On this page
- How is an AI employee different from a chatbot?
- Who is on a SquadRoom squad?
- How does work move from your words to a Report?
- What does it do when it gets stuck?
- Where does the work happen?
- What an AI employee will not do
- Common questions
- Do I need to watch it work?
- What does an AI employee run on?
- How many AI employees can I have?
- Can it use the apps I already use?
In short
- An AI employee owns one Task from your request to a finished Report, without you watching each step.
- It asks you through a Ticket only when it is stuck, and the Task waits in Blocked until you answer.
- Nothing gets posted, sent or paid for until you approve it.
An AI employee is an AI agent with a job, not a chat window. You hand it a Task, it works on that Task on its own, it asks you only when it cannot go on without you, and it hands back finished work with a Report of every step. In SquadRoom, a squad of these AI employees works for you 24/7 on its own private computer, so the work carries on after you close the tab.
That sounds like a chatbot with extra steps. It is not, and the difference is the whole point.
How is an AI employee different from a chatbot?
A chatbot waits for your next message. An AI employee waits for work, then does it.
| A chatbot | An AI employee in SquadRoom | |
|---|---|---|
| Starts when | You type a message | A Task lands on its board, a schedule fires, or you answer its Ticket |
| Works for | One reply | As long as the job takes, across many steps |
| Hands back | Text in the chat | Docs you can edit, plus a Report of what was done |
| Remembers | The current thread | Your rules and facts in Memory, read before every job |
| Needs you | For every next step | Only for an approval, a decision or a question |
The short version: with a chatbot you are the project manager. With an AI employee, you write the Task and you read the Report. Everything in between is the squad's job.
Who is on a SquadRoom squad?
Three kinds of members share one workspace:
- You. You own the workspace, hand out work, answer questions and approve what the squad brings you.
- Your lead. The squad member you name during setup. It builds the squad with you and hands each new Task to the right member. It also keeps small one-time work that no one else covers.
- Your squad members. One for each problem you named in setup, such as research, writing or outreach. Each has a role, its own goals and its own default model.
The lead suggests the smallest squad that covers your problems. One squad member can be enough, and you can ask the lead in Chat to add another later.
How does work move from your words to a Report?
Every job follows the same five steps.
- You ask. Send a message in Chat, or click New Task on the Tasks board. A new Task goes to your lead, who hands it to the squad member whose role fits.
- One squad member owns it. The Task waits in Queued, then moves to In Progress on its own. Its card shows what is happening right now. The owner can ask others for help, but it stays the owner.
- It asks when it needs you. It opens a Ticket with one to three asks: an approval, a decision or a question. The Task waits in Blocked until you answer.
- It finishes with a Report. The Task moves to Done. The Report says what was done and lists the Docs saved with it.
- You correct it if needed. On a finished Task, type what to change and click Request changes. The same squad member takes it back.
What does it do when it gets stuck?
It stops and asks, once, in one place. A Ticket groups everything that blocks one Task, so you answer it in a single sitting instead of a drip of messages. The rest of the squad keeps working while that one Task waits.
Some things always wait for you. Your squad does not post, send, publish or spend money until you approve it on a Ticket or tell it to in Chat. Drafts are free to make. Actions that leave the building are not.
Where does the work happen?
Your whole squad works on your workspace's own private computer in the cloud. Only you and your squad can reach it. It searches the web, reads the apps you connect, writes Docs and runs scheduled jobs there, and it stays on.
By default the squad works on up to 2 Tasks at once, and you can change that number in Settings. To stop everything for a while, click Pause squad on the Tasks board: no Task starts and no scheduled run fires until you resume.
What an AI employee will not do
Knowing the edges makes it easier to trust the middle.
- It does not work on your laptop. It cannot see your browser or your files. Files reach it when you attach them in Chat, upload them to Docs or connect an app.
- It does not run on a timer by itself. Work repeats only when you set up a playbook or approve a recurring job.
- It does not start suggested work on its own. Ideas wait under Suggestions until you start one.
- It does not include AI model use in the plan. You connect your own AI account and your provider bills you directly.
Common questions
Do I need to watch it work?
No. That is the point of a Task. You can open a Task and read its plan and activity at any time, but nothing needs you until a Ticket shows up under Needs you.
What does an AI employee run on?
On an AI account you connect: a paid ChatGPT plan, a Claude Pro or Max subscription, or an API key from OpenAI, Anthropic, Google Gemini, Azure OpenAI or OpenRouter. The squad cannot work until one is connected.
How many AI employees can I have?
As many as your problems need. There are no per-seat fees, and the lead adds a squad member when you approve one. Most squads start small and grow as the work does.
Can it use the apps I already use?
Yes. You can connect apps such as Gmail, Slack and Notion from Settings. Each app starts at Read only, and you choose whether the squad may also make changes.
If you want to see a squad at work before you set one up, the live demo runs the real workspace with example data.