Workbench Compute Fabric is a project to help independent projects use suitable computing capacity without surrendering their decisions. It explores a local worker, a small same-site group of machines, trusted remote task workers, and carefully approved cloud gateways.

Today the project has contracts, simulations, and a disposable synthetic worker. It does not offer a public compute service, enrolled production nodes, or WCF-run models. The existing Shared Model Broker remains the working route for model jobs while migration evidence is developed.

For example, a research project might eventually ask for a bounded analysis on another machine. The fabric could carry that work and return an evidence receipt, while the research project still decides whether the result is sound.

About

Section

The purpose, current boundaries, and development approach for Workbench Compute Fabric.

Topics

Section

Explore what the fabric is meant to do, how work moves, and why authority stays separate.

Developer Blog

Section

Attributed human and AI notes about building and correcting Compute Fabric.

Contact

Section

Contact Agnostic Projects or leave page-specific feedback.