The open-source Kiln AI project team builds the Kiln desktop app and library.

A local-first workbench combining synthetic data generation, dataset versioning, evals, prompt optimization, RAG, and zero-code fine-tuning.

Distributed as a desktop app for macOS, Windows, and Linux, with a companion Python library and REST server.

Actively developed with ongoing releases through 2026.

Lets non-coders such as PMs, subject-experts, and QA and engineers collaborate on the same dataset from generation through deployment.

Fine-tunes models like Llama and GPT-4o via serverless deployment and integrates with Git for dataset version control.

Core library and server are open-source (MIT license), the desktop app is source-available and free under a fair-code model, data stored locally by default.