Leeroo open-sources Kapso for enterprise data experiments
Original title: Show HN: Kapso – Iterative learning from enterprise data
Authors claim first place on RelBench and a top open-source rank on OpenAI’s MLE-bench at submission.
Leeroo open-sourced Kapso, an engine that takes an objective and success criteria, runs experiments, deploys a solution, and updates a git-backed bank of lessons. It is model- and coding-agent-agnostic and can adapt a solution to stacks such as Docker, Modal, BentoCloud and LangGraph. The authors say it led RelBench, ranked first among open-source systems on MLE-bench at submission, and placed in the top three AI systems at the International Olympiad in AI 2026.
Unconfirmed: Benchmark and competition rankings are claims by the Kapso authors, not independently confirmed in the post.
- Lessons carry a credence score, conditions and evidence, updated by learn().
- A verifier the experiment cannot access checks results.
- The public Leeroopedia bank is described as about 100k pages and about 6M requests a month.
- A commenter notes another popular service already uses the name Kapso.
From the original article
6 min read at news.ycombinator.com[](https://news. ycombinator. com/)Hacker Newsnew past comments ask show jobs submitlogin [](https://news.
By alireza_msh · news.ycombinator.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.