Kiln

Kiln is an AI workbench for building, evaluating and improving AI systems. It combines a desktop app for macOS, Windows and Linux with an open source Python library, so a task designed in the app can be called from production code. The feature set covers retrieval augmented generation with document extraction, chunking, embedding and indexing, plus skills, tools, MCP connections and sub agents. For quality work it offers evals with LLM judges, golden datasets and human ratings, an AI eval builder, synthetic data generation, fine tuning to distill into smaller models, and automatic optimization of prompts against evals. An AI assistant can run experiments and optimize a system through conversation. Datasets live on the user's machine and sync through a git repository the team controls, and the vendor lists support for cloud and local models. Kiln is aimed at engineers, data scientists and non coding contributors such as product managers, QA staff and subject matter experts, who can add feedback, ratings and evals without writing code. The Individual plan is free and includes the Python library (MIT licensed), the source available desktop app, local datasets with git sync, Kiln Pro with standard models at rate limits, and community support. The Team plan is available by request and adds enhanced Kiln Pro models, higher limits, automatic agent optimization and email support. Enterprise pricing is custom and adds SSO and SAML, annual contracts, an SLA and a dedicated solutions engineer. Kiln Pro features run on the vendor's servers and are the part that is charged for.