--- title: Building a Speaker Arena With Kilo, Jev, and Enkrypt description: A developer built the 'Speaker Arena', a browser-based 3D conference world with an AI guide, for the All Day AI event that drew 12,313 registrants and 140... canonical: https://daily.dev/posts/building-a-speaker-arena-with-kilo-jev-and-enkrypt-nhx30yrfr twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Building a Speaker Arena With Kilo, Jev, and Enkrypt | daily.dev og:description: A developer built the 'Speaker Arena', a browser-based 3D conference world with an AI guide, for the All Day AI event that drew 12,313 registrants and 140... og:url: https://daily.dev/posts/building-a-speaker-arena-with-kilo-jev-and-enkrypt-nhx30yrfr og:image: https://api.daily.dev/og/posts/NHX30YRfr.png og:image:alt: Building a Speaker Arena With Kilo, Jev, and Enkrypt og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Building a Speaker Arena With Kilo, Jev, and Enkrypt **[Kilo Blog](https://daily.dev/sources/kilo-ai-blog)** · 7 min read · 2 upvotes · 0 comments ## Summary A developer built the 'Speaker Arena', a browser-based 3D conference world with an AI guide, for the All Day AI event that drew 12,313 registrants and 140 speakers. He used Kilo as his coding agent in VS Code, connecting his OpenAI subscription to iterate from a 2D prototype into a Three.js 3D park. The live app split model duties: Jev, a small decision model, classified questions as on-topic or off-topic, while Gemini 2.5 Flash generated answers, both routed through Kilo Gateway with one server-side key. Enkrypt AI enforced a 'Code of Conduct' policy through input and output guardrails, red-teaming, and logging, and the author disabled an overly broad PII detector that flagged public speaker names. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How can I combine a small classification model with a larger generation model to reduce LLM costs? Route the question to a small decision model first to classify it as relevant or off-topic, then only call the larger generation model for relevant questions. In this setup, a small model called Jev returned either 'talk_question' or 'off_topic', skipping the Gemini 2.5 Flash answer call entirely for off-topic requests to save on generation cost and latency. _Developers weighing model-routing strategies for cost control can track real implementations like this on daily.dev._ ### How do I stop an LLM guardrail from flagging public figures' names as PII? Disable the overly broad PII detector for cases where public names and organizations are expected content. During live testing of an AI conference guide, broad PII settings flagged benign speaker-related content such as public names, so the guardrail was adjusted by disabling that detector and then retested against normal questions, off-topic requests, and injection attempts. _Anyone tuning guardrail false positives can follow real-world fixes like this on daily.dev._ ### What is Kilo Gateway and what does it provide for routing between multiple AI models? Kilo Gateway is a gateway that supports over 500 models and lets a single server-side API key cover multiple inference calls without managing separate provider credentials for each model. In one application it routed calls to a small decision model called Jev through a TypeSafe System One endpoint and to Gemini through an OpenAI-compatible chat-completions endpoint. _Teams comparing model-gateway options for multi-model apps can follow this kind of setup on daily.dev._ ## Similar posts on daily.dev - [Building for builders](https://daily.dev/posts/building-for-builders-hqwxhn8k6) · Kilo Blog · 0 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#google-gemini](https://daily.dev/tags/google-gemini), [#kilo-code](https://daily.dev/tags/kilo-code) [View this post on daily.dev](https://daily.dev/posts/building-a-speaker-arena-with-kilo-jev-and-enkrypt-nhx30yrfr) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Building a Speaker Arena With Kilo, Jev, and Enkrypt","url":"https://daily.dev/posts/building-a-speaker-arena-with-kilo-jev-and-enkrypt-nhx30yrfr","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/building-a-speaker-arena-with-kilo-jev-and-enkrypt-nhx30yrfr"},"datePublished":"2026-10-27T21:18:45.480Z","dateModified":"2026-10-27T21:39:43.369Z","description":"A developer built the 'Speaker Arena', a browser-based 3D conference world with an AI guide, for the All Day AI event that drew 12,313 registrants and 140...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/5167676e9236094c4b4e6bdd8441ae4f?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/5167676e9236094c4b4e6bdd8441ae4f?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Kilo Blog","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Kilo Blog","logo":"https://media.daily.dev/image/upload/s--x3pmrf8D--/f_auto,q_auto/v1774964071/logos/kilo-ai-blog?_a=BAMAMiWQ0","url":"https://daily.dev/sources/kilo-ai-blog"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/building-a-speaker-arena-with-kilo-jev-and-enkrypt-nhx30yrfr","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,google-gemini,kilo-code","timeRequired":"PT7M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Kilo Blog","item":"https://daily.dev/sources/kilo-ai-blog"},{"@type":"ListItem","position":3,"name":"Building a Speaker Arena With Kilo, Jev, and Enkrypt"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/building-a-speaker-arena-with-kilo-jev-and-enkrypt-nhx30yrfr#faq","mainEntity":[{"@type":"Question","name":"How can I combine a small classification model with a larger generation model to reduce LLM costs?","acceptedAnswer":{"@type":"Answer","text":"Route the question to a small decision model first to classify it as relevant or off-topic, then only call the larger generation model for relevant questions. In this setup, a small model called Jev returned either 'talk_question' or 'off_topic', skipping the Gemini 2.5 Flash answer call entirely for off-topic requests to save on generation cost and latency. Developers weighing model-routing strategies for cost control can track real implementations like this on daily.dev."}},{"@type":"Question","name":"How do I stop an LLM guardrail from flagging public figures' names as PII?","acceptedAnswer":{"@type":"Answer","text":"Disable the overly broad PII detector for cases where public names and organizations are expected content. During live testing of an AI conference guide, broad PII settings flagged benign speaker-related content such as public names, so the guardrail was adjusted by disabling that detector and then retested against normal questions, off-topic requests, and injection attempts. Anyone tuning guardrail false positives can follow real-world fixes like this on daily.dev."}},{"@type":"Question","name":"What is Kilo Gateway and what does it provide for routing between multiple AI models?","acceptedAnswer":{"@type":"Answer","text":"Kilo Gateway is a gateway that supports over 500 models and lets a single server-side API key cover multiple inference calls without managing separate provider credentials for each model. In one application it routed calls to a small decision model called Jev through a TypeSafe System One endpoint and to Gemini through an OpenAI-compatible chat-completions endpoint. Teams comparing model-gateway options for multi-model apps can follow this kind of setup on daily.dev."}}]} ```