Discover the latest news, updates, and articles from the team and community.
Before we shipped a single line of AI Sessions, we had already spent months supervising AI coding agents on our own codebase, through Webmux, the open-source dashboard we built for ourselves. Here is how that daily use shaped the product.
How do I automate Windows servers without Docker? Windmill runs natively on Windows as a service with PowerShell, C#, MSSQL Kerberos auth, and Teams integration.
How do I build an AI Discord bot that accesses Stripe without leaking API keys? Use Windmill sandboxed tools to give the AI capabilities, not credentials.
Introducing workflow-as-code: define complex workflows in TypeScript or Python with checkpointing, parallelism, and fault tolerance built in.
Introducing improved Git sync and workspace forks: sync workspaces to Git, fork for feature development, and deploy via CI/CD.
Introducing AI sandboxes: run Claude Code, Codex, or custom agents in isolated environments with persistent volumes on Windmill.
Introducing Data Tables and Ducklake: managed SQL storage and S3-backed data lakehouse in Windmill.
Introducing full-code apps: build React or Svelte frontends connected to backend runnables in 25+ languages, with a typed API and local dev workflow.
How does Webmux help engineers run parallel AI coding agents across git worktrees from a single web dashboard?