The questions you should ask us.
We already pay for ChatGPT and Claude. Why would we pay you?
You are not buying access from us: you already have it, and you should keep buying it directly. You are buying the four hours in which someone configures it against your actual work while you watch, and the written record that lets you do it again for the next person who joins.
Why Claude?
Because it is the tool we are certified on, and the one we know well enough to set up properly. We are Claude-first, not Claude-only: Claude does the reading, sorting and drafting; an automation connects it to the tools you already use. If Claude is not the right fit for your task, we will say so on the call.
Is our data safe?
We configure on your machine and on your account. We do not take copies of your files, we do not hold your credentials, and nothing is moved to us. In session one we scope Claude to one bounded working folder rather than the whole disk, and we set read and write rights per connector. Reads are the low-risk half; writes are the half that changes things and gets treated accordingly.
Anthropic documents the safety model directly, and it is worth reading before you buy anything, from us or from anyone: using Claude Cowork safely.
Where does my data go?
It stays where it is. For a set-up, we work on your machine and in your own Claude account: we take no copies of your files and nothing is moved to us. Claude is Anthropic’s service, so what you send to it is processed under the terms of the plan you buy from them. For an automation, where the flow runs and what it can reach are agreed in writing before we build it.
Do I own what you build?
Yes. A set-up lives in your own Claude account: the skill, the live page and the instructions are yours, and you can edit them. The written guide is yours to keep. For automations and websites, the handover is part of the scope we agree in writing before we start.
Does the AI Act require a certificate?
The AI Act asks you to take reasonable steps so your staff understand the tools they use. Our sessions and written guide are a documented step. No certificate is required.
Nobody here is technical.
Nothing in either session requires code. A skill is a page of plain instructions; a working folder is a folder. The most technical thing you will do is decide which tools Claude may write to, which is a judgement about your firm rather than about software.
Two hours, twice. Is that really enough?
It is enough to get one workflow genuinely running and to teach the method that produces the next one. It is not enough to automate a department.
Do we need a particular Claude plan?
Cowork runs on the paid plans (Pro, Max, Team and Enterprise) and you need the desktop app for anything that touches local files. On Team and Enterprise an administrator may need to enable it first. We will tell you which plan fits before you buy one, on the introductory call, and we take no commission either way.
Current details are on Anthropic’s own page.
What happens if it does not stick after you leave?
That is the usual outcome of AI training, and it is why the format is what it is. Two things fight it. The first is that everything is built on your own live work rather than on an example, so what exists at the end of session two is already in use. The second is the written guide: the sessions are not the deliverable on their own, the document that lets you re-run them is.
Can you sign an NDA?
Yes, before the first session and at no cost. Send yours, or ask and we will send a short mutual one.