Apps are getting more complex, MVP scope is growing, and developers are being asked to build more than ever faster than ever. New information systems are also expected to be compatible with agentic features rather than creating friction. Most importantly, developers are required to be accountable for their work despite producing more code than ever.

How do you build reasonable systems with AI when the default mode of working with Agentic Coding leads to architectural drift and slop? Is there a way to bottom out AI decision making at a structural level? Can Markdown files ever be enough?

In this talk, I show you how to use ideas from software engineering, a small set of foundational rules, composable building blocks, and an immutable single source of truth to constrain AI agents and produce reliable behavior, based on our experience in production.

You'll come away from the talk understanding how to create successful software in the post-AI world.

Cameron Barre

Cameron Barre is a long time Clojure professional currently solving problems in the realm of AI assisted coding and building systems that integrate AI features as CTO of ObneyAI