Before people even wake up, a DevOps/SRE agent might replay recent system failures, identify the underlying reasons, and recommend solutions. LangChain's ambient agent handles email and productivity by organizing a user's inbox, setting priorities, and creating responses automatically.\\xa0\"}),/*#__PURE__*/e(\"p\",{children:\"Additional use cases include supply-chain monitors (which reorder products when thresholds are reached), code-review assistants (which automatically check pull requests for problems), and legal-compliance monitors (which scan contracts and highlight concerns).\"}),/*#__PURE__*/t(\"p\",{children:[\"Business-wise, this change allows teams to concentrate on strategy by automating repetitive tasks. \",/*#__PURE__*/e(n,{href:\"https://www.oracle.com/artificial-intelligence/ai-agents/ai-agent-use-cases\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Oracle\"})}),\", for instance, points out that by eliminating human tasks, AI agents, a broad category that includes ambient agents, can enhance procedures like financial forecasting, document analysis, and hiring.\"]}),/*#__PURE__*/e(\"h2\",{children:\"Integration and Complementarity\"}),/*#__PURE__*/t(\"p\",{children:[\"Ambient and chat agents often complement each other. For example, a chatbot may be used as the ambient system's interface. An employee could ask the chatbot, \\\"What's wrong with our server?\\\" and the study of the ambient agent (from logs) is summed back under the hood. On the other hand, when an ambient agent notices a problem, it may initiate a support ticket or send a chat message to a team member. \",/*#__PURE__*/e(n,{href:\"https://medium.com/@bijit211987/ambient-agents-e80d3c77518e\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"According to one article\"})}),', \"ambient agents are step two: AI that works with the system, for the humans, in the flow of processes, if the chatbot was step one.\" Although LLMs are at the heart of both paradigms, chatbots are designed for natural conversation and question-answering, whereas ambient agents are designed for action and planning across various data sources.']}),/*#__PURE__*/e(\"h2\",{children:\"Considerations and Risks\"}),/*#__PURE__*/e(\"h3\",{children:\"1. Chat Agents\\xa0\"}),/*#__PURE__*/e(\"p\",{children:\"Chat agents face several limitations tied to the nature of conversational AI. Accuracy is a big worry because big language models can confidently generate inaccurate answers, which is a major problem in delicate fields like healthcare or finance. It is also difficult to maintain context throughout lengthy or intricate interactions unless memory systems are integrated. Because chat interfaces have the potential to unintentionally expose sensitive information, privacy and data governance add even more complexity. Furthermore, chat agents rely heavily on prompts; the quality of their output is determined by the wording of inputs or the design of prompts. Continuous, timely tuning, training, and cautious safeguards to guarantee dependability are frequently necessary for effective deployment. Last but not least, consumer trust is crucial - bad or automated responses can drive users away, necessitating careful design and ongoing iteration.\"}),/*#__PURE__*/e(\"h3\",{children:\"2. Ambient Agents\"}),/*#__PURE__*/t(\"p\",{children:['Ambient agents, on the other hand, introduce new concerns. They can introduce hidden complexity since they behave independently. Debugging may be challenging because of their somewhat opaque decision logic (\"Why did the agent scale down the database?\"). If an agent\\'s model or rules are out of date, it may operate inappropriately because policies and data can drift out of alignment. Teams that rely too much on ambient agents run the danger of omitting important reviews. Security is also important since an agent could be tricked by harmful material, such as poisoned logs. Designers emphasize human oversight and auditability because of these factors. Ambient agents should include unambiguous fail-safes that need human consent for high-risk changes, stringent guardrails (rules), and full observability (recording every activity). According to one ',/*#__PURE__*/e(n,{href:\"https://medium.com/@bijit211987/ambient-agents-e80d3c77518e\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"expert\"})}),', they should be viewed as \"smart co-pilots\"\u2014strong automation that nevertheless permits human intervention.']}),/*#__PURE__*/e(\"p\",{children:\"In reality, companies that use ambient agents have to spend money on infrastructure and surveillance. They require AIOps tools, knowledge graphs, or databases for context, and streaming data platforms (Kafka, cloud event buses). This can result in far more automation, but it is a bigger lift than just putting a chatbot in place. Businesses should consider whether jobs are better managed interactively vs those that require continuous monitoring. According to several observers, it will become commonplace to combine the two, for instance, by using chat interfaces to monitor ambient agent activity.\"}),/*#__PURE__*/e(\"h2\",{children:\"Conclusion\"}),/*#__PURE__*/e(\"p\",{children:\"In 2025 and beyond, both chat and ambient agents will advance in parallel. As conversational AI advances (multimodal GPT-4o, improved memory), chatbots become more practical and organic. Ambient agents, meanwhile, represent a rapidly expanding field: Major labs and venture capital-backed businesses are developing frameworks (such as Microsoft's Copilot Studio, ZBrain, and LangGraph) to orchestrate event-driven agents at scale. Understanding the use case is crucial for business and product leaders. Ambient agents are the next great lever if you need to automate background processes and integrate AI deeply into your operations. Chat agents continue to be the preferred option if you must primarily serve or communicate with end users. A hybrid strategy will frequently be used, in which chat interfaces allow humans to interact with ambient agents or override them as necessary while ambient agents covertly do the hard lifting.\"}),/*#__PURE__*/e(\"p\",{children:\"In conclusion, ambient agents are a step up from chatbots, advancing AI from solely user-driven conversations to ongoing, context-aware automation. Though they necessitate careful design and supervision, they promise increased efficiency and additional possibilities (such as 24/7 self-healing systems). Businesses can benefit from the best of both worlds when well-designed ambient and chat agents collaborate: intelligent, conversational user interfaces for people and intelligent, automated procedures for the computers that support them.\"})]});export const richText16=/*#__PURE__*/t(o.Fragment,{children:[/*#__PURE__*/t(\"p\",{children:[\"AI, ZBrain. \u201CAmbient Agents Explained: Applications, Architecture, and Building with ZBrain.\u201D \",/*#__PURE__*/e(\"em\",{children:\"ZBrain\"}),\", 19 June 2025, zbrain.ai/ambient-agents/. \"]}),/*#__PURE__*/t(\"p\",{children:[\"Chase, Harrison. \u201CIntroducing Ambient Agents.\u201D \",/*#__PURE__*/e(\"em\",{children:\"LangChain Blog\"}),\", 14 Jan. 2025, blog.langchain.com/introducing-ambient-agents.\"]}),/*#__PURE__*/t(\"p\",{children:[\"\u201CConversational Agent.\u201D \",/*#__PURE__*/e(\"em\",{children:\"DeepAI\"}),\", 17 May 2019, deepai.org/machine-learning-glossary-and-terms/conversational-agent.\"]}),/*#__PURE__*/t(\"p\",{children:[\"Ghosh, Bijit. \u201CAmbient Agents - Bijit Ghosh - Medium.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Medium\"}),\", 9 May 2025, medium.com/@bijit211987/ambient-agents-e80d3c77518e. \"]}),/*#__PURE__*/t(\"p\",{children:[\"Ricadela, Aaron. \u201C23 Real-World AI Agent Use Cases.\u201D \",/*#__PURE__*/e(\"em\",{children:\"Oracle.com\"}),\", Oracle, 21 May 2025, www.oracle.com/artificial-intelligence/ai-agents/ai-agent-use-cases. \"]})]});export const richText17=/*#__PURE__*/e(o.Fragment,{children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(n,{href:\"https://www.walturn.com/author/abdullah-ahmed\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!0,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Abdullah Ahmed\"})}),\", \",/*#__PURE__*/e(n,{href:\"https://www.walturn.com/author/hashim-hayat\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Hashim Hayat\"})}),\", \",/*#__PURE__*/e(n,{href:\"https://www.walturn.com/author/daheem-hayat\",motionChild:!0,nodeId:\"RG6I9Jvqh\",openInNewTab:!1,relValues:[],scopeId:\"contentManagement\",smoothScroll:!1,children:/*#__PURE__*/e(i.a,{children:\"Daheem Hayat\"})})]})});export const richText18=/*#__PURE__*/e(o.Fragment,{children:/*#__PURE__*/t(\"ul\",{children:[/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Prompting is Linguistic Tuning\"}),\": Prompt engineering influences output using language phrasing, examples, and reasoning patterns.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Context is Systems Thinking\"}),\": Context engineering assembles tools, memory, and data to enrich AI\u2019s task-specific environment.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Prompting is Accessible but Brittle\"}),\": It's easy to use but sensitive to small changes, lacking persistence and generalization.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Context Enables Multi-Turn Reliability\"}),\": Context systems maintain state, reduce hallucinations, and scale across complex workflows.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"They\u2019re Complementary, Not Competing\"}),\": Combining strong prompting within robust context setups yields optimal AI results.\"]})}),/*#__PURE__*/e(\"li\",{\"data-preset-tag\":\"p\",children:/*#__PURE__*/t(\"p\",{children:[/*#__PURE__*/e(\"strong\",{children:\"Strategic Shift Toward Context\"}),\": Serious AI applications increasingly require context engineering for accuracy and alignment.\"]})})]})});\nexport const __FramerMetadata__ = {\"exports\":{\"richText4\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText6\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText5\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText14\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText12\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText7\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText18\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText9\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText15\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText2\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText10\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText11\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText8\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText16\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText3\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText13\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText17\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"richText1\":{\"type\":\"variable\",\"annotations\":{\"framerContractVersion\":\"1\"}},\"__FramerMetadata__\":{\"type\":\"variable\"}}}"],
  "mappings": "uUAAgS,IAAMA,EAAsBC,EAAIC,EAAS,CAAC,SAAsBC,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAEG,EAAE,CAAC,KAAK,wDAAwD,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,gBAAgB,CAAC,CAAC,CAAC,EAAE,KAAkBJ,EAAEG,EAAE,CAAC,KAAK,8CAA8C,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,cAAc,CAAC,CAAC,CAAC,EAAE,KAAkBJ,EAAEG,EAAE,CAAC,KAAK,8CAA8C,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,cAAc,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeC,EAAuBL,EAAIC,EAAS,CAAC,SAAsBC,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,oBAAoB,CAAC,EAAE,4GAA4G,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,kBAAkB,CAAC,EAAE,sFAAsF,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,qBAAqB,CAAC,EAAE,6FAAwF,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,qBAAqB,CAAC,EAAE,uFAAuF,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,kBAAkB,CAAC,EAAE,qGAAqG,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,YAAY,CAAC,EAAE,kGAAkG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeM,EAAuBJ,EAAID,EAAS,CAAC,SAAS,CAAcD,EAAE,KAAK,CAAC,SAAS,cAAc,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,wwBAAwwB,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,2BAAwCF,EAAEG,EAAE,CAAC,KAAK,+CAA+C,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,oBAAoB,CAAC,CAAC,CAAC,EAAE,6MAA6M,CAAC,CAAC,EAAeF,EAAE,IAAI,CAAC,SAAS,CAAC,kOAA+OF,EAAEG,EAAE,CAAC,KAAK,sBAAsB,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,4GAA4G,CAAC,CAAC,CAAC,EAAE,4CAA0DJ,EAAEG,EAAE,CAAC,KAAK,+CAA+C,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,0BAA0B,CAAC,CAAC,CAAC,EAAE,qNAAqN,CAAC,CAAC,EAAeJ,EAAE,KAAK,CAAC,SAAS,uCAAuC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,kvBAAiwBF,EAAEG,EAAE,CAAC,KAAK,+CAA+C,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,QAAQ,CAAC,CAAC,CAAC,EAAE,iBAAiB,CAAC,CAAC,EAAeF,EAAE,IAAI,CAAC,SAAS,CAAC,gIAA6IF,EAAEG,EAAE,CAAC,KAAK,+CAA+C,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,SAAS,CAAC,CAAC,CAAC,EAAE,gkBAAgkB,CAAC,CAAC,EAAeJ,EAAE,IAAI,CAAC,SAAS,iiBAAkiB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,6BAA6B,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,8cAA8c,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,snBAAsnB,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,yCAAiDF,EAAEG,EAAE,CAAC,KAAK,6DAA6D,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,iBAAiB,CAAC,CAAC,CAAC,EAAE,iEAA4D,CAAC,CAAC,EAAeJ,EAAE,KAAK,CAAC,SAAS,4BAA4B,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,+SAA+S,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,yBAAyB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,iUAAiU,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,gCAAgC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,sWAAsW,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,0BAA0B,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,0WAA0W,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,wSAAwS,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,gDAA2C,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,4HAA4H,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,0CAA0C,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,ulBAAulB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,oCAAoC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,srBAAwrB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,iCAAiC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,8tBAA8tB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,qCAAqC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,kpBAAopB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,uBAAuB,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,8YAA4ZF,EAAEG,EAAE,CAAC,KAAK,sBAAsB,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,oEAA+D,CAAC,CAAC,CAAC,EAAE,qIAAqI,CAAC,CAAC,EAAeJ,EAAE,IAAI,CAAC,SAAS,6sBAA6sB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,qBAAqB,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,oXAAiYF,EAAEG,EAAE,CAAC,KAAK,mDAAmD,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,0EAA0E,CAAC,CAAC,CAAC,EAAE,mNAAmN,CAAC,CAAC,EAAeJ,EAAE,IAAI,CAAC,SAAS,uEAAuE,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,kBAAkB,CAAC,EAAE,oNAAoN,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,oBAAoB,CAAC,EAAE,kRAAkR,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,sBAAsB,CAAC,EAAE,sPAAsP,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,oBAAoB,CAAC,EAAE,sRAAsR,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,4UAAwU,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,8CAAoC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,oTAAqT,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,uDAAoEF,EAAEG,EAAE,CAAC,KAAK,0DAA0D,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,QAAQ,CAAC,CAAC,CAAC,EAAE,0LAAuMJ,EAAEG,EAAE,CAAC,KAAK,0DAA0D,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,4BAA4B,CAAC,CAAC,CAAC,EAAE,kLAAkL,CAAC,CAAC,EAAeJ,EAAE,IAAI,CAAC,SAAS,ubAAub,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,YAAY,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,mgBAAmgB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,kkBAAkkB,CAAC,CAAC,CAAC,CAAC,EAAeO,EAAuBL,EAAID,EAAS,CAAC,SAAS,CAAcC,EAAE,IAAI,CAAC,SAAS,CAAC,0FAA6FF,EAAE,KAAK,CAAC,SAAS,SAAS,CAAC,EAAE,qEAAqE,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,2GAAkGF,EAAE,KAAK,CAAC,SAAS,mBAAmB,CAAC,EAAE,wGAAwG,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,uBAA0BF,EAAE,KAAK,CAAC,SAAS,YAAY,CAAC,EAAE,uBAAuB,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,uEAA0EF,EAAE,KAAK,CAAC,SAAS,yBAAyB,CAAC,EAAE,oDAAoD,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,iGAAoGF,EAAE,KAAK,CAAC,SAAS,cAAc,CAAC,EAAE,+CAA+C,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeQ,EAAuBR,EAAIC,EAAS,CAAC,SAAsBC,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,sBAAsB,CAAC,EAAE,yHAAyH,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,iBAAiB,CAAC,EAAE,sGAAiG,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,eAAe,CAAC,EAAE,4GAA4G,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,iBAAiB,CAAC,EAAE,8FAA8F,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,gBAAgB,CAAC,EAAE,yFAAyF,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,qBAAqB,CAAC,EAAE,+FAA+F,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeS,EAAuBP,EAAID,EAAS,CAAC,SAAS,CAAcD,EAAE,KAAK,CAAC,SAAS,cAAc,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,oZAAoZ,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,8bAA8b,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,iCAAiC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAEG,EAAE,CAAC,KAAK,0BAA0B,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,aAAa,CAAC,CAAC,CAAC,EAAE,siBAA4hB,CAAC,CAAC,EAAeJ,EAAE,IAAI,CAAC,SAAS,upBAA6oB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,oPAAoP,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,0tBAA0tB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,0DAAqD,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAEG,EAAE,CAAC,KAAK,yBAAyB,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,WAAW,CAAC,CAAC,CAAC,EAAE,oaAA0Z,CAAC,CAAC,EAAeF,EAAE,IAAI,CAAC,SAAS,CAAC,8SAA2TF,EAAEG,EAAE,CAAC,KAAK,kGAAkG,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,gBAAgB,CAAC,CAAC,CAAC,EAAE,iZAAiZ,CAAC,CAAC,EAAeF,EAAE,IAAI,CAAC,SAAS,CAAC,uIAAoJF,EAAEG,EAAE,CAAC,KAAK,gCAAgC,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,WAAW,CAAC,CAAC,CAAC,EAAE,26BAA26B,CAAC,CAAC,EAAeJ,EAAE,KAAK,CAAC,SAAS,yCAAyC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,YAAyBF,EAAEG,EAAE,CAAC,KAAK,sBAAsB,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,aAAa,CAAC,CAAC,CAAC,EAAE,qYAAmZJ,EAAEG,EAAE,CAAC,KAAK,gDAAgD,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,WAAW,CAAC,CAAC,CAAC,EAAE,8KAA8K,CAAC,CAAC,EAAeJ,EAAE,IAAI,CAAC,SAAS,ixBAAixB,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,8hBAAsiBF,EAAEG,EAAE,CAAC,KAAK,wEAAwE,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,SAAS,CAAC,CAAC,CAAC,EAAE,4XAA4X,CAAC,CAAC,EAAeJ,EAAE,KAAK,CAAC,SAAS,6BAA6B,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAEG,EAAE,CAAC,KAAK,2BAA2B,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,MAAM,CAAC,CAAC,CAAC,EAAE,sgBAAsgB,CAAC,CAAC,EAAeJ,EAAE,IAAI,CAAC,SAAS,6uBAA6uB,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,yDAA4DF,EAAEG,EAAE,CAAC,KAAK,kCAAkC,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,aAAa,CAAC,CAAC,CAAC,EAAE,uiCAAwiC,CAAC,CAAC,EAAeJ,EAAE,KAAK,CAAC,SAAS,mBAAmB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,aAAa,CAAC,EAAeA,EAAE,MAAM,CAAC,IAAI,GAAG,UAAU,eAAe,OAAO,OAAO,IAAI,8FAA8F,OAAO,icAAic,MAAM,CAAC,YAAY,aAAa,EAAE,MAAM,MAAM,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,6BAA6B,CAAC,EAAeA,EAAE,MAAM,CAAC,IAAI,GAAG,UAAU,eAAe,OAAO,OAAO,IAAI,8FAA8F,OAAO,icAAic,MAAM,CAAC,YAAY,aAAa,EAAE,MAAM,MAAM,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,YAAY,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,2tBAA2tB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,guBAAguB,CAAC,CAAC,CAAC,CAAC,EAAeU,EAAuBR,EAAID,EAAS,CAAC,SAAS,CAAcC,EAAE,IAAI,CAAC,SAAS,CAAC,wDAA2DF,EAAE,KAAK,CAAC,SAAS,MAAM,CAAC,EAAE,kDAAkD,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,iGAA+FF,EAAE,KAAK,CAAC,SAAS,wBAAwB,CAAC,EAAE,6HAA6H,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,gEAAmEF,EAAE,KAAK,CAAC,SAAS,eAAe,CAAC,EAAE,gDAAgD,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,wBAA2BF,EAAE,KAAK,CAAC,SAAS,aAAa,CAAC,EAAE,qDAAqD,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,gDAAmDF,EAAE,KAAK,CAAC,SAAS,aAAa,CAAC,EAAE,8FAA8F,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeW,EAAuBX,EAAIC,EAAS,CAAC,SAAsBC,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,wBAAwB,CAAC,EAAE,wFAAwF,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,sBAAsB,CAAC,EAAE,qGAAqG,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,6BAA6B,CAAC,EAAE,kGAAkG,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,kCAAkC,CAAC,EAAE,mGAAmG,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,yBAAyB,CAAC,EAAE,sGAAsG,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,oBAAoB,CAAC,EAAE,uHAAuH,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeY,EAAuBV,EAAID,EAAS,CAAC,SAAS,CAAcD,EAAE,KAAK,CAAC,SAAS,cAAc,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,0eAA0e,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,wmBAAwmB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,2BAA2B,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,eAAe,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,kVAA+VF,EAAEG,EAAE,CAAC,KAAK,8CAA8C,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,yBAAyB,CAAC,CAAC,CAAC,EAAE,+RAA+R,CAAC,CAAC,EAAeJ,EAAE,IAAI,CAAC,SAAS,kFAAkF,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,qBAAqB,CAAC,EAAE,iMAAiM,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,kBAAkB,CAAC,EAAE,iRAAiR,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,0BAA0B,CAAC,EAAE,iZAAiZ,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,yBAAyB,CAAC,EAAE,4PAA4P,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,yBAAyB,CAAC,EAAE,mZAAmZ,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,gBAAgB,CAAC,EAAE,6bAA6b,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,sBAAsB,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,kXAA0XF,EAAEG,EAAE,CAAC,KAAK,2EAA2E,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,+CAA+C,CAAC,CAAC,CAAC,EAAE,kYAAkY,CAAC,CAAC,EAAeF,EAAE,IAAI,CAAC,SAAS,CAAC,sfAAmgBF,EAAEG,EAAE,CAAC,KAAK,2EAA2E,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,kDAAkD,CAAC,CAAC,CAAC,EAAE,yGAAyG,CAAC,CAAC,EAAeJ,EAAE,KAAK,CAAC,SAAS,iBAAiB,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,+BAA+B,CAAC,EAAE,sRAAmSA,EAAEG,EAAE,CAAC,KAAK,+CAA+C,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,wGAAwG,CAAC,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,EAAeF,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,kBAAkB,CAAC,EAAE,yVAAyV,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,yBAAyB,CAAC,EAAE,sZAAsZ,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,oCAAoC,CAAC,EAAE,wVAAwV,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,gCAAgC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,0HAA0H,CAAC,EAAeA,EAAE,MAAM,CAAC,IAAI,GAAG,UAAU,eAAe,OAAO,MAAM,IAAI,4FAA4F,OAAO,ybAAyb,MAAM,CAAC,YAAY,aAAa,EAAE,MAAM,MAAM,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,gBAAgB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,mFAAmF,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,mBAAmB,CAAC,EAAE,yMAAyM,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,iBAAiB,CAAC,EAAE,yOAAyO,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,qBAAqB,CAAC,EAAE,6TAA6T,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,kBAAkB,CAAC,EAAE,+WAA+W,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,sBAAsB,CAAC,EAAE,yVAAyV,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,qRAAqR,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,yBAAyB,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,uCAAoDF,EAAE,SAAS,CAAC,SAAS,QAAQ,CAAC,EAAE,wNAAqOA,EAAEG,EAAE,CAAC,KAAK,iCAAiC,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAsBJ,EAAE,KAAK,CAAC,SAAS,uDAAuD,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,2PAA2P,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,0kBAA0kB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,4lBAA4lB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,2CAA2C,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,2aAA2a,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,gQAA6QF,EAAEG,EAAE,CAAC,KAAK,+GAA+G,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,MAAM,CAAC,CAAC,CAAC,EAAE,iXAAiX,CAAC,CAAC,EAAeJ,EAAE,IAAI,CAAC,SAAS,6mBAA+mB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,YAAY,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,ohBAAohB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,gnBAAinB,CAAC,CAAC,CAAC,CAAC,EAAea,EAAuBX,EAAID,EAAS,CAAC,SAAS,CAAcC,EAAE,IAAI,CAAC,SAAS,CAAC,wDAA2DF,EAAE,KAAK,CAAC,SAAS,+CAA+C,CAAC,EAAE,8EAA8E,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,8FAAiGF,EAAE,KAAK,CAAC,SAAS,YAAY,CAAC,EAAE,gEAAgE,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,iGAAoGF,EAAE,KAAK,CAAC,SAAS,SAAS,CAAC,EAAE,yGAAyG,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,kEAAqEF,EAAE,KAAK,CAAC,SAAS,WAAW,CAAC,EAAE,kGAAkG,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,iFAAoFF,EAAE,KAAK,CAAC,SAAS,cAAc,CAAC,EAAE,wCAAwC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAec,EAAwBd,EAAIC,EAAS,CAAC,SAAsBC,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAEG,EAAE,CAAC,KAAK,+CAA+C,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,eAAe,CAAC,CAAC,CAAC,EAAE,KAAkBJ,EAAEG,EAAE,CAAC,KAAK,8CAA8C,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,cAAc,CAAC,CAAC,CAAC,EAAE,KAAkBJ,EAAEG,EAAE,CAAC,KAAK,8CAA8C,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,cAAc,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeW,EAAwBf,EAAIC,EAAS,CAAC,SAAsBC,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,0BAA0B,CAAC,EAAE,0GAA0G,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,8BAA8B,CAAC,EAAE,uGAAkG,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,8BAA8B,CAAC,EAAE,+FAA+F,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,8BAA8B,CAAC,EAAE,8FAA8F,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,gCAAgC,CAAC,EAAE,6FAA6F,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,iCAAiC,CAAC,EAAE,iGAAiG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAegB,EAAwBd,EAAID,EAAS,CAAC,SAAS,CAAcD,EAAE,KAAK,CAAC,SAAS,cAAc,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,wgBAAyf,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,2aAA2a,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,qVAAqV,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,iBAAiB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,8YAA8Y,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,gFAA6FF,EAAEG,EAAE,CAAC,KAAK,wCAAwC,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,MAAM,CAAC,CAAC,CAAC,EAAE,0CAAuDJ,EAAEG,EAAE,CAAC,KAAK,wCAAwC,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,OAAO,CAAC,CAAC,CAAC,EAAE,+VAA+V,CAAC,CAAC,EAAeF,EAAE,IAAI,CAAC,SAAS,CAAC,0BAAuCF,EAAE,SAAS,CAAC,SAAS,YAAY,CAAC,EAAE,iKAA8KA,EAAE,SAAS,CAAC,SAAS,WAAW,CAAC,EAAE,QAAqBA,EAAE,SAAS,CAAC,SAAS,UAAU,CAAC,EAAE,+IAA+I,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,8aAA8a,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,kTAAkT,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,aAAa,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,wnBAAwnB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,wBAAwB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,wPAAwP,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,WAAW,CAAC,EAAE,2NAA2N,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,UAAU,CAAC,EAAE,6QAA6Q,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,WAAW,CAAC,EAAE,qPAAqP,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,aAAa,CAAC,EAAE,sQAAiQ,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,MAAM,CAAC,EAAE,gQAAgQ,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,wBAAwB,CAAC,EAAE,8PAA8P,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,aAAa,CAAC,EAAE,gOAAgO,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,gBAAgB,CAAC,EAAE,oFAAiGA,EAAE,KAAK,CAAC,SAAS,KAAK,CAAC,EAAE,0KAA0K,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,gMAAgM,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,uBAAuB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,yNAAyN,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,kBAAkB,CAAC,EAAE,6YAA6Y,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,mBAAmB,CAAC,EAAE,8lBAAylB,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,mBAAmB,CAAC,EAAE,sSAAsS,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,gCAAgC,CAAC,EAAE,qUAAqU,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,0OAA0O,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,aAAa,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,oTAA4TF,EAAEG,EAAE,CAAC,KAAK,uCAAuC,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,kBAAkB,CAAC,CAAC,CAAC,EAAE,KAAkBJ,EAAEG,EAAE,CAAC,KAAK,0CAA0C,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,8BAA8B,CAAC,CAAC,CAAC,EAAE,SAAsBJ,EAAEG,EAAE,CAAC,KAAK,4BAA4B,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,0BAA0B,CAAC,CAAC,CAAC,EAAE,iFAAiF,CAAC,CAAC,EAAeJ,EAAE,KAAK,CAAC,SAAS,8BAA8B,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,4nBAAknB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,YAAY,CAAC,EAAeA,EAAE,MAAM,CAAC,IAAI,GAAG,UAAU,eAAe,OAAO,MAAM,IAAI,4FAA4F,OAAO,ybAAyb,MAAM,CAAC,YAAY,aAAa,EAAE,MAAM,MAAM,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,2BAA2B,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,wLAAqMF,EAAE,KAAK,CAAC,SAAS,0EAAgE,CAAC,EAAE,+KAA4LA,EAAE,KAAK,CAAC,SAAS,0EAAgE,CAAC,EAAE,wIAAqJA,EAAE,KAAK,CAAC,SAAS,6EAAmE,CAAC,EAAE,2NAAwOA,EAAE,KAAK,CAAC,SAAS,wEAA8D,CAAC,EAAE,6HAAwH,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,SAAS,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,kXAAkX,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,qBAAqB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,QAAQ,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,2GAA2G,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,iQAAiQ,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,uUAAuU,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAC,sLAAmMF,EAAE,SAAS,CAAC,SAAS,qBAAqB,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,yLAAyL,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,+NAAqN,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,gLAAgL,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,iBAAiB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,eAAe,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,sNAAiN,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,4UAAkU,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,sYAAsY,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,uYAAkY,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,mYAA8X,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,0TAA0T,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,wZAAmZ,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,yOAAyO,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,gNAA2M,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,gBAAgB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,oOAAoO,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,wbAA8a,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,sSAAsS,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,4TAA4T,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,+PAA+P,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,0aAA0a,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,uBAAuB,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,wEAAqFF,EAAE,KAAK,CAAC,SAAS,QAAG,CAAC,EAAeA,EAAEG,EAAE,CAAC,KAAK,gGAAgG,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAsBJ,EAAE,KAAK,CAAC,SAAS,kCAAkC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,SAAI,CAAC,EAAE,oRAA+Q,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,QAAQ,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,kIAAyG,CAAC,CAAC,CAAC,EAAeE,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,oHAA+G,CAAC,EAAE,0OAAgO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,8DAA8D,CAAC,EAAE,iOAAiO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,yEAAyE,CAAC,EAAE,gQAAsP,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAsBA,EAAE,SAAS,CAAC,SAAS,iBAAiB,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAsBA,EAAE,SAAS,CAAC,SAAS,gEAAgE,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAsBA,EAAE,SAAS,CAAC,SAAS,gHAAgH,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,oNAAoN,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,kEAAkE,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,qMAAqM,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAsBA,EAAE,SAAS,CAAC,SAAS,0CAA0C,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,yKAAyK,CAAC,CAAC,CAAC,EAAeE,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,mCAAmC,CAAC,EAAE,kSAAkS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,sDAAsD,CAAC,EAAE,8NAA8N,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,iCAAiC,CAAC,EAAE,gWAAgW,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,gBAAgB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,0DAA0D,CAAC,CAAC,CAAC,EAAeE,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,kCAAkC,CAAC,EAAE,uYAAuY,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,kDAAkD,CAAC,EAAE,qeAAue,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,sCAAsC,CAAC,EAAE,qbAAqb,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,gBAAgB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,QAAQ,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAC,iDAAyDF,EAAE,SAAS,CAAC,SAAS,sBAAsB,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAC,oBAAiCF,EAAE,SAAS,CAAC,SAAS,gBAAgB,CAAC,EAAE,gGAAmGA,EAAE,SAAS,CAAC,SAAS,6DAA6D,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAC,yCAAsDF,EAAE,SAAS,CAAC,SAAS,6BAA6B,CAAC,EAAE,8HAAoH,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAC,2CAAwDF,EAAE,SAAS,CAAC,SAAS,sBAAsB,CAAC,EAAE,gEAA2D,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,iBAAiB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,uPAAuP,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,+TAAkU,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,oSAAsS,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,4NAA4N,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,uOAAuO,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,gBAAgB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,0GAA0G,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,4BAAqB,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAC,qDAAkEF,EAAE,SAAS,CAAC,SAAS,MAAM,CAAC,EAAE,8BAA2CA,EAAE,SAAS,CAAC,SAAS,OAAO,CAAC,EAAE,SAAS,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,wLAAiL,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAC,oBAAiCF,EAAE,SAAS,CAAC,SAAS,WAAW,CAAC,EAAE,4KAA6K,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAC,yDAAsEF,EAAE,SAAS,CAAC,SAAS,MAAM,CAAC,EAAE,eAA4BA,EAAE,SAAS,CAAC,SAAS,WAAW,CAAC,EAAE,kCAA+CA,EAAE,SAAS,CAAC,SAAS,OAAO,CAAC,EAAE,0HAA0H,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,8BAA8B,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,sBAAsB,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,mGAAyF,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBE,EAAE,KAAK,CAAC,SAAS,CAAC,uIAAoJF,EAAE,SAAS,CAAC,SAAS,WAAW,CAAC,EAAE,mBAAgCA,EAAE,SAAS,CAAC,SAAS,WAAW,CAAC,EAAE,2EAA2E,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAsBA,EAAE,KAAK,CAAC,SAAS,uHAAuH,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,oBAAoB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,QAAQ,CAAC,EAAeA,EAAE,MAAM,CAAC,UAAU,qBAAqB,MAAM,CAAC,OAAO,OAAO,MAAM,MAAM,EAAE,SAAsBA,EAAEiB,EAAE,CAAC,oBAAoB,wEAAwE,SAASC,GAAgBlB,EAAEmB,EAAE,CAAC,GAAGD,EAAE,KAAK;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,0BAAmlB,SAAS,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,EAAelB,EAAE,KAAK,CAAC,SAAS,iBAAiB,CAAC,EAAeA,EAAE,MAAM,CAAC,UAAU,qBAAqB,MAAM,CAAC,OAAO,OAAO,MAAM,MAAM,EAAE,SAAsBA,EAAEiB,EAAE,CAAC,oBAAoB,wEAAwE,SAASC,GAAgBlB,EAAEmB,EAAE,CAAC,GAAGD,EAAE,KAAK;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,wBAAusH,SAAS,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,EAAelB,EAAE,KAAK,CAAC,SAAS,gBAAgB,CAAC,EAAeA,EAAE,MAAM,CAAC,UAAU,qBAAqB,MAAM,CAAC,OAAO,OAAO,MAAM,MAAM,EAAE,SAAsBA,EAAEiB,EAAE,CAAC,oBAAoB,wEAAwE,SAASC,GAAgBlB,EAAEmB,EAAE,CAAC,GAAGD,EAAE,KAAK;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,mBAAouC,SAAS,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,EAAelB,EAAE,KAAK,CAAC,SAAS,UAAU,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,qBAAqB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,isBAAurB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,+gBAA0gB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,uBAAuB,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,2GAAwHF,EAAE,KAAK,CAAC,SAAS,+CAAqC,CAAC,EAAE,+dAA+d,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,gBAAgB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,kjBAAyhB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,6jBAAmjB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,oBAAoB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,okBAA0jB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,YAAY,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,uhBAAuhB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,khBAA6gB,CAAC,CAAC,CAAC,CAAC,EAAeoB,EAAwBlB,EAAID,EAAS,CAAC,SAAS,CAAcD,EAAE,IAAI,CAAC,SAAS,gJAAsI,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,kHAAkH,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,gNAAsM,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,qIAAqI,CAAC,CAAC,CAAC,CAAC,EAAeqB,EAAwBrB,EAAIC,EAAS,CAAC,SAAsBC,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,6BAA6B,CAAC,EAAE,0EAA0E,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,2BAA2B,CAAC,EAAE,iFAAiF,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,mBAAmB,CAAC,EAAE,qFAAqF,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,wBAAwB,CAAC,EAAE,+FAA+F,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,gBAAgB,CAAC,EAAE,yFAAyF,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,sBAAsB,CAAC,EAAE,0EAA0E,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAesB,EAAwBpB,EAAID,EAAS,CAAC,SAAS,CAAcD,EAAE,KAAK,CAAC,SAAS,cAAc,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,iyBAAiyB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,gCAAgC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,oaAAkbF,EAAEG,EAAE,CAAC,KAAK,8EAA8E,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,iEAA4D,CAAC,CAAC,CAAC,EAAE,kZAAkZ,CAAC,CAAC,EAAeJ,EAAE,IAAI,CAAC,SAAS,4iBAA4iB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,yCAAyC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,sJAAmKF,EAAEG,EAAE,CAAC,KAAK,wDAAwD,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,qBAAqB,CAAC,CAAC,CAAC,EAAE,mYAAoY,CAAC,CAAC,EAAeJ,EAAE,IAAI,CAAC,SAAS,4tBAA4tB,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,sxBAAqyBF,EAAEG,EAAE,CAAC,KAAK,oCAAoC,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,MAAM,CAAC,CAAC,CAAC,EAAE,yHAAyH,CAAC,CAAC,EAAeJ,EAAE,KAAK,CAAC,SAAS,iBAAiB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,2GAA2G,CAAC,EAAeA,EAAE,MAAM,CAAC,IAAI,GAAG,UAAU,eAAe,OAAO,MAAM,IAAI,6FAA6F,OAAO,6bAA6b,MAAM,CAAC,YAAY,aAAa,EAAE,MAAM,MAAM,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,ibAAib,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,wBAAwB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,oZAAoZ,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,kGAAkG,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,kBAAkB,CAAC,EAAE,oHAAoH,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,kBAAkB,CAAC,EAAE,oIAAoI,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,gBAAgB,CAAC,EAAE,qLAAgL,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,wBAAwB,CAAC,EAAE,iGAAiG,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,4BAA4B,CAAC,EAAE,qKAAqK,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,sBAAsB,CAAC,EAAE,2JAAsJ,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,sBAAsB,CAAC,EAAE,iGAA4F,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,8BAA8B,CAAC,EAAE,2JAA4I,CAAC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,miBAAmiB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,uBAAuB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,wCAAwC,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,kqBAAkqB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,mBAAmB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,6jBAA6jB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,sQAAsQ,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,sGAAmHF,EAAEG,EAAE,CAAC,KAAK,8EAA8E,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,QAAQ,CAAC,CAAC,CAAC,EAAE,yMAAyM,CAAC,CAAC,EAAeJ,EAAE,KAAK,CAAC,SAAS,iCAAiC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,qZAAoaF,EAAEG,EAAE,CAAC,KAAK,8DAA8D,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,0BAA0B,CAAC,CAAC,CAAC,EAAE,0VAA0V,CAAC,CAAC,EAAeJ,EAAE,KAAK,CAAC,SAAS,0BAA0B,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,oBAAoB,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,q7BAAq7B,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,mBAAmB,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,y1BAAu2BF,EAAEG,EAAE,CAAC,KAAK,8DAA8D,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,QAAQ,CAAC,CAAC,CAAC,EAAE,mHAA8G,CAAC,CAAC,EAAeJ,EAAE,IAAI,CAAC,SAAS,2lBAA2lB,CAAC,EAAeA,EAAE,KAAK,CAAC,SAAS,YAAY,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,w6BAAw6B,CAAC,EAAeA,EAAE,IAAI,CAAC,SAAS,+hBAA+hB,CAAC,CAAC,CAAC,CAAC,EAAeuB,EAAwBrB,EAAID,EAAS,CAAC,SAAS,CAAcC,EAAE,IAAI,CAAC,SAAS,CAAC,2GAA8GF,EAAE,KAAK,CAAC,SAAS,QAAQ,CAAC,EAAE,6CAA6C,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,4DAA+DF,EAAE,KAAK,CAAC,SAAS,gBAAgB,CAAC,EAAE,gEAAgE,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,qCAAwCF,EAAE,KAAK,CAAC,SAAS,QAAQ,CAAC,EAAE,qFAAqF,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,oEAAuEF,EAAE,KAAK,CAAC,SAAS,QAAQ,CAAC,EAAE,qEAAqE,CAAC,CAAC,EAAeE,EAAE,IAAI,CAAC,SAAS,CAAC,kEAAqEF,EAAE,KAAK,CAAC,SAAS,YAAY,CAAC,EAAE,8FAA8F,CAAC,CAAC,CAAC,CAAC,CAAC,EAAewB,EAAwBxB,EAAIC,EAAS,CAAC,SAAsBC,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAEG,EAAE,CAAC,KAAK,gDAAgD,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,gBAAgB,CAAC,CAAC,CAAC,EAAE,KAAkBJ,EAAEG,EAAE,CAAC,KAAK,8CAA8C,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,cAAc,CAAC,CAAC,CAAC,EAAE,KAAkBJ,EAAEG,EAAE,CAAC,KAAK,8CAA8C,YAAY,GAAG,OAAO,YAAY,aAAa,GAAG,UAAU,CAAC,EAAE,QAAQ,oBAAoB,aAAa,GAAG,SAAsBH,EAAEI,EAAE,EAAE,CAAC,SAAS,cAAc,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAeqB,EAAwBzB,EAAIC,EAAS,CAAC,SAAsBC,EAAE,KAAK,CAAC,SAAS,CAAcF,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,gCAAgC,CAAC,EAAE,mGAAmG,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,6BAA6B,CAAC,EAAE,wGAAmG,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,qCAAqC,CAAC,EAAE,4FAA4F,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,wCAAwC,CAAC,EAAE,8FAA8F,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,2CAAsC,CAAC,EAAE,sFAAsF,CAAC,CAAC,CAAC,CAAC,EAAeA,EAAE,KAAK,CAAC,kBAAkB,IAAI,SAAsBE,EAAE,IAAI,CAAC,SAAS,CAAcF,EAAE,SAAS,CAAC,SAAS,gCAAgC,CAAC,EAAE,gGAAgG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAC72qI0B,EAAqB,CAAC,QAAU,CAAC,UAAY,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,UAAY,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,UAAY,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,WAAa,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,WAAa,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,UAAY,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,WAAa,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,UAAY,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,WAAa,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,UAAY,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,WAAa,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,WAAa,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,UAAY,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,WAAa,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,UAAY,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,SAAW,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,WAAa,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,WAAa,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,UAAY,CAAC,KAAO,WAAW,YAAc,CAAC,sBAAwB,GAAG,CAAC,EAAE,mBAAqB,CAAC,KAAO,UAAU,CAAC,CAAC",
  "names": ["richText", "p", "x", "u", "Link", "motion", "richText1", "richText2", "richText3", "richText4", "richText5", "richText6", "richText7", "richText8", "richText9", "richText10", "richText11", "richText12", "ComponentPresetsConsumer", "t", "CodeBlock_default", "richText13", "richText14", "richText15", "richText16", "richText17", "richText18", "__FramerMetadata__"]
}