

How to Build Production-Ready Voice AI Agents with Coval
Learn how to build Production Ready Voice Agents!
A four-hour deep dive to learn how to build, deploy, evaluate, and optimize production-grade voice AI agents.
As part of the upcoming AI Agent Builder Summit in — we’re hosting a hands-on, expert-led workshop for a limited group of practitioners.
This session is led by Brooke Hopkins, founder of Coval- one of the leading voices in voice-first agent development and Ben Cherry, LiveKit's product and engineering leader.
Who This Is For -
Machine Learning Engineers or Applied AI Engineers with at least 2-3 years of experience
Product-focused engineers interested in deploying AI to production
Voice UX designers curious about the technical build + eval process
What You'll Learn -
Architecting and building reliable voice agents with LLMs, TTS and STT components.
Integrating voice input/output into agent workflows
Optimize for naturalness, latency, and robustness
Deployment and performance tuning in real-world environments
Optimize for naturalness, latency, and robustness
Pre-Workshop Requirements -
Familiarity with Python with Python 3.9+ installed
Basic understanding of LLMs and APIs
Some experience building AI applications, chatbots, or voice systems is helpful but not required
Access to an OpenAI API key or equivalent LLM provider
Clone the provided starter repo (we’ll share this beforehand)
A personal laptop
Workshop Structure + Activities -
The workshop is a hands-on deep dive into building and deploying voice agents—starting from a basic STT → LLM → TTS pipeline to production-ready systems with robust error handling, memory, evaluation, and deployment strategies.
You’ll also get to troubleshoot live issues, run A/B tests across components, and handle edge cases like noise and interruptions. Check out the full workshop structure and activities here.
FOOD WILL BE PROVIDED AS PART OF THE TICKET. Kindly email us at minushi@mlops.community to inform us of any allergies or dietary restrictions.
What You’ll Walk Away With -
A working voice agent demo
Evaluation/observability templates for voice agents
Real-world design tips from people deploying this in production
Confidence to extend or build your own voice-first AI systems
Meet Your Instructor -
Brooke Hopkins founded Coval in 2024 to help teams test and improve AI voice and chat agents at scale. She’s all about making sure these systems actually work in the real world.
Check out Coval here.
Coval is the leading simulation and evaluation platform for AI voice and chat agents. Test, monitor, and optimize your AI agents at scale.
This is a paid workshop with limited capacity — registration is $200 for the next 3 weeks so snatch that early bird discount at half off!
If you’re serious about building voice agents and want hands-on exposure to best practices and deployment techniques, this session is for you.