Lenny Rachitsky · 3 min read · Gen AI Tools
🎙️ This week on How I AI: I gave Claude Code our entire codebase. Our customers noticed.
Read the original on lennysnewsletter.com →
Al Chen shows how Claude Code can ingest Galileo's entire codebase across 15 repositories and, when combined with Confluence and Slack, deliver accurate, real-time answers for customers without repeatedly pinging engineering. The approach treats code as the truth, turns customer questions into scalable knowledge, and aims to reduce interruptions while preserving human validation.
Key Takeaways
- Write a daily 16-line script generated by Claude Code to pull the latest main branch from all 15 repositories, ensuring AI answers stay current.
- Maintain a 'customer quirks' page in Confluence with each enterprise customer's deployment requirements to tailor AI responses.
- Use a multi-source context flow where Claude Code first checks Confluence for deployment docs, then queries code repositories for implementation details.
- Turn Slack support threads into knowledge base articles automatically using Pylon to capture real customer questions as current, in-depth content.
- Reduce engineering interruptions by letting AI answer common questions, with human proofreading and validation for complex cases.
Excerpt
I gave Claude Code our entire codebase. Our customers noticed. | Al Chen (Galileo) Listen now on YouTube • Spotify • Apple Podcasts Brought to you by: Orkes —The enterprise platform for reliable applications and agentic workflows Tines —Start building intelligent workflows today Al Chen is a field engineer at Galileo who works directly with enterprise customers asking highly technical questions. In this episode, Claire talks with Al about how he uses Claude Code to query Galileo’s entire codebase across 15 repositories, combine it with Confluence and Slack, and deliver accurate, real-time answers without relying on documentation or constantly pinging…