Studio in your own AI assistant
Your own AI assistant, with your business data at hand.
Do you work in ChatGPT, Claude, Mistral Vibe or Microsoft Copilot? Here’s how it works with your Studio workspace.
Ask what came in this week, or have it draft a proposal for a new screen. Without your data simply walking out the door.
- ChatGPT
- Claude
- Mistral Vibe
- Microsoft Copilot
What you use it for
Ask in the assistant you already use
Three examples. The answer comes from your own Studio workspace.
Just ask.
“How many pallets came in this week, per supplier?” in Teams via Copilot.
Sum it up for the meeting.
“Put this month’s totals next to last month’s” in Claude.
Capture an idea straight away.
“Draft a proposal for a returns screen” in ChatGPT: the proposal is then waiting in Studio for you to review.
How many pallets came in this week, per supplier?
148 pallets came in this week.
| Supplier A | 64 |
|---|---|
| Supplier B | 51 |
| Supplier C | 33 |
| Total | 148 |
From Studio: totals only
Draft a proposal for a returns screen.
Proposal ready in Studio
Returns screen
What it can do, and what it deliberately can’t
Limited on purpose: so your data stays yours.
Your own assistant can search your workspace, ask for totals and counts, and make a proposal. Nothing more, and that is a choice.
| Topic | Your own ChatGPT, Claude, Vibe or Copilot |
|---|---|
| Where it runs | At that provider, under your account there |
| Data | Totals and counts only |
| Personal data | Never |
| Building | Makes a proposal; you take it over |
| On or off | Off by default, the admin switches it on |
Why
- An external assistant runs at another company. We have no agreements with that company about what happens to your data.
- That is why we give an external assistant only what you can safely share, and let it change nothing by itself.
How we built it
Four choices, no jargon
The technical term is in brackets, for anyone looking for it.
An open standard.
Studio speaks the language all these assistants understand (MCP, the Model Context Protocol). So one connection works for all of them.
Sign in with your own Studio account.
The assistant gets no password or key from you: you sign in with us, give consent and choose the workspace. Access is short-lived and you can revoke it whenever you want.
Only what is safe.
Every question passes the same gate as the rest of Studio, which knows per field what is sensitive. Only totals and counts go out, no individual rows or personal data.
A proposal instead of a change.
If the assistant wants to build something, it becomes a proposal in Studio. You or a colleague review it in the builder and take it over, or not.
You stay in control
Off by default. You switch it on.
It is off by default. Only an admin of the workspace switches it on, and switches it off again with one click.
Every question from an assistant goes into the log: which assistant, what, when. Not the content.
In Studio you see which assistants are connected.
Questions
What people often ask
Why can’t I see individual rows through ChatGPT?
Because ChatGPT runs at another company, and we have no agreements with it about your data. That is why an external assistant only gets totals and counts. You look at individual rows in Studio itself.
Do OpenAI, Anthropic, Mistral or Microsoft read my data?
Only what the assistant asks for in a conversation, so totals and the names of apps and pages. That falls under your own account and terms with that company.
Can the assistant change my apps?
No, it can only make a proposal. The proposal waits in Studio, and you decide whether to take it over.
How do I switch it off?
An admin of the workspace switches it off in Studio with one click. After that, no connected assistant can ask for anything. You can also revoke your own assistant’s access yourself, whenever you want.
Curious what an assistant can do for your work?
Tell us what you want to ask your assistant. We show you how that works with Studio, and where the limits are.