Internal Tools · Claude Code
Three internal AI tools, shipped to production
Designed and built three AI tools that are in daily use across product, sales, and engineering — covering the three core LLM application patterns: summarization, agentic workflows, and retrieval.
All three are shipped and in regular use.
Goal
Stop talking about where AI could help and ship tools that actually do — built hands-on with Claude Code, in production, used by real teams.
What I built
A release-notes generator. It turns a set of engineering tickets into clear, customer-facing release notes. What used to be a manual writing chore is now a fast, consistent step in the release process.
A customer-interview-to-spec workflow. An end-to-end pipeline: it takes a recorded customer call, transcribes it, summarizes it, pulls out themed insights, and produces structured input for a product spec. It compresses the slowest part of discovery — turning conversations into decisions.
A retrieval assistant for the sales team. A RAG tool that surfaces specific technical details buried inside product spec sheets, so the sales team can answer customer questions in seconds instead of digging through PDFs.
Why it matters
Those three tools aren’t a grab bag — they cover the three patterns most LLM applications are built on: summarization, agentic workflows, and retrieval. Most product managers have shipped none of these. Building all three, in production, is what “builder PM” actually means.