Windmill

Windmill takes a deliberately different angle from the connector-first automation tools: it starts with the script, not the integration. You write a Python, TypeScript, Go, Bash, or SQL function; Windmill wraps it in a production runtime that handles scheduling, secrets injection, error handling, and execution history. Chain those scripts together in the flow editor and you get a DAG-based workflow without having to build the orchestration layer yourself. The backend is written in Rust, which gives the job runner a performance ceiling noticeably higher than comparable Python-based platforms.

The App builder is the feature that tends to surprise engineers most. Any script or flow can be surfaced as a drag-and-drop internal UI — data tables, form inputs, charts — without writing a single line of frontend code. That makes Windmill a credible replacement for lightweight Retool-style internal tools as well as for job orchestration, which is an unusually wide surface area for a single open-source project. The AGPL-3.0 license means the full source is available and free to self-host, with a commercial license for organisations that need proprietary deployment.

The tool is clearly aimed at engineering teams rather than operations or marketing. The mental model is "scripts are the atomic unit," which is powerful if your team writes code and limiting if they do not. Pre-built SaaS connectors are fewer than Zapier's or n8n's, so connecting to common tools like Salesforce or HubSpot often means writing the API calls yourself. For data engineering, internal tooling, and developer-owned workflow automation, that trade-off typically favours Windmill. For general business process automation driven by non-technical staff, it does not.