@DavidKPiano That’s exactly what I’ve built, a bit different approach, using DAG.
Something like terraform for agents, separating plan and execution from each other:
https://t.co/1gt0xV9vkl
Current AI agents:
• Open-ended loops
• Wasted API keys
The Rigorix way:
• Task ➡️ Deterministic DAG
• Strict budget & policy bounds
The LLM generates code; the graph controls the flow. Built in Rust 🦀
Repo:
https://t.co/1gt0xV9vkl