To confirm your spot, please complete your registration on our official Luma event page
You’ll receive a confirmation email once your seat is approved.
Building Voice & Video AI agents that work in a demo is one thing. Making real-time conversations feel natural, coordinating multiple agents, and keeping systems reliable in production is a much bigger engineering challenge. These systems need to respond with minimal latency, handle interruptions, maintain conversational context, and integrate with real business workflows. In this meetup, we'll explore both sides of the challenge: the architecture behind a real-time multi-agent video experience and the practical lessons from deploying Voice AI agents in production.
On the Agenda
18:00 to 18:30
Welcome Drinks & Networking
18:30 to 19:15
Real-Time Multi-Agent Voice AI Architecture
Yanai Franchi, Agentic Platform Architect, Tikal
What does it take to bring a moderator and three AI experts, covering frontend, backend, and DevOps, into one live conversation and make the interaction feel natural?
In this session, Yanai will take us inside the architecture of the Consulting Room, a multi-agent video experience where AI consultants listen, decide who should respond, speak in under a second, and collaborate on a shared architecture canvas.
He'll explain the engineering decisions behind the system, including routing conversations before users finish speaking, racing LLMs against each other to reduce latency, filling silence without sounding robotic, and implementing echo-aware voice activity detection in the browser.
We'll also share a surprising lesson about measuring AI quality along the way.
19:15 to 20:00
From Voice AI Demos to Real Business Impact
Yakov Sinwani, Senior Agentic AI Engineer, Tikal
Getting a Voice AI demo to work is relatively straightforward. Making it reliable enough to handle real conversations and business processes in production introduces a different set of challenges.
In this session, Yakov will share practical experience building and deploying Voice AI agents in real-world environments.
Through live demos and real examples, he'll explore how to build natural conversations, integrate agents with existing business systems, introduce human supervision when needed, and ensure reliable business workflows.
The focus will be on the engineering and operational challenges encountered in production, the decisions made along the way, and the lessons learned from real deployments.
The talks will be held in Hebrew.
Who Should Attend?
This meetup is intended for AI Engineers, Software Architects, Tech Leads, Engineering Managers, and developers building or exploring Voice AI, real-time conversational agents, and multi-agent applications.
To confirm your spot, please complete your registration on our official Luma event page
You’ll receive a confirmation email once your seat is approved.