AI Engineers Tech Talk: August

Recap

In August, SF AI engineers got together for another night of tech talks, pizza, and shop talk. Three speakers walked through systems they'd actually built, and two big questions ran through the whole evening: how do you give AI agents memory that holds up over time, and how do you build a product that survives the next wave of new models? If you missed it, the recordings are below.

Full Event Recording

Here's the whole evening in one place, all three talks plus the Q&A. It's worth watching start to finish if you're building agents that need to know more than what fits in a prompt.

Adam Malone, Senior Director at PromptQL
Building a Company Brain for AI Agents

Adam argued that vector databases alone fall short when company knowledge keeps changing. He showed PromptQL's self-organizing, Wikipedia-style knowledge layer, which agents can pull from and help keep up to date. The big takeaway was that you can give agents durable domain knowledge without fine-tuning or cramming everything into a giant context window.

Ming Wu, Advisor at Parcle
Building Your Company Ontology for Context-Aware Agent Systems

Ming took on a problem anyone using more than one AI tool has run into: every new session starts from scratch. He shared how Parcle is building a shared memory layer that carries your context across ChatGPT, Claude, Hermes, and OpenClaw. If you've ever re-explained your project to a chatbot for the fifth time, this one's for you.

Luke Zhan, Software Engineer at Altara
Defeating "Infinite Churn": A Plug-and-Play Architecture for AI Startups

Luke's talk was less about agents and more about staying sane. New models and providers show up constantly, and Altara didn't want to rebuild their platform every time. He broke down how they separated product logic from infrastructure using abstractions and adapters, so trying a new AI tool or cloud provider is a small job instead of a rewrite.

See You at the Next One

Want to catch the next talk in person? Head to fonzi.ai to grab a spot, get intros to AI startups, or find your next engineering hire.