Kiln AI is a workbench for teams to build, evaluate, fine-tune, and optimize models using an open-source Python library and desktop application.
What it is
Kiln AI is a desktop application and open-source Python library designed to help development teams build, evaluate, fine-tune, and optimize systems. It combines local data management with git synchronization, allowing engineers, data scientists, and product managers to collaborate on testing and improving models.
Key features
- Build retrieval-augmented generation pipelines, custom skills, and sub-agents using a visual interface
- Evaluate model quality through LLM-as-a-judge scoring, auto-generated evaluations, and golden datasets
- Automatically optimize prompts and agent designs to match specific evaluation criteria
- Fine-tune smaller models using generated synthetic data and custom datasets
- Chat with an assistant to create experiments, run tasks, and troubleshoot configurations
- Deploy tasks into production using an MIT-licensed Python library
Who it's for
Kiln AI is built for software engineers, data scientists, and product managers who need a collaborative environment to test, score, and refine models without writing complex infrastructure code from scratch.
Pricing
Kiln AI offers a free individual tier that includes the open-source Python library, desktop application, and local datasets with git synchronization. Paid team and custom enterprise tiers are available for organizations requiring advanced server-side features, higher rate limits, SSO, and priority support.
