WHAT WE GOT WRONG
We optimized the model. We should have optimized the context.
Better prompts
Model settings
Output compression
A retrieval layer between codebase and agent
AI text/layout recreation from video frame; verify against source image.