Tale

Tale is an MIT-licensed workspace for people and AI agents working on shared projects. It brings task context, agent execution, and human review into the same project. Teams can self-host it or choose a managed deployment.

What it does

Teams describe work as tasks with context, owners, and completion criteria. An agent can be assigned suitable work and started from the task. Its report and deliverables return to the task for review. Teammates can request revisions before accepting the result.

Projects keep related files, instructions, and discussions together. Agents use configured runtimes and models, with work executed in persistent sandbox workspaces. Parallel execution depends on available capacity. Compatible runtimes can use supported API credentials or existing subscriptions; compatibility depends on the selected runtime and provider.

Who it is for

Tale is intended for teams coordinating agent work across software delivery, marketing campaigns, research, document preparation, and operations. It suits teams that want shared visibility into assignments and results, and those that want to operate an open-source workspace on their own infrastructure. Self-hosting requires deployment administration and model-provider configuration.

Standout capability

The task review loop connects agent execution to a shared project board. Completed agent work produces a report and deliverables, then enters review; a human reviewer decides whether it is done. Persistent project workspaces let agents reuse working files across tasks while teammates keep direction and review attached to the work.