In this episode we tackle a hard problem live: designing and implementing parts of an agentic system. Topics include agent design and control flow, compiler improvements, full-turning completeness, arbitrary object instantiation, garbage collector and heap allocator work, F5 bridges, and the rebuilt code layer. They also explain a refined workflow that reduces reliance on fragile prompt hacks—replacing “magic words” with clearer control-flow patterns informed by their 12-factor approach to coding agents.
Chapters:
Intro
Architecture Overview
Tooling & CI/CD
AI Research Questions
VM & Bytecode
Prompt Engineering
Design & Phases
Code Review Workflow
Wrap-up / Outro
Check out our github: https://www.github.com/boundaryml/baml
AI That Works repo: https://github.com/ai-that-works/ai-that-works
Socials:
X: https://x.com/boundaryml
Discord: https://discord.com/invite/yzaTpQ3tdT
LinkedIn: https://www.linkedin.com/company/boundaryml/
Transcript
I've watched this one closely. Ask me anything about it and I'll point you to the exact moment.