English · Deutsch · Français
Tale — The open-source workspace for teams and AI agents
Turn company problems into tasks your team and AI agents can solve together.
Tale gives teammates and AI agents a shared project workspace. Add tasks to the board, assign people or agents, follow the work, and review reports and delivered files. Keep the brief, discussions, project knowledge, and results together. Use agents to research a question, review documents, prepare reports and marketing materials, or build a website, app, or internal tool.
Choose each agent’s runtime, model, skills, and tools. Equip a manager agent to delegate ready tasks and coordinate follow-up work. Agents run in persistent sandbox workspaces, with concurrency limited by your configured capacity.
Use your own provider API keys or supported subscriptions with compatible agent runtimes. See runtime and credential support for the available combinations.
Self-host Tale on your own infrastructure or use the managed Cloud service. The code is MIT-licensed. Community and Enterprise include the same product features; Enterprise adds professional operation and support. See plans and pricing for the current offer.
Select a screenshot to view it at full size. The captures show the English interface.
From tasks to reviewed results
- Describe the work. Create a project task with the problem, source files, and completion criteria.
- Assign people and agents. Configure project agents for the work, choose their tools, and start the tasks you want them to handle.
- Coordinate and review. Follow progress on the board, steer agents with @mentions in task comments, and inspect their reports and files. A manager agent can delegate ready work when granted the required tools.
- Repeat a defined process. Use a versioned automation when the work needs scheduled starts, workflow steps, or connector approvals.
Example: review a launch brief
With a project agent and compatible model credentials configured, adapt this illustrative task brief to your own source files:
Compare the attached launch brief and meeting notes. Produce a Markdown report listing conflicting dates, missing owners, and open decisions. Cite the source file and passage for each finding. Separate confirmed facts from questions, and leave the source files unchanged.
Before accepting the result, open the delivered report, verify its citations against both files, and check that each requested category is covered. Ask for corrections in the task when evidence is missing. The task review guide explains how to request changes or accept completed work.
Evaluate Tale for your team
Which agent runtimes can I use? Tale includes Claude Code, Codex, Cursor, Gemini CLI, Hermes, OpenClaw, OpenCode, Pi, and Qwen Code. Availability depends on your deployment, credentials, and sandbox capacity. Check the runtime compatibility matrix for credential paths, tools, and conversation limits.
Can I use an API key or an existing subscription? Stored provider API keys use Tale's model gateway. Supported vendor subscriptions work only with compatible runtimes, cannot power ordinary Chat, and bypass Tale's gateway metering and spending caps. Read the credential and cost details before choosing a connection.
What does self-hosting require? Start with Docker and Compose, storage for images and persistent data, and credentials for a supported model provider. Production also needs DNS, TLS, backups, and access controls. The self-hosted quickstart covers the local setup and links to production preparation.
Where does my data go? Application records, searchable knowledge, and original files have separate storage settings. Model providers, connectors, and external tools can process data outside those stores; self-hosting alone does not keep every request local. Review data residency and the runtime's credential and network behavior.
What differs between Community and Enterprise? Both include the same product features under the MIT license. Enterprise adds professional operation and support. See plans and pricing for the current service terms.
This is the beginning of the README. Read the rest at the original. Read the full README ↗
