✨ TLDR
Helped Ragas (open source evals framework) revamp their docs and built automation so docs stay in sync with code changes.
They now have AI-ready docs and a workflow that auto-updates documentation when code changes.
🎯 The Problem
Ragas is one of the most popular open source evals frameworks — 10k+ GitHub stars, 11M+ downloads.
They were launching a new SDK version and needed:
- A complete docs overhaul for the new paradigm
- Systems to keep docs in sync long-term
💡 The Approach
Jump in, deliver value fast, then leave systems behind.
The goal: help them ship better docs, then set up automation so the work continues without me.
🛠️ What I Did
Docs Revamp
- Restructured everything around an experiments-first paradigm
- Fixed broken tutorials and guides
- Wrote new how-to guides: aligning LLM-as-Judge, evaluating RAG apps
- Simplified existing guides that were too complex
AI-Ready Automation
This is where it gets interesting:
- llms.txt generation — LLM-friendly documentation format so AI tools can easily consume their docs
- Copy-to-LLM button — One-click to copy docs content formatted for AI assistants
- Auto-update workflow — When code changes, it checks if docs need updating and proposes changes automatically
The auto-update workflow is the real unlock. Docs drift is inevitable — now they have a system that catches it.
🧠 What I Learned
Systems > one-time work. The docs revamp was valuable, but the auto-update workflow is what compounds. Now every code change triggers a docs check. That’s leverage.
🔗 Related
- AI Consulting — More about my consulting work