Tau
tau is an open-source git-native PaaS (Platform as a Service) that can be run locally and on self-hosted infrastructure. The project is available at github.com/taubyte/tau. It combines serverless function execution, object storage, a content delivery layer, DNS management, and a p2p networking substrate into a single self-hostable stack. Deployment is triggered by pushing to a git repository — no separate CLI deploy commands or dashboard interactions are required.
The "run locally" capability is a design objective: tau simulates the full production platform on a developer's laptop, so the same git push workflow used in production functions identically in a local environment. Applications are defined via configuration files in the repository; the platform resolves DNS, routes traffic, executes functions, and serves static assets without manual provisioning steps.
Placed in Platforms / Assess / center.
tau addresses the same gap as managed PaaS offerings (Heroku, Render, Railway) — reducing deployment complexity to a git push — but as a self-hostable stack with no per-seat or per-compute pricing. The combination of serverless functions, object storage, CDN, and DNS in one deployable unit is broader than most comparable self-hosted tools, which typically handle one of these concerns.
The local simulation capability is a meaningful engineering property: development and production environments share the same execution model, which eliminates a class of "works on my machine" deployment bugs.
Center position reflects early adoption signal relative to the ambition of the scope. The platform covers significant ground — more moving parts means more surface area for operational issues, and the community track record is thinner than comparable tools like Windmill (inner) or Kestra (inner) already on this radar. The concept warrants watching; inner position requires a clearer production deployment story.
Trial gate: tau deployed locally serving at least one serverless function and one static site, with a verified git-push deployment cycle end to end.
