Windmill is a Y Combinator-backed open-source platform that combines a workflow engine, an internal app builder, and a script execution runtime into a single self-hostable product. The backend is written in Rust with stateless API servers and workers pulling jobs from a PostgreSQL queue, which gives it low overhead — a typical lightweight job completes in around 100ms total with only ~50ms of queue latency. Independent benchmarks show it outperforming Airflow, Prefect, and Temporal on both lightweight parallel tasks and long-running sequential workflows. Scripts can be written in Python, TypeScript, Go, PHP, Bash, C#, SQL, Rust, or run inside arbitrary Docker containers.
The platform works at three levels: individual scripts get automatic parameter parsing and UI generation so any function becomes a shareable tool; flows compose scripts into multi-step workflows with branching, retries, error handling, and approval steps via a visual editor or YAML; and the app builder creates data-centric dashboards and internal tools using drag-and-drop components or full React/Svelte code. Local development is supported through a CLI and VS Code extension that sync with Git repos, and workspace forks enable parallel teamwork with built-in diffs and audit trails before merging to production. Sandboxing uses nsjail for filesystem and PID namespace isolation to prevent jobs from accessing worker memory.
Self-hosting requires just three files — docker-compose.yml, Caddyfile, and .env — backed by PostgreSQL (compatible with Aurora, Cloud SQL, Neon, and Azure PostgreSQL). Scaling goes from a single-node VPS to 1,000-node Kubernetes clusters with auto-scaling worker groups. The Community Edition is AGPLv3 and free with no seat limits. Enterprise Edition adds SSO/SAML, Git sync, audit logs, distributed S3-backed dependency caching, and dedicated support, with pricing based on compute units measuring worker-GB-months. Windmill is also available on the AWS Marketplace. Current production users include Pave running 100+ scripts and 15+ crons across multiple teams, and Investing.com orchestrating AI content processing and ETL pipelines.