Skip to main content
Tuck AI answers questions about the thoughts, todos, Awaits, and people already in Tuck. It works from the screen you’re on, and you can see exactly what it gets before you ask. Tuck AI is in the web app on Tucker and Founder. It’s on by default. To turn it off, open Settings → Tuck AI. The switch follows your account, so turning it off on one computer turns it off on the others too.

What Tuck AI is sent

Your unlocked device builds what Tuck AI sees. Before you ask, the panel shows a What’s leaving list with each item that will be sent. The first time you use Tuck AI, a note explains where it goes. When you send a question, those items go through Tuck’s server to Anthropic’s Claude models, which write the answer. Tuck AI doesn’t receive your vault key, and Tuck’s server doesn’t read your vault to answer. Tuck’s logs keep counts and costs, not your questions or answers. These stay on your device:
  • Email addresses, phone numbers, and links. Tuck removes them or swaps in placeholders such as [email].
  • Files and attachments.
  • Anything in a shared brain.
  • Items your device couldn’t decrypt.
Mail is sent only from a person’s card, when you turn on Include recent mail. See Summarize a person. To review what you’ve sent, open Settings → Tuck AI and check the Sent to AI log. It follows your account, and clearing it clears it on your other computers too.

Ask about the screen you’re on

Choose Ask Tuck AI, or press ⌘J. What’s sent depends on the screen:
  • A thought: the thought, its open todos and Awaits, and the people on it (name, company, title, and notes).
  • Today: what needs your attention and why, your schedule with titles and times only, todos due today, and Awaits to check on. Meeting locations, guests, descriptions, and links stay on your device.
  • Todos: the todos in view, in the page’s order, with their thought’s title, brain, people, and sub-items. Up to 50 are sent.
  • A brain: the brain and its nested brains, its open todos and active Awaits, and its 20 newest thoughts, pinned first. A shared brain sends nothing.
  • A person: see Summarize a person.
The panel starts with a few suggested questions. Under each answer, you can copy it, save it as a new thought, or, on a thought, add it as a comment labeled Tuck AI. A chat holds 10 questions. After that, choose New chat to start again. Each follow-up includes the chat so far, so Tuck AI remembers what you’ve talked about. Tuck AI sticks to what’s on screen and the work it points to. If you ask something unrelated, it suggests tucking the question as its own thought instead.

Search your whole vault

Turn on Search my whole vault in the chat panel when the answer might be somewhere else. Your device searches your thoughts, todos, and people, and adds up to 8 thoughts, 8 todos, and 4 people to what’s sent. They’re marked in What’s leaving, and a line under your question says what was searched and added. The search runs on your device. Mail isn’t searched, and matches in shared brains are counted but never sent.

Let Tuck AI add things for you

In a chat, you can ask Tuck AI to add a todo, an Await, a thought, or a tag. It answers with a card instead of text to copy. It only offers cards when you ask, or after you say yes to a suggestion. On a card you can:
  • Edit the title.
  • Set a date on a todo or Await.
  • Change a new thought’s text and pick its brain. Shared and archived brains aren’t offered.
Nothing is saved until you choose Add. Choose Dismiss to skip a card. On a thought’s page, a todo or Await you add links to that thought. Tuck checks each card again when you add it. If the thought it belongs to has moved into a shared brain or been archived, Tuck tells you and doesn’t add it. Tuck AI never creates a new person. Someone who isn’t in People is only named in the text.

Draft from a thought or todo

Draft appears on a thought, next to Ask Tuck AI, and on open todos that read like writing work, such as “email Sam” or “finish the blog post”.
  1. Choose Draft. The dialog shows What’s leaving, then the draft.
  2. Read it over.
  3. Choose Save as thought, Copy, or Try again.
Tuck AI is sent the todo or thought you’re drafting from, the thought, its other open todos, and the people on them. Email addresses, phone numbers, and links are replaced with [email], [phone], and [link], and the draft keeps those placeholders for you to fill in. Nothing is kept until you save. A saved draft goes into the same brain with the same people. It starts with a link back to its source, and the source thought links to the draft. Hashtags and @ mentions in a draft stay as text, so they don’t create tags or people.

Prep for a meeting

On Today, choose Prep on an upcoming meeting to open a brief. On a meeting’s thought, the brief shows above the text until the meeting is over. Prep isn’t offered for calendars set to For reference, or after a meeting ends. A brief has a short summary, a line for each attendee who’s in People, open todos and Awaits worth raising, and up to four things to bring up. It doesn’t change anything on its own. Tuck AI is sent:
  • The meeting’s title, time, guest names, and description, with links, addresses, and phone numbers removed. The location is never sent.
  • Your notes for this meeting and up to three earlier meetings in the same series, with their open todos and Awaits.
  • For each attendee in People, their open todos and Awaits and up to five recent thoughts they’re on.
No mail is sent. To keep a brief:
  • Save as comment adds it to the meeting’s thought as a Tuck AI comment.
  • Add to thought as quote adds it to the end of the thought’s text as a quote.
From Today, these save to the meeting’s thought, or start one if there isn’t one yet. Each brief can be saved once. Briefs are kept on your device, encrypted with your vault key, and don’t sync. A brief says when it’s out of date, and you can refresh it at most once an hour.

Summarize a person

On a person’s page, choose Summarize (it shows their name, such as Summarize Sam). The card sums up where things stand and links their open todos and Awaits. By default, it uses only thoughts, todos, Awaits, and the details already in Tuck. To include mail, turn on Include recent mail on that card. It starts off for every new card, and the card remembers your choice when you refresh. With mail off, What’s leaving shows how many recent emails stay on your device. With mail on, Tuck looks at the last 60 days: up to 10 emails with this person, then up to 6 from their company. Free email providers and your own domains don’t count as a company. For each email, Tuck AI is sent the sender and recipient names, the date, the subject, and up to 1,500 characters of the message, with email addresses and phone numbers removed. Addresses themselves are never sent. With mail on, the card can also suggest todos and Awaits. Choose Add to create one. Suggestions you add or dismiss don’t come back when you refresh. Calendar events aren’t sent. Pressing ⌘J on a person’s page opens chat with the same context, following the same mail choice. The card is kept on your device, encrypted with your vault key, and doesn’t sync. It shows Out of date when what it was made from changes. After a card is made, Refresh waits an hour and shows when it’s ready, such as Refresh at 5:06 AM.

Your monthly allowance

Tuck AI includes a monthly allowance, plus limits on how many questions you can ask in a minute and in a day. Tuck shows your use as a percentage.
  • From 80% used, the chat panel shows a warning.
  • At 100%, Tuck AI pauses until the first of next month and tells you when it’s back, such as It’s back on December 1.
  • Settings → Tuck AI shows a This month bar once you’ve used half, or while Tuck AI is paused.
If something looks wrong with an answer or the allowance, contact support and we’ll take a look.