Windmill replaces Retool and Airflow for developer teams

Windmill replaces Retool and Airflow

Windmill replaces Retool and Airflow by combining script orchestration and internal app building into one platform. Teams moving from n8n often seek a truly open-source alternative to n8n’s fair-code model, which restricts reselling or offering the tool as a managed service. Windmill uses a code-first philosophy where developers write scripts in Python, TypeScript, Go, or Bash instead of dragging nodes on a canvas. These scripts become building blocks that users chain into flows or expose as apps through autogenerated UIs. If you already manage complex infrastructure, you will appreciate how Windmill treats automation as a software engineering discipline. Panther Labs used this approach to separate business logic from platform orchestration, allowing 60% of their mission-critical cybersecurity workflows to run on Windmill. The platform provides triggers via webhooks, schedules, HTTP routes, Kafka, WebSockets, and emails. Users can define different script kinds, including Action, Trigger, Approval, Error handler, or Preprocessor. The platform turns scripts into APIs, background jobs, and UIs.

High performance script orchestration

The Rust backend enables high-performance execution, achieving speeds up to 13 times faster than Airflow in specific benchmarks. Windmill manages dependencies automatically by parsing imports and spawning jobs to associate crates with a lockfile. For developers needing granular control, the platform supports runtimes like Bun, Deno, and Python. Python users benefit from dependency management through uv. You can use the VS Code extension to sync scripts from your local environment to your instance. Engineering teams often migrate to Windmill because the platform allows them to define tasks in multiple languages directly in the UI, which simplifies deployment and provides a level of control that visual node-based tools cannot match. Windmill uses nsjail for filesystem and resource isolation and PID namespace isolation to prevent jobs from accessing worker process memory. Every script version in Windmill is uniquely identified by a hash, and referring to a script by path refers to the latest deployed hash at that path. The interface generates a UI from the JSON schema of the script’s signature. The backend consists of stateless API servers and workers pulling jobs from a Postgres queue.

Deployment and pricing

Self-hosting Windmill via Docker Compose or Kubernetes requires managing your own infrastructure. Some users find the pricing model for self-hosted deployments problematic because it charges based on the number of workers or memory limits on infrastructure they already own. The app builder lacks the sophistication of Retool when building complex internal tools with real-time updates or advanced filtering. While Windmill excels at DevOps runbooks and backend automation, can it truly replace a dedicated data engineering platform for heavy ML pipelines? Windmill supports several languages including PHP, C#, Java, and Ansible. The platform also handles secrets via a built-in variable store. The Enterprise plan starts at $120 per month and adds SAML, audit logs, and autoscaling. For those avoiding infrastructure management, the Cloud Team tier costs $10 per month for up to 10 seats, while developers cost $20 per month and operators cost $10 per month. Nonprofits and SMBs access the platform from $48 per month with a 16% discount when billed annually.