Louisa
Creator
Open-Source Developer Tools · 2026
Built Louisa, an open-source AI workflow that automatically generates polished release notes for GitHub and GitLab repositories — eliminating manual changelog writing by using Claude to enrich PRs at merge time and synthesize user-facing release notes on tag push.
Technology Partners
Challenge
Engineering teams waste hours every release cycle on changelog maintenance — pulling from vague PR titles, inconsistent commit messages, and half-written descriptions to produce something readable for users. The gap between raw commits and user-facing release notes was a consistent pain point across Arthur's own repositories.
Solution
Built Louisa as a Node.js workflow running on Vercel and GitHub Actions. Claude Haiku enriches each PR at merge time, rewriting descriptions with structured context — problem, solution, and user impact. On tag push, Claude Opus synthesizes those pre-computed summaries into polished, grouped release notes organized by product area. Instrumented end-to-end with OpenTelemetry and Arthur Engine for full observability into AI execution.
Impact
- —Eliminated changelog maintenance entirely — push a tag, production-quality release notes appear automatically
- —Designed a two-model AI strategy that balanced speed and quality at every stage of the pipeline
- —Built observability-first: every AI call instrumented with OpenTelemetry and Arthur Engine from day one
- —Shipped GitHub and GitLab support out of the box — usable by any engineering team, immediately
- —Extended to auto-generated blog posts and changelog publishing with zero additional engineering effort
Results
- —Published as open-source under the arthur-ai org on GitHub
- —Became a reference implementation for AI workflow design vs. autonomous agent design
- —Featured in Arthur AI blog on building production-ready AI systems
- —Demonstrated observability-first approach to shipping AI in the real world
Design
