Question 1

Can we not just install it ourselves?

Accepted Answer

You can. The work is deciding what it may touch and which actions must wait for a person. OpenClaw's computer-command approvals do not automatically stop an email send or a change through a connected tool. We build and test those review steps around your work, along with access limits and support after launch.

Question 2

Is OpenClaw free?

Accepted Answer

The software is free under an open licence. Running it still involves hardware or hosting, AI usage and upkeep. You can use a computer you already have; cloud AI usage is billed by the provider you choose. Costs depend on the work and usage, so we estimate them with you before you commit.

Question 3

Where does our data go?

Accepted Answer

OpenClaw stores its working state on the machine you choose. Content sent to a cloud AI is processed by that provider; messages and records in Slack, WhatsApp, email or other connected services are also handled by those services. Running the AI locally avoids cloud-model requests, but does not make connected services local. We review the whole route your data takes before launch.

Question 4

Will it run on Windows?

Accepted Answer

Yes - Mac, Windows and Linux are all supported. What matters more than the operating system is that the machine can stay switched on, because the assistant is only reachable while it is running. A laptop that closes at six will not answer at nine. Most teams use a spare desktop, a small office machine, or a cheap rented server.

Question 5

How long before it is actually useful?

Accepted Answer

One to two weeks to set up, and it does real work from the first week. You approve everything at the start, which feels slow for a few days and is the point - you are teaching it where your lines are. The corrections tail off as it settles into how you work, and most people stop approving the routine things by the end of the first month.

Question 6

Can more than one person use it?

Accepted Answer

Yes. You name who is allowed to talk to it and it ignores everyone else, including in group chats, where it stays quiet unless spoken to directly. Most teams start with one or two people, watch how it behaves for a fortnight, then widen it. You can give different people different limits.

Question 7

What happens when it gets something wrong?

Accepted Answer

You see it before anyone else does, because the things that reach a client are the things we hold for approval. When a draft is wrong you correct it and tell us if it keeps happening - usually it means a limit is set in the wrong place, which we adjust. The history is readable in plain language, so working out what it did and when is not a forensic exercise.

Question 8

What if we want to stop working with you?

Accepted Answer

Everything keeps running. The computer, the accounts and the software are yours, the software is open, and there is no licence for us to withdraw or a hosted service for us to switch off. We will hand over the configuration and what it has learned so somebody else can pick it up.

Task	Agent	You
Catching you up on a thread and flagging what needs a decision
Drafting the reply in Gmail
Copying invoice totals into your spreadsheet
Sending anything to a client
Paying, or changing anything live

OpenClaw setup, on a machine you own.

What is OpenClaw?

What an OpenClaw agent does, and what stays with you

Which apps does OpenClaw work with?

Is OpenClaw safe?

Review for the actual job

Only the people you name

Only the accounts you connect

OpenClaw, Hermes or Grok Bot?

How do you get OpenClaw set up?

A call

Setup

Live

After

See the work in practice.

Common questions about OpenClaw