/ About this event
The demo always works. Production breaks. And usually it's not the model that fails — it's the information the agent is acting on. An agent is only as good as what it knows, so here's the question worth an evening: how does an AI agent get the latest, most reliable, verifiable external information to act on, at agentic speed?
FounderCoHo is teaming up with Google Cloud, and DeepVista for an evening of lightning talks and honest, in-the-trenches conversation about what it actually takes to run an agent in production.
So that's what we're spending the night on. Founders and engineers actually shipping this stuff will share field notes on:
Getting agents the right information at the right moment — and grounding their decisions in real, current sources instead of guesswork Keeping state from drifting across long, multi-step runs Building multi-agent systems that hold up instead of deadlocking Token economics — pushing completion rates up and API bills down
Speakers Annie Wang (Qingyue Wang) — Developer Relations Engineer @ Google, focused on AI agents: multi-agent patterns, agent memory, self-evolving agents, and graph workflows with ADK. Runs live agent-building sessions with big audiences.
Jay W. — Co-Founder of OpusClip (the #1 AI video-clipping tool). Shipped "AI Producer," an agent that edits video end-to-end — a real production agent with a large user base.
Samer Masterson — Staff Engineer @ Harvey (the AI platform for legal work), previously at Google. Builds production agents for high-stakes legal workflows — the kind of domain where "mostly right" isn't good enough, so reliability and long-horizon behavior are the whole game.
Philip Lan — Member of Tech Stuff @ Harvey, previously at GitHub and Microsoft. Brings deep developer-tools and systems experience to shipping AI agents that actually hold up inside real legal workflows.
Moderator Jing Conan Wang — FounderCoHo co-founder & podcast host, former Google Brain researcher, founder of DeepVista AI.
Agenda
5:00 PM–5:30 PM Check-in, drinks, and casual networking 5:30 PM–5:50 PM Inside a 30-Turn Agent Run: Where Things Break—and Where Tokens Disappear 5:50 PM–6:30 PM Founder Panel: What It Takes to Scale Stateful, Multi-Agent Systems 6:30 PM–8:00PM Open discussion, founder Q&A, and networking
Who Should Join
This gathering is designed for: • Founders and technical leaders building autonomous, vertical, or developer-focused AI agents • Engineers working on agent reliability, multi-step orchestration, retrieval, and context management • Builders who want to improve response quality while keeping inference costs sustainable • Anyone curious about how agent systems move from impressive demos to dependable products
Whether your agent is already in production or still taking shape on a whiteboard, you’ll leave with practical ideas—and hopefully a few new people to exchange notes with along the way.
Space is limited so we can keep the discussion thoughtful, interactive, and genuinely useful. Priority will be given to active founders, engineers, and technical operators.
Meet Your Host
FounderCoHo is a community and media platform where founders share hard-earned lessons and build meaningful connections wih others on the same journey. With 220,000+ subscribers across YouTube and Substack, FounderCoHo has featured conversations with the co-founder of Lyft, leading CRISPR researchers, an AlphaChip co-author, successful founders, and senior technology leaders.
We also host monthly in-person gatherings where founders and builders can exchange practical insights, meet new collaborators, and enjoy conversations that go beyond the usual networking small talk. Explore more: https://foundercoho.com/
This event is a part of #SFTechWeek—a week of events hosted by VCs and startups to bring together the tech ecosystem. Learn more at www.tech-week.com
