Stop burning cash on API tokens.

Tired of paying good money for API tokens just so a cloud model can confidently tell you about functions that don't exist in your codebase? Same. Let's stop doing that.

The GPU market is a hellscape and RAM costs more than a kidney. This talk we will break down how to survive the current hardware hellscape by running continuous, local-first AI coding agents on whatever hardware you already own. It can run always-on, local-first AI coding agents, and I'll show you how.

Small local models are like enthusiastic interns. They're fast, cheap, eager to help, and they'll absolutely set the kitchen on fire if you hand them anything complicated. We'll cover the execution guardrails that keep them from wandering off a cliff, and the routing strategies that send each job to the right brain. That's how you get a small model to reliably build big, production-grade projects without draining your bank account.

Come for the cost savings. Stay for the networking. Leave with a slightly concerning urge to buy more fans.