Open-source AI coworker · Local-first · Windows

Delegate the outcome.
Keep the control.

LocalAI Cowork is the open-source, local-first alternative for agentic work alongside Claude Cowork and Microsoft Copilot Cowork. Bring your own model, files, and tools into one inspectable Windows workspace.

No LocalAI Cowork account Bring your model Approval controls Apache-2.0
LocalAI Cowork desktop app with a local Ollama model ready for a task
LocalAI Cowork running locally with an Ollama model. Interface language and provider are configurable.

One workspace, less tool switching

Built for work that needs to finish.

Chat is only the starting point. LocalAI Cowork keeps the task, context, tools, progress, and review surface together.

Outcome-driven tasks

Describe the result you want, keep the plan visible, and follow progress instead of reconstructing it from chat history.

Files in context

Choose the files and folders a task can use. Keep relevant local material close to the work.

Bring your model

Use Ollama locally or connect compatible providers such as OpenRouter and OpenAI-style endpoints.

MCP and local tools

Connect MCP servers, terminals, and reusable skills when the task needs more than a text response.

Review before risk

Permission policies and approval states keep sensitive operations visible before they run.

Reusable workflows

Turn repeated instructions into templates, skills, and pipelines instead of starting from zero.

Local-first by design

Your workspace should stay yours.

LocalAI Cowork is a desktop application, not a hosted account. You decide which model and tools receive a request.

Credentials use the OS vault

Provider keys and connector secrets are stored through Windows Credential Manager instead of ordinary app settings.

Secrets are removed from persistence and logs

Sensitive fields are stripped from persisted frontend state and redacted from bounded audit output.

No website tracking

This site ships no analytics, advertising scripts, third-party fonts, or tracking cookies.

Inspect the implementation

The source, release workflow, dependency audits, checksums, SBOM, and build attestations are public.

A visible working loop

From request to reviewed result.

Choose the context

Select a workspace, relevant files, a model, and the tools the task may use.

Describe the outcome

State what “done” looks like. LocalAI Cowork keeps execution and progress in one place.

Review and approve

Inspect important actions, outputs, and audit information before shipping the result.

Choose the workspace that fits

An open-source route to AI cowork.

LocalAI Cowork focuses on inspectable source, local model support, and user-controlled tools. These guides explain the trade-offs honestly—without pretending every product solves the same problem.

Built in public

Use it. Inspect it. Improve it.

LocalAI Cowork is early software and welcomes concrete bug reports, feature ideas, documentation fixes, and focused pull requests.

Useful details

Questions before you install.

What is LocalAI Cowork?

LocalAI Cowork is an open-source Windows desktop workspace for AI-assisted work. It combines chat, tasks, file context, model configuration, MCP servers, tools, approvals, and audit information in one application.

Is LocalAI Cowork part of the LocalAI project?

No. LocalAI Cowork is an independent project and currently uses Ollama for local model execution. It is not affiliated with or endorsed by the separate LocalAI project at localai.io, Anthropic, or Microsoft.

Can it use a local AI model?

Yes. Ollama is supported for local model execution. You can also configure OpenAI-compatible or OpenRouter endpoints when a hosted model fits the task better.

Does my data ever leave the device?

The app itself runs locally and the current release has no implemented telemetry sender. Requests leave the device only when you choose a remote model endpoint, connector, MCP server, or web-enabled tool. Those services receive the information needed for the action you requested.

Is LocalAI Cowork free?

Yes. The application is available under the Apache License 2.0. A remote model provider may charge for its own API usage; local Ollama use does not require a provider subscription.

Which operating systems are supported?

The current installer and smoke tests target Windows 10 and Windows 11. The Tauri foundation can support more platforms later, but Windows is the maintained release target today.