Dystr

Dystr is an AI analysis tool designed to help engineering teams increase velocity. It supports fast iteration on projects by organizing work in isolated project workspaces, so teams can run code and use AI workers without disrupting active efforts.

A core capability is managing how analysis runs and how work gets dispatched. Dystr can handle one-off AI requests, and it can also create background workers to automate repetitive processes. To support ongoing collaboration, it enables teams to preserve and fork analysis, helping keep datasets and calculations aligned while multiple contributors build on the same work.

Dystr also provides a hybrid way to operate AI workers. Work can be triggered by email, scheduled runs, or external systems, and you maintain control while integrating these automated tasks into existing engineering workflows.

Dystr’s ecosystem includes RunMat, a GPU-accelerated runtime for math. Run math in the browser, on desktop, or via a CLI, with automatic GPU/CPU scheduling and MATLAB input syntax. RunMat positions itself as extremely fast for large elementwise math workloads.

What does Dystr help engineers do?

Dystr helps engineers accelerate analysis workflows, so teams can iterate faster and spend more time on higher-priority work.

How does Dystr manage work without disrupting ongoing projects?

Dystr uses isolated project workspaces that act like sandboxes for running code and AI workers.

Can Dystr automate tasks instead of only handling one-off requests?

Yes. Dystr can create background workers to automate repetitive processes, not just respond to single AI requests.

How are Dystr AI workers triggered?

AI workers can be triggered by email, scheduled runs, or external systems, fitting into existing workflows.

What is RunMat within the Dystr ecosystem?

RunMat is a GPU-accelerated runtime for math that runs in the browser, desktop, or as a CLI, with MATLAB input syntax and automatic GPU/CPU scheduling.

Is RunMat designed for elementwise math workloads?

RunMat is positioned for very large elementwise math, highlighting speed improvements for that use case.