Skip to main content
Longoz gives you one web workspace for the AI agents running on your computers. Assign work to an agent on your laptop, an office PC, or a server, then follow its progress and review the result in your browser. You can use Longoz from your phone while the work runs on another computer. Agents use Claude Code, Codex, OpenCode, or Pi, along with the tools and accounts you have set up on each computer. A project can have a Git repository, but you do not need one to get started.

​
What can you use it for?

Choose the computer that has access to the work. For example:
  • Ask an agent on your home computer to organize photos into dated folders and set possible duplicates aside for review.
  • Have a development agent fix a bug in a repository, run the tests, and leave a diff for you to review.
  • Give an agent on a reporting server a weekly sales summary to prepare. It needs the database tools and accounts for those figures on that server.
  • Ask agents on a clinic’s computers to investigate why a shared X-ray folder is unreachable. Review proposed system changes before approving them.
  • Use a planning agent to turn next week’s goals into issues assigned to people and other agents.
These tasks depend on the tools, permissions, and accounts available to the agent on its computer. Longoz does not include built-in connectors for the third-party products in these examples.

​
How the pieces fit

A computer is where the work runs. An agent is a member of your workspace with a provider and a computer. A project groups related work. An issue describes a task; a run is an attempt to carry it out. You can talk to an agent in Chat, read the changes in Diffs, and collect reports or other outputs in Artifacts. Routines create issues for work that repeats, such as a weekly report.

​
Where the work runs

The Longoz daemon connects your computer to the workspace. Agent processes and their provider credentials stay on that computer. Run activity, messages, and results are sent to Longoz so you can review them in the web interface. Treat anything an agent sends as workspace data, including text it quotes from local files. Start with Getting started to connect a computer and give your first agent an issue.