Cisco Foundation AI Intros FAFO: Sparse User Feedback Powers a Recursive Agent Self-Improvement Loop
aminkarbasi · x · 2026-10-10
Cisco Foundation AI introduced FAFO (Fully Automated Flow Optimization), a data pipeline that turns production agent traces and sparse user feedback into automated evaluation and optimization.
- The problem: teams collect thousands of agent traces but user feedback covers only a small subset; LLM-generated rubrics alone may treat agent mistakes as acceptable
- The approach: extract reusable guidelines from the few feedback-labeled traces, then apply them to build rubrics for all traces — amplifying limited feedback into large-scale eval sets
- FAFO extends their earlier FAPO prompt-optimization work, closing the loop from eval case creation to harness optimization, enabling recursive self-improvement grounded in real-world data
More from coding & agent
- Watch Claude build a GTA clone in hours: video shows how the AI iterates and debugs on its own — Daniel_Farinax · 2026-10-10
- dotConferences talk: how speculative decoding speeds up llama.cpp — ngxson · 2026-10-10
- Developer ships first App Store app Save4Two built with Cursor and Grok — chrisfleck · 2026-10-10
- Dev clones Photoshop and the whole Adobe suite in Rust using Claude — iamfakhrealam · 2026-10-10
- When AI Writes the Code, What Replaces Design Patterns? — JnBrymn · 2026-10-10
- Google Cloud launches App Optimize remote MCP server to query cloud costs from any AI assistant — rseroter · 2026-10-10