In this episode, we dive into how real-world AI systems handle messy receipt data, structured extraction, and eval pipelines. We cover why reliability matters, how invariants can replace hand-labeled datasets, and what it takes to scale an extraction workflow from dozens of receipts to thousands.
Chapters:
Intro
AI Reliability
Receipt Data Basics
Evals Overview
System Architecture
Invariants Approach
Core Loop & Scaling
Prompts & Improvements
Final Thoughts & Wrap-Up
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.