AI agents for work you do on repeat

Hand off recurring work. Stay in control.

Build AI agents around your tools and the way your team works. Run them in the background, set approval points, and review the results.

Download for MacApple Silicon

Set up with CLImacOS · Windows · Linux · Node.js 22+

See the work and the decisions waiting on youProduct walkthrough

The demo shows the operations dashboard, an agent detail page, and a pending approval review. Agents work in the background and pause at the approval points you set.

From an AI task to a team process

Turn a successful AI task into an agent you can run again.

Use Codex or Claude Code to create and improve your agent files. Run them in AgentUse on a schedule, inspect the results, and approve the actions you choose. The workflow stays in your repository, ready to review and change.

Example run · weekly team report · illustrative data

Keep recurring work moving

Run the same defined job manually, on a schedule, from an event, or in CI.

weekly-report.agentuserun #218done

triggerschedule · every monday 08:00also runsmanual · on: new-ticket · ci: pull_requestdefinitionagents/weekly-report.agentuse · unchanged since run #204

Keep consequential actions under review

Put sending, publishing, deleting, and other external changes behind approval.

08:00:04read linear.issues (updated this week)08:00:11read github.pulls (merged this week)08:00:37write report.md

pausedsend slack.post #leadershipwaiting for approval · 1h 12m

approved by Leon Ho · 09:12:40

09:12:41send slack.post #leadership · ok

Review the result, not just the run status

Open the deliverable, check what needs attention, and inspect the tool calls and approval decisions behind it.

report.md · example deliverable

Weekly team report

Completed: Release notes drafted from merged pull requests and this week’s Linear updates.
Needs attention: The onboarding issue has no owner. Confirm who will take the next step.
After approval: Report shared to #leadership. No issues were reassigned.

Result

report.md

Tool calls

4 · 0 failed

Artifacts

1 file

Approvals

1 approved

full transcript, tool inputs and outputs, and the approval decision are kept with run #218

Get started

Start running your agents.

Start with the Mac app, use the CLI and web dashboard on Windows, Linux, or macOS, or host AgentUse on your own server.

Mac app

Download for Apple silicon

The AgentUse runtime is included. Move the app to Applications, open it, and follow the guided setup.

Download for Mac

CLI + web dashboard

Start with one command

Requires Node.js 22+. Use the guided browser setup locally or Terminal mode for servers and CI.

$npx -y agentuse@latest setup

Agent fleet · Markdown-defined agents ready to run

AgentUse agent fleet in light mode showing Markdown-defined agents ready to run

For work you need to trust

Test it before you leave it running.

Test without real side effects

Exercise workflows and approval paths with isolated state.

Keep the operating process

Readable agent files live in your repository and change with review.

Workflow guides

Start with a workflow. Make it yours.

Each guide includes a prompt for Codex or Claude Code to build and test an AgentUse agent around your tools, process, and approval points.

Copy the prompt → Build and test with your coding agent → Run in AgentUse

Email · Google Drive · Card statement

Collect the month’s invoices

You get: An invoice bundle matched to your statement, with missing documents flagged.
You review: Check unmatched charges and approve any vendor follow-up.
Runs: Monthly, when your statement is ready

Read workflow guide

DataForSEO · Change log

Track what changed in SEO

You get: A weekly journal connecting crawl and ranking changes to your recorded decisions.
You review: Resolve conflicting evidence and review proposed changes in direction.
Runs: Weekly, on your schedule

Read workflow guide

Slack · Notion · Linear

Turn feedback into a backlog

You get: A prioritized proposal linking customer feedback to existing work and new issue drafts.
You review: Approve the exact backlog changes before they’re applied.
Runs: Weekly, before backlog review

Read workflow guide

Need it implemented?

Need the workflows designed and implemented for you?

Start with one recurring function and launch it in your environment with AgentUse Studio.

Explore AgentUse Studio →

Need agents running while your Mac is offline?