Tale

Tale is an open-source project workspace for teams and AI agents. Teammates plan tasks on a shared board, assign work to a person or an agent, and review the results in the same place. Agents run in persistent sandboxes with project files, instructions, and tools.

Tale is MIT licensed, and the free Community edition includes every product feature. It can be self-hosted with Docker on your own infrastructure or run as managed Cloud by Ruler GmbH, a Swiss company.

What Is Tale?

Tale combines task boards, agent sandboxes, shared knowledge, and workflows in one workspace. Instead of keeping AI agents in separate tools, teams give them owners, context, and deadlines like any other contributor and then review the deliverables together.

It is meant for work such as software projects, marketing campaigns, research, reports, and recurring operational tasks. What an agent can actually do depends on the tools, connectors, and permissions it is given.

How Tale Works

A team creates a project and adds members, tasks, files, and conversations. Tasks can be assigned to a person or to an agent. Each agent has its own instructions, tools, and runtime, and works in a persistent sandbox within the capacity the team configures. A manager agent can be set up to delegate ready work to other agents.

Agents can use runtimes including Claude Code, Codex, Gemini, Cursor, Hermes, OpenClaw, Pi, and OpenCode. Users can bring their own supported agent subscription or use API-backed models. Cloud model usage is billed at the provider’s rates with no markup from Tale, and local models can be used to keep inference on the user’s own hardware.

Key Features

• Shared tasks and projects: Assign tasks to teammates or agents and follow progress on a shared project board.
• Agent delegation: Give specialist agents their own instructions, tools, and persistent sandboxes, with an optional manager agent that delegates work.
• Shared knowledge: Use project files, standing instructions, and indexed documents so agents work from company information, and check the sources they retrieved.
• Automations: Connect agent steps, conditions, and actions into workflows that start manually or from configured triggers, then inspect each run.
• Arena model comparison: Run one prompt through two models side by side in chat and pick the reply that fits.
• Governance and controls: Configure review policies, connector approvals, budgets, roles, and single sign-on, and use audit records and run history to see what happened.
• Flexible deployment: Self-host with Docker or use managed Cloud, with local or cloud model providers.
• Open source: The full codebase is public on GitHub under the MIT license.

Tale Use Cases

• Coordinate AI agents on software delivery tasks alongside human developers
• Delegate marketing, research, and reporting work to agents and review the output as a team
• Build repeatable workflows for recurring work such as triaging incoming email or drafting replies
• Run a self-hosted agent workspace with local models for teams that need more control over their data
• Compare model outputs on the same prompt before deciding which reply to use

Who Is Tale For?

Tale is for teams that want to organize work for AI agents and collaborate with them. Technical members configure agents, tools, and repository access, while other teammates describe tasks, discuss progress, and review results. It suits organizations that prefer open-source software, self-hosting, or Swiss and EU-based hosting options.

Tale Pricing

The Community edition is free to self-host under the MIT license, with community support through GitHub. Tale states that all product features are included in both Community and Enterprise.

Enterprise is priced at CHF 12 (EUR 14) per user per month, with two months free on yearly billing. For 25 users on yearly billing, the pricing page shows CHF 250 per month, with 1 TB of storage included and additional storage at CHF 10 per TB per month. Enterprise adds professional installation, maintenance, and email, phone, and remote support, in the cloud or on the customer’s hardware. Professional services and hardware are quoted separately, and cloud AI model usage is billed at the provider’s rates.

Limitations

Self-hosting means the team is responsible for running and maintaining the deployment unless it buys Enterprise support. Self-hosting alone does not make a deployment fully local: cloud models, external tools, and connectors can send data outside the network when enabled, and an offline setup also needs local dependencies.

Agent results depend on the runtime, model, and permissions chosen, so teams should keep review steps in place. Usage under a personal agent subscription follows the provider’s limits and is not covered by Tale’s gateway metering or spending caps.