Windmill

Windmill lets developers build internal apps, data pipelines and AI automation around ordinary scripts. Teams can share that code through browser interfaces and run it on infrastructure they control.

• Scripts as apps and APIs: Python, TypeScript, Go, Bash, SQL and Rust scripts can become API endpoints and generated forms, or serve as the backend for custom React and Svelte apps.
• Workflows: Combine tasks in a visual editor or write workflows in Python or TypeScript, with retries, approval steps and parallel branches.
• AI agents: Agents can use scripts and MCP servers as tools, with a choice of external model providers or your own model behind an OpenAI-compatible endpoint.
• Data pipelines: Run DuckDB and Python tasks or bring an existing dbt project, with model assets and lineage tracking.
• Run monitoring: Inspect inputs, results and logs, review workflow timelines, and send failure alerts through Slack, Teams or email.

Development works in the browser or from a local editor. Shared workspaces provide permissions and encrypted credentials, with integrations such as PostgreSQL, Snowflake and Slack.

Windmill can run on your own servers through Docker or Kubernetes; Windmill Cloud is hosted by the company. Self-hosting places the platform and job execution on your infrastructure, while agents using external model providers send requests to those services. The open-source code uses AGPLv3, and commercial licensing and support are also available.