Schedule AI agent tasks on your Mac

ShiftHands lets you schedule AI agents to handle recurring work on your Mac. Give an agent a daily or weekly task, and it runs on schedule, posts a result in its chat and shows you anything that needs review. Your Mac must stay on with ShiftHands open for work to run on time.

ShiftHands is a Mac app for running and supervising a team of AI agents. The agents run on your own Apple-silicon Mac; you can monitor their work, chat with them and respond to flagged command approvals from the Mac app or a desktop or phone browser.

What can you schedule an AI agent to do?

Anything repeatable with a result you can check. Here is what the instructions look like and where the result lands.

Inbox triage

Every morning

“Sort my unread email, flag anything urgent and draft replies.”

Result: A digest and drafted replies in the agent’s chat.

Competitor check

Weekly

“Check these five pricing pages and changelogs for changes.”

Result: What changed, with links to the sources.

Looking for a tool to handle recurring SEO research and reports? Our AI SEO agents comparison covers nine options for research, content work, automation and reporting.

Need a repeatable competitor watch instead of general scheduled work? Our AI agent for competitor monitoring guide shows how to check named websites, compare findings with earlier reports and review source-linked briefings.

Website uptime check

Hourly on Pro

“Load our main pages and report anything down or slow.”

Result: A status summary, with problems marked Needs review.

Recurring research

Daily or weekly

“Find new posts on this topic and summarise the useful ones.”

Result: Findings and suggested follow-up tasks.

These are illustrative set-ups, not tested results or accuracy guarantees. An agent can only use the accounts, tools and browser access you give it.

Want an agent to prepare social posts and research on a recurring schedule? Our AI agent for social media guide shows how to produce drafts, review the results and keep publishing decisions in your hands.

See it working

See scheduled AI agents working

This is the real ShiftHands app. Open the Automations tab for scheduled jobs and their result cards; the other tabs show the chats, projects and browser those jobs use.

Virtual Environment

Your AI team, in a 3D office.

Every ShiftHands plan, Free included, comes with the Virtual Environment: a 3D office in the app where your agents work at their desks, grouped by project. Walk over, pull up a chair, and watch your crew get it done.

Walking through the ShiftHands Virtual Environment: realistic agents typing at their desks in colour-coded project areas, glass walls and the city outside the windows.
The Automations Wall showing Upcoming, Running now and Completed automations above a long row of agent desks.

The Automations Wall

A big screen shows today’s Upcoming, Running now and Completed automations. Step up to the tablet to take a closer look at any run.

Seated at your desk in the 3D office, with the desk panel listing Upcoming, Completed, Currently working and New replies.

Your boss desk

Sit down to see every project and every agent at a glance, and chat with anyone without leaving your chair.

Close up in the 3D office: realistic agents with real faces and hair working at their desks, with name tags above their heads.

Watch your agents work

Every agent has a desk in its project area. Agents running an automation head to the Automations room. Walk up to any agent and chat.

  • Free and Pro
  • No extra AI usage for the 3D view
  • Mac app and web, including phone browsers

Rendered in the ShiftHands demo workspace with made-up names. Open the Virtual Environment from the sidebar or from Chief of Staff in the app. Agents still run on your Mac and use your connected AI accounts.

How scheduled AI agent tasks work

A schedule belongs to an agent, so every run uses that agent’s project, model and settings.

1. Schedule

Give an agent a task and choose when it runs. The time follows your account’s time zone.

2. Run on your Mac

At that time, the agent does the work on your Mac, using the model and tools you set up.

3. Result in chat

The run posts a result card in the agent’s chat, marked Done or Needs review.

4. You review

Read the result, reply with a next step, or answer a flagged command if the agent asked.

Approval only comes up when a command is flagged under that agent’s settings. If a run fails or needs attention, check its status and result in Automations, then give the agent its next instruction.

Set up your first scheduled AI agent task

New to running agents on your Mac? Start with our AI agent for Mac guide to set up your workspace, connect a model and create your first agent.

Six steps, all inside the Mac app.

1. Open ShiftHands

Use an Apple-silicon Mac with macOS 13 or later. See where ShiftHands agents run.

2. Pick a project and agent

Choose an existing agent or create one with a name and a one-line role.

3. Write the task

In the agent’s automations, tap Add automation. Give it a name and fill in “What to do each time”.

4. Choose When

Pick Daily, Weekdays, Weekly or One time and set the time. On Pro, Hourly and Every few hours are also available.

5. Check permissions

Decide whether “Ask before risky steps” stays on. If the job uses a browser, install Ego Lite first.

6. Check the first run

When it runs, find it in Automations and open its result card in the agent’s chat.

You can also ask your Chief of Staff in chat to create an automation for an agent.

Run results

Review upcoming, running and completed automations

The automations view lists what is coming up, what is running now and what has finished.

Upcoming3

  1. Uptime checkSite MonitorRunning
  2. Support queue sweepSupport Inbox12:45 PM
  3. Ad spend pacing checkAds Manager1:15 PM

Completed4

  1. Expense receipts syncBookkeeper · 11:32 AMDone
  2. Outreach follow-upsOutreach Assistant · 10:07 AMNeeds review
  3. Backup checkSite Monitor · 9:31 AMDone
  4. Weekly SEO reportSEO Assistant · 8:39 AMDone
Completed automations in ShiftHands: finished and needs-review jobs, and the Backup check result card in the Site Monitor chat.

Examples from the ShiftHands demo workspace.

  • Upcoming jobs show their next run time.
  • Running jobs show live progress, and Watch live when they use a browser.
  • Finished runs are marked Done or Needs review.
  • Each run’s result card sits in that agent’s chat, with the rest of its history.
  • Turn an automation off, edit its schedule or delete it from the same list.

Prefer one screen for every agent and job? See the ShiftHands AI agent dashboard.

Daily schedules, hourly runs and plan limits

The plan sets how many automations you keep and how often they can run.

Free, $0

2 automations at a time. Each runs at most once a day: Daily, Weekdays, Weekly or One time.

Pro, $49.99 a month or $499 a year

No plan cap on automations. Adds Hourly and Every few hours (2, 3, 4, 6, 8 or 12 hours).

Still limited by

Fair-use caps, your AI provider’s limits and what your Mac can handle.

Full plan details are on ShiftHands pricing.

Can AI agents work while you sleep?

Yes, if the Mac that hosts them stays on, awake, connected and running ShiftHands. It is your computer doing the work, so it has to be available.

Plugged-in desktop Macs

A Mac mini or Mac Studio left on runs automations on time, even at night.

If the Mac sleeps

Agents pause. Scheduled work missed while the Mac slept runs when it wakes.

If power or internet drops

Nothing runs until the Mac is back. Messages you send meanwhile are delivered when it reconnects.

Planning a dedicated machine? Read the Mac mini AI agent setup guide.

Approvals and browser

Browser work and unattended approvals

Unattended doesn’t mean unrestricted. A scheduled run can stop and wait for you.

Approval behaviour depends on each agent’s settings. Not every file change or send is gated.

Ads Manager needs your OK: raise the Fall retargeting daily budget from $40 to $60, with the exact command and Allow once and Deny buttons.

An approval card from the ShiftHands demo workspace, with made-up names.

  • A flagged command shows a plain-English line, a Low, Medium or High risk level and the real command.
  • Allow once or Deny. Typing “yes” in chat never counts as approval.
  • Browser jobs need Ego Lite, a free agent browser from Citro Labs. It’s a separate download with its own terms, macOS only.
  • If a site needs a sign-in or code, take control of the browser, do that step, then Return to agent.
  • A run waiting on you is not finished; it shows as needing review.
  1. Watch it liveA live view of the agent’s browser beside the chat, with the site it’s on. Running automations have Watch live too.
  2. One tab per browserWhen helpers browse too, each browser gets its own tab, so you can switch between them.
  3. Take controlType into the page, use the address bar, sign in. The agent waits.
  4. Return to agentOne tap hands control back and the task carries on. When the agent is done with the browser, you see a short summary.
The live browser view in ShiftHands: the Support Inbox browser on a store admin page asking for a 6-digit code, with Return to agent, Can’t do it and Cancel task buttons.

Phone review

Check scheduled work from your phone

Open shifthands.com in your phone’s browser and sign in. Your Mac still runs the agents; the phone is where you check in.

  • See upcoming and completed automations, marked Done or Needs review.
  • Open a result card in the agent’s chat and reply with the next step.
  • Answer flagged commands with Allow once or Deny.

More on this in our guide to manage AI agents from your phone.

A phone showing the Completed automations list, each run marked Done or Needs review.

Models, usage and running costs

ShiftHands doesn’t sell model credits. Scheduled work uses the AI accounts you connect.

Your model account

Sign in with Claude or ChatGPT, or add an API key. Each provider’s limits and pricing apply.

Run frequency

More frequent or longer jobs use more of your provider’s allowance.

Message limit

On Free, automation runs don’t count toward the 25 agent messages a day. Only messages you send do.

Your Mac

It has to stay on, which costs power. Heavy schedules can slow other work on the same Mac.

No cost savings are promised; what you spend depends on your provider, schedule and Mac.

Who should schedule AI agent tasks?

It suits repeatable work on your own Mac, with output a person can check.

Scheduling fits if

  • You repeat the same research, checks or inbox work every day or week.
  • You want each run’s result in one place, with a chat to follow up.
  • You have an Apple-silicon Mac that can stay on.
  • You want flagged commands to wait for your decision.

Look elsewhere if

  • You need a cloud scheduler that runs with no computer of your own.
  • You need webhook or event triggers between other apps.
  • You need guaranteed exact-minute timing.
  • You only have Windows or an Intel Mac.

Frequently asked questions

How often can a scheduled AI agent task run?

On Free, at most once a day: Daily, Weekdays, Weekly or One time. Pro adds Hourly and every 2, 3, 4, 6, 8 or 12 hours.

How many automations does the Free plan include?

Two at a time, along with 25 agent messages a day, 1 project and 3 agents.

Do automation runs count toward my message limit?

No. The Free limit counts messages you send. Automation runs, approvals and an agent’s own follow-up steps don’t count.

What happens if my Mac is asleep at the scheduled time?

The agent can’t run while the Mac sleeps. Missed scheduled work runs when it wakes.

Can a scheduled agent run without approvals?

It depends on that agent’s settings. With “Ask before risky steps” on, it pauses for your OK on risky commands. It can still browse and send messages on its own.

Can scheduled jobs use a browser?

Yes, with Ego Lite installed. Running browser jobs have Watch live, and you can take over for a sign-in.

Can I review scheduled work on my phone?

Yes. Sign in at shifthands.com in your phone’s browser to see results and answer flagged commands.

Does scheduling cost extra?

Not on top of your plan. Runs use your own AI provider account, under its pricing and limits.

Give an agent its first recurring task

Download the Mac app, pick an agent and set when it runs. Check the result in its chat.

Download ShiftHands for Mac

Needs an Apple-silicon Mac (M1 or newer) with macOS 13 or later. For an always-on host, see the Mac mini AI agent setup guide.

Screens and video come from the ShiftHands demo workspace with made-up names. Features and prices were checked on October 27, 2026 and can change.