Cisco AI FAPO: Automated LLM Prompt Optimization with Claude

Cisco AI has introduced FAPO, a new system designed to optimize large language model pipelines. FAPO stands for Fully Automated Prompt Optimization and uses Claude Code to achieve target accuracy. Here's how it works: you provide a dataset and an initial prompt. FAPO then evaluates, identifies failures, suggests new variations, validates them, and repeats the process. This entire loop is managed by Claude Code agents. The project is open source under an Apache 2.0 license. In Cisco's evaluation, FAPO outperformed GEPA, another prompt optimizer, in 15 out of 18 model-benchmark comparisons. When FAPO made pipeline changes, it showed a mean gain of over 33% compared to GEPA. The bottom line is that FAPO aims to make developing reliable LLM applications easier by automating a critical and often difficult part of the process.