Stop Fighting the LLM: Mental Models for Fast, Frustration-Free App Building

I've built a lot of software with LLMs over the past year. Some of it went great. Some of it turned into five prompts of debugging hallucinated code while I questioned every life choice that brought me to that moment. For most people, the biggest bottleneck is the expectations they bring with them. Specifically, it's the mental models we bring to the work.

In this session I'll share the frameworks I've landed on for actually shipping apps with LLMs, not just generating code and hoping for the best. We'll talk about how to treat an LLM like a tireless intern instead of a magic wand, how to recognize and escape context collapse before your project goes sideways, and the specific tooling and workflows that bridge the gap between a raw prompt and something you can actually deploy.

If "vibecoding" hasn't been working for you, or if you're an experienced engineer who keeps getting stuck in loops with AI, this talk is for you. You'll leave with a repeatable approach to designing, prompting, and shipping real software without the frustration.

• Beginner
• Development
• Startups
• Emerging Tech
• AI/ML
• Product