My DIY #C08QMJD9J9G project: https://livekit-users.slack.com/archives/C08QMJD9J9G/p1746198983997959

I am excited to see this course come out about "Building AI Voice Agents for Production". It has a few familiar faces 😉 https://x.com/realavatarai/status/1920157517860254066 https://www.deeplearning.ai/short-courses/building-ai-voice-agents-for-production

Loved the video from @flaky-sundown-18688 on Tavus, so I made one too! Although, his was way cooler with the quiz and flashcards Cheers!

How to easily build & deploy Livekit Voice AI and 🚀Telephony Assistants without Render.com or Cloud Init or Downloading Docker Desktop 🤮

Let's redefine customer support—African style. Imagine if your AI support agent didn't just understand what your customers say—but how they say it. Welcome to Spitch Agent — the future of intelligent, local-first customer service. Built for voice. Built for Africa. Built for you. From English to Hausa, Igbo to Yoruba, Spitch Agent gets it. No more misinterpretations. Just smooth, smart, culturally fluent conversations that make your customers feel truly heard. Here's what Spitch Agent unlocks: ✨ Natural voice + text support that understands African accents and expressions 🧠 Smart conversations 24/7, powered by real-time AI 📈 Continuous learning—gets better every time you use it Try the live demo: https://chat.spitch.app/

so i built an agent to lowball watch dealers

Hey #C05L395US9L one of our users built an Personal AI Secretary using Livekit for the Voice AI functionality (it is currently Live on the📲 App Store) https://apps.apple.com/us/app/iassist-ai/id6745621868 Best of All its absolutely FREE (my favorite price 😅) We have reached the 🥇first page of Product Hunt on our first day. Thank you @magnificent-art-43333 and @dry-elephant-14928

I made this video that shows how to get your LiveKit ai voice agent to fill your website's forms instead of manually typing, which you can also do and the agent will understand either way.

@adorable-dentist-65637 That's really cool. This would really be useful in real world.

Hey #C05L395US9L, if you're looking to launch a video conferencing solution with LiveKit and would like a boost to your production timetables, check out Hylist! We offer a licensable, production-ready, general purpose video conferencing solution built for iOS, Android, and Web. Focus on your unique market use-case without worrying about the fundamentals of video-conferencing. For more information, visit https://docs.hylist.app or try the software out for yourself at https://hylist.app.

Build a new MCP demo! This one shops with you: https://x.com/shayneparlo/status/1925973507940262021

We are using livekit for some time on the https://jobmojito.com platform (AI job interviews), you can try it using our sample interview. https://try.interviewmojito.com/position/49cb2c01-f34f-4166-8340-4328314be0d7 Let me know, what you think 🚀

Just published a blog post about how we've built proactive AI agents (using livekit!) https://open.substack.com/pub/bryanhoulton1/p/building-proactive-ai-agents?r=1gvtm6&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

I have created an AgentSession(as seen below). I want to extend teh llm to implement my CustomAzureOpenAILLM class(see image below). Is this allowed. If yes is there any documentation or url you could provide where I can get this information?

image.png

Hello dears, I have a question, Im working on testing phone call appointment scheduling agent uses STT - LLM - TTS to check latency, call health, ..etc built using Live-Kit plugins, what is the best way to automate a test for it, creating static pipeline with .wav or mp3 mock files to answer the agent questions, creating another agent using basic OpenAI voice agent, or built the same agent but oppositely to start talking with the original agent or if theres any recommendations to do that, please help!

I want to validate my idea. This is my product: https://agentfied.com Can you share some feedback on this project? This is an MVP

Hey devs! We just launched Avatario – real-time AI avatars powered by LiveKit. Your livekit agent can now talk, react, and look human (minus the caffeine addiction). Low-latency, API-first, and dev-approved. Perfect for support bots, AI tutors, virtual astrologers, virtual assistance — you name it. 60s live demo (free to try): avatario.ai 100 minutes of free usage on signup. You can integrate it seamlessly in under 10 minutes. Would love your thoughts or roast 🙃

Hey guys I launched PACE. An AI Copilot for freight brokerage, 1. Comprehensive Multi-Tenant Architecture • Organization-based data isolation ensuring complete security between different freight companies • Role-based access control with 5 distinct user roles (ADMIN, BROKER, CARRIER, CONSIGNOR, CONSIGNEE) • Selective PIN authentication - 80% reduction in PIN prompts while maintaining security for critical operations 2. Advanced AI & Voice Integration • Voice-first automation with for hands-free operations • Natural Language Processing for extracting tracking numbers from conversations • AI-powered route optimization and predictive analytics • Voice status updates without authentication interruptions 3. Real-Time Tracking & Monitoring • Enhanced tracking system with synthesized insights about shipments • Multiple identifier support (tracking numbers, BOL, container numbers, reference numbers) • Real-time status updates with location tracking • Automated delay detection and proactive alerting 4. Automated Business Operations • Intelligent quote generation with PDF automation • Automated follow-up campaigns (quote follow-ups, delivery confirmations, feedback requests) • Performance analytics and reporting with KPI calculations • Document processing (BOL, invoices, shipping documents) • Smart notification system (email, SMS, voice alerts) 5. Complete Freight Brokerage Workflow • Quote-to-delivery lifecycle management • Carrier assignment and management • Rate calculation engine with service type multipliers • Document management system with cloud storage • Proof of delivery handling 6. Advanced Analytics & Intelligence • Real-time dashboard with metrics and KPIs • On-time delivery rate calculations • Commodity analysis and trends • Carrier performance monitoring • Revenue and performance metrics Freight is a huge market from statistics 2024- 2025 Global Logistics Market Size: • 2024: USD $9.98 trillion • 2025: USD $11.23 trillion • 2034 (Projected): USD $23.14 trillion • Growth Rate: 8.36% CAGR (2025-2034) Please check it out https://gopace.app

Hey Livekit community 🙂 A couple weeks back, I was brainstorming ideas for a product when an idea that I liked crossed my mind. What if I built a voice agent that guides you in writing your resume. So I went ahead and built it. Took me a month. But I believe I am starting to see good results. I am giving away free sessions with the agent to people in this community. And I’d love to get your feedback. You can find it at Vocalcv.com If you have any questions about how i built it, feel free to drop a comment — I’ll be happy to share! take a look ⬇️

As part of scheduled cleanups I want to remove inactive/closed livekit sessions. How can i do?

Hey folks, I’m currently trying to decide between open source repos like Pipekat and LiveKit, and I’m a bit stuck. I’d really appreciate some insight into the pros and cons of each. Here’s what matters most to me: 1. Full control over the entire STT → LLM → TTS pipeline. 2. Low latency is absolutely key. 3. How hackable and easy to navigate the codebase is. 4. How easily I can support multiple agents in real time. 5. How cloud-agnostic the solution is — ideally, I want smooth deployment on AWS or any other cloud without being tied to their native infra. If you were in my shoes, which would you pick and why?

Demo from Shayne for cartesia_ai STT https://x.com/shayneparlo/status/1932840478036066341

Hey people! We are offering a free round of user interviews with launchskylight.com, voice agents for automated user interviews. If you have a product with many users and no PMF yet, asking 100x the users about their experience and needs is obviously insanely useful. Just DM me here and well sort you out in 24 hours. Have a nice day!!

I created an mcp server that adds bidirectional voice to anything that supports MCP. It defaults to microphone and speakers if livekit server is not available. mbailey/voice-mcp : Voice Mode for Claude Code Add two way voice to Claude Code (or any tool that supports MCP). - Includes support for locally hosted ASR and TTS (e.g. Kokoro-FastAPI) - free and open source http://voice-mcp.com/

Hey everyone, I have attempting to create a no-code way of creating voice agents with livekit. Meet Vaani – a configuration-driven voice-AI framework built on top of the LiveKit Agents • Define any agent with a single JSON file (no code) • Support multiple providers: elevenlabs, deepgram, sarvam, cartesia, openai, etc • Worker pulls that config at runtime, so updates are instant • Built-in webhooks for transcripts & metrics • Optional RAG (Pinecone + LlamaIndex) and long-term memory with Mem0 Repo: https://github.com/namish800/vaani Do check it out and provide feedback. Thanks

Introducing MIA – Your Gateway to No-Code AI Automation Unlock the power of intelligent automation with zero coding required. ✨ Powered by the LiveKit framework, MIA delivers seamless, real-time interaction capabilities to supercharge your AI experiences. Why MIA? • All-in-One No-Code Platform: Build ChatBots, VoiceBots, and deploy *AI Agents*—no tech skills required. • AI Agent Marketplace: Access a growing library of ready-to-use agents for tasks like MOM generation, content creation, RAG, diarization, social posting, and more. • VoiceBot Telephony Integration: Connect effortlessly with systems like Twilio, Asterisk, and Plivo to manage real-world calls—both inbound and outbound. • Workflow Orchestration (Coming Soon!): Coordinate multi-agent tasks with powerful orchestration tools—all within MIA. • Flexible Subscription Plans: Tailored options with rollover credits and enterprise-ready packages with GDPR & HIPAA compliance. • Scalable by Design: Architected with LiveKit + Kubernetes + Docker for high reliability and dynamic load management. • B2B2C Friendly: Onboard your clients, assign projects, and help them build their own smart assistants. Link: https://supermia.ai/ Do check it out and provide feedback.

If you're looking to build your first AI voice agent using LiveKit, I just finished making a video that walks through the process. I tried to make it a clear and helpful guide. You can check it out here:

/

Hi everyone, I’ve benchmarked the various encoders available for screen sharing with LiveKit and documented the process and findings here. I hope this comparison proves useful!