> ## Documentation Index > Fetch the complete documentation index at: https://meetdonna.xyz/docs/llms.txt > Use this file to discover all available pages before exploring further. # Memory: what Donna keeps from your work > What carries forward when you finish working, the full list of what Donna stores, and the controls you have over it. Memory is the context Donna keeps from your work, so the next related task starts with what you already did instead of from a blank page. When you draft, list, or write in Donna's work view, the work is saved with its Outcome. When you finish, Donna distills what mattered and uses it when it builds later Actions toward the same Outcome. ## Work stays in the app Text-shaped work runs inside Donna's work view: you draft, list, and write there instead of opening a separate doc, switching away, and coming back to tick a box. Staying in one place removes a second cold start, which helps when switching tasks is the hard part. The work view itself (the editor, the timer, Donna's inline moves) is covered in [Working an Action](/docs/daily/working-an-action). This page is about what happens to the work afterward. ## What becomes memory Two kinds of things persist. The work itself is saved with its [Outcome](/docs/concepts/outcomes): the concrete details you worked out (the three companies you listed, the recruiter's name, the angle you settled on) are still there the next time a related Action comes up. Donna also picks up your patterns as you work with it: when your energy is sharp, what kind of framing gets you started, what tends to stall. There is no profile to fill out; both kinds of memory come from working with Donna. ## A concrete example Say one Action is "list three companies worth cold-emailing." You write them in the work view and move on. A week later you start a related Action, "draft an intro to one of them." Donna already has the three companies. You do not type them again, and the draft it helps you write uses the names you chose rather than placeholders. ## Everything Donna keeps, listed This is the full inventory: | What | What it contains | Where it comes from | | ----------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------- | | Your work | Drafts and notes in the work view, saved with their Outcome, and documents you drop in | You write it | | Your conversations | The transcript of what you and Donna say, whether you typed it or dictated it, plus a short summary that carries between sittings. Dictation is kept as its text; the audio is never stored | You say it | | Your structure | Outcomes, milestones, Cycles, Areas, Habits, and captures, in your own words | You and Donna build it together | | An inferred profile | Basics: name, role, timezone, life context. Your ADHD profile as you have described it: diagnosis status, medication timing, challenges, strengths. Patterns: energy, what tends to motivate you, typical capacity, avoidance and flow triggers, sensitivity to rejection, how your time estimates tend to run. Preferences: directness, humor, message length, words to avoid. People you mention, and their context | Inferred from working with you, recompiled nightly | | Individual observations | Specific learnings with a confidence level, like "long unstructured tasks stall; a named first step works." New ones start as low-confidence candidates and only shape Donna's behavior once later evidence confirms them | Inferred when you finish working | | Behavioral signals | Skips, event history of started and finished work, and Donna's current read of your energy and load. Used to change the approach, never shown to you as a score | Computed as you use Donna | | Settings | Voice preset, actions per day, work hours, ping times, what Donna calls you, and the work-surface sharing toggle | You set them | Two things the table does not spell out: * **Retiring, not erasing.** When a belief turns out wrong, Donna retires it rather than deleting the record, and the event history is append-only. * **The work-surface toggle controls reading, not storage.** Turning off "share your work surface" stops Donna reading the work view while you work. Notes you have typed are still saved with their Outcome. ## What you control Your writing is saved with its Outcome, and you can share a note as a link when you choose to (see [Sharing your work](/docs/daily/sharing-your-work)). The "share your work surface" toggle in settings is the off switch for that reading: Donna still knows you are working, just not what is on the page. If Donna has something wrong in a conversation, say so and it works from your correction for the rest of that conversation. Durable beliefs shift a different way: they revise as later evidence confirms or contradicts them, the retiring described above. You can delete your account, and with it everything Donna has learned, anytime from settings. More on both in [Your data and privacy](/docs/reference/your-data). The work view itself: the editor, the timer, and Donna's inline moves. How Donna reads the day you are having and adjusts what it asks.