The basics
How is this different from ChatGPT or the Claude app?
Chat apps answer questions and forget you between sessions. An operator is set up around you: it keeps permanent memory, holds real access to the accounts you choose, reaches out first (briefs, digests, alerts), and completes tasks โ sends the email, books the meeting โ instead of drafting suggestions. You keep using a chat app to talk to it; the difference is everything behind the chat.
Why does it need to know so much about me?
Because context is king. The single biggest lever on whether an AI answer is generic or genuinely useful isn't the model โ it's how much of your actual life it knows when it answers. A stateless chatbot starts from zero every session, so it gives you the average answer for anyone. Your operator starts every reply already knowing your calendar, your people, your open threads and your history with them โ so "should I take this meeting" gets a real answer, not a shrug. That's the trade you're making on purpose: feed it real context, get an answer worth reading. Feed it nothing, and it's just a chatbot with extra steps.
Do I need to install anything?
No. Your operator lives in the messenger you already use. Setup is about 10 minutes, and the first thing your operator does is interview you for 10 minutes to learn names, routines and boundaries.
What languages does it speak?
Swedish and English from day one. It mirrors whichever you use, and you can mix freely.
Can I change what it does later?
Yes, any day. There is one plan, so there is nothing to upgrade or switch: just tell your operator what you want help with. Cancel any time.
Privacy & your data
Where is my data actually stored?
In one private space that belongs to you: your operator's handbook, memory, files and logs, all as plain readable files. Your calendar stays your calendar and your inbox stays your inbox โ the operator holds credentials to act on them, not copies of their contents. Operated from Sweden under EU rules. Full detail:
data & storage architecture.
Is my data used to train AI models?
No. Conversations are processed under commercial API terms that exclude training. There is also no cross-customer anything โ no shared index, no "users like you".
What does the AI provider see?
Each message plus whatever files your operator opened for that task travel to the model for that turn, transiently, then fall out of context. The durable copy of your data only ever lives in your space.
Can other customers' operators see my data?
No. Every operator runs in its own isolated space and physically starts inside your folder โ it cannot read outside it. Isolation is enforced by the architecture, not by a promise.
Can I take my data and leave?
Always. Because everything is plain files, export is literally an archive of your space โ readable the moment you open it. After cancellation your export link stays live for 30 days, then the space is destroyed.
Can the humans who run the service read my conversations?
The audit log and chat mirror exist in your space so that you can hold your operator accountable. Operationally we only look inside a space with your explicit permission, for support you've asked for โ and we take nothing out of it. If that trust boundary matters to you (it should), ask us about it before you buy.
Control & safety
Can it send emails or book things without asking me?
Not by default. Everything outbound โ emails, invites, posts โ is previewed to you first. Once you trust a specific ability, you can loosen the preview per ability ("just send my lunch confirmations"). Tightening it back takes one message.
What if it makes a mistake?
It will, occasionally โ it's capable, not infallible. That's why previews are on by default, why every action is logged, and why it insists you review anything high-stakes. Design principle: mistakes should be small, visible and reversible.
Can I pause or stop it instantly?
Yes. "Pause my operator" stops everything on the spot; resume the same way. There is also a hard kill switch on our side that disables the whole operator and locks its credentials until a human turns it back on.
What can't it do?
It won't impersonate you deceptively, break laws or platform rules, or give financial, legal or medical advice (it will tell you to see a professional). It also won't touch abilities you haven't enabled โ an operator without the mail scoop cannot read mail, structurally.
Billing
Is there a binding period?
No. Monthly subscription, cancel any month by chat or from your account page. Your operator handles its own offboarding, including your data export.
What happens to my operator's memory if I cancel?
You get the full export (plain files). The export link lives for 30 days after cancellation; then your space is destroyed. If you come back later, you can restore from your export and your operator picks up where it left off.
Why is it more expensive than a chatbot subscription?
Because it's a different product: a private always-on service provisioned for you alone โ isolated space, real integrations, scheduled work at night, previews and audit logging. You're not buying answers; you're staffing a role.