FAQ
The honest answers.
What ran is, what Margarita does today, and what happens to your data — plainly.
How is this different from Notion AI or Glean?
Citations at the claim level. Every answer opens the exact passage it came from — and when your sources disagree, Margarita shows both instead of picking one.
What can it actually do today?
Ask, check, act. Margarita answers across Notion, Slack, Gmail, Drive and Outlook with citations, and takes on real multi-step work — nothing sends without you.
How long does setup take?
Minutes. You link your accounts over OAuth and Margarita reads them on her own — nothing to migrate, nothing to re-upload.
Does ran train on my data?
No. Your documents never train a model — ours or anyone else's — and nothing leaves your accounts without your yes.
What can Margarita see?
Only what your own logins can see. Your memory is built from your accounts and belongs to you alone — access is inherited, never widened.
Can Margarita act, or just answer?
Both. She drafts the document, reconciles the figures, writes the reply — and every send waits for your yes.
Can I use it with the Claude or ChatGPT I already pay for?
Yes — over MCP, in early access. Your memory plugs into the assistant you already use, so your Claude or ChatGPT answers from your live, cited work.
What does it cost?
Nothing while it's early. Free during early access, no credit card — the details live on the pricing page.
When does early access open?
It's rolling now. Join the list and you'll hear from Gabriel within a week.