DoneThat

DoneThat vs Keep notes

Use personal notes for memory; searchable but relies on discipline.

By Don, DoneThat's AI coach · verified

AI Memory: side-by-side

Tools to archive your digital life and make it available for later use by eg AIs.

DoneThat vs Keep notes comparison
ToolInput dataPlatform supportData locationAPI
Privacy-first work tracker and AI coach
Best for: Calendar-like timeline of your work.
Screen, audio, appsmacOS, Windows, LinuxCloudYes
Keep notes
Use personal notes for memory; searchable but relies on discipline.
Best for: Small knowledge bases with deliberate curation.
Manual records-N/A-N/A-N/A

Keep notes is best for

Small knowledge bases with deliberate curation.

DoneThat is best for

Calendar-like timeline of your work.

What Keep notes does well

  • You choose what is worth keeping, so the record is already filtered to what mattered to you.
  • Writing is thinking — the act of taking a note processes the idea, which capture never does.
  • Notes hold intent, reasoning and conclusions, none of which appear on a screen.
  • Portable, plain, permanent, and owned by you in whatever tool you already use.
  • No capture of any kind: nothing runs in the background, nothing is processed.

Where Keep notes falls short

  • Only records what you stopped to write, which is a tiny fraction of what you did.
  • The moments you most need later are usually the ones you were too busy to note.
  • No timeline: notes rarely say how long anything took or what surrounded it.
  • Requires sustained discipline, and the habit breaks precisely during busy periods.
  • Retrieval depends on how well you titled and filed things months ago.

See the full ai memory comparison or visit Keep notes.

The short answer

Notes and automatic capture record different things, and confusing them is why people bounce off both.

Notes hold what you decided — the reasoning, the conclusion, the thing you want to remember on purpose. Capture holds what happened — what you worked on, for how long, in what order, including all the parts you would never have written down. Neither one contains the other, and the useful setup is usually both.

What notes are good at

They hold intent. Why you chose this approach, what you ruled out, what you concluded. None of that is visible in a screen recording, an app name or a window title. If you do not write it down, it is genuinely gone, and no amount of capture recovers it.

Writing is thinking. The act of taking a note processes the idea. Capture is passive by design and does none of that work for you.

They are already filtered. A note exists because you decided it mattered. That is a signal automatic systems have to infer, badly.

They ask nothing. No background process, no capture, no vendor. If minimising what any software sees is the priority, a text file is unbeatable.

Where notes fall down as a memory system

One structural problem: notes only exist where you stopped.

Stopping is expensive. It happens when work is calm and you have slack, and it does not happen when you are deep in something, firefighting, or context-switching every ten minutes. So the coverage of your notes is inversely correlated with how eventful the period was — the weeks you most want to reconstruct later are the weeks you wrote least.

Three consequences follow:

  • No timeline. Notes rarely record how long anything took, what preceded it, or what interrupted it. "Fixed the deploy issue" could be twenty minutes or two days.
  • Silent gaps. Time you did not note does not appear as a gap. It appears as nothing, so the record reads as though those hours never happened.
  • Retrieval decays. Finding something in six months depends on how well you titled and filed it at a moment when you were busy and did not care.

What capture adds

DoneThat reconstructs the day automatically, so the record covers the whole of it — including the days you wrote nothing. That gives your notes the substrate they assume but never state: what surrounded the decision, how long the work took, what you were actually doing in the two hours before you concluded anything.

On the trade, since notes capture nothing at all: screenshots are never stored. They are processed in real time and immediately deleted, and only the derived structured activity persists. Excluded apps are redacted on your machine before anything is sent. With your own Gemini or OpenAI-compatible model, raw activity never reaches our servers. The desktop app is open source.

Both, feeding one assistant

The strongest version of this setup is not choosing.

Keep writing notes for reasoning and conclusions. Let capture handle the timeline. Then point an AI assistant at both: DoneThat exposes captured work history through an API and an MCP server, so an assistant can reason over what you did as well as what you wrote — instead of relying on you to remember the context and retype it into a prompt.

That is the actual argument for automatic capture in a notes-first workflow. Not that notes are insufficient, but that they were never meant to be a record of your time, and using them as one quietly loses most of it.

Where notes are the better choice

  • You need reasoning preserved, not activity. Only writing does that.
  • Your work is mostly thinking, and the screen is incidental to it.
  • You want zero capture. A text file is the most private tool in this comparison.
  • You already have a habit that works. A functioning system beats a theoretically better one.
  • You need the record to outlive any vendor. Plain files do.

Where DoneThat is the better choice

  • You want to know where the time went. Notes almost never answer this.
  • Your note-taking lapses when you get busy. That is universal, and capture does not lapse.
  • You want the days you wrote nothing to still be recorded.
  • You want AI tools to have your work history as long-term memory through MCP and an API.
  • You want coaching. Don nudges you onto goals, blocks calendar time, flags low-value work while you can still fix it, checks your mood, reminds you of meetings, suggests breaks. Notes have no opinion.
  • You want to share progress without writing an update. Summaries can go to followers, a team, Slack or Discord — all opt-in, with visibility set per summary.

How we compared these

There is no vendor here: "keep notes" is a practice, not a product, so this compares approaches rather than feature lists. Claims about DoneThat are first-hand — we build it.

We are not arguing you should stop taking notes; that would be a bad recommendation and an obvious one to see through. The narrower and more defensible claim is that notes and captured activity are different records, and most people are asking their notes to be both.

Questions people ask about Keep notes

Are notes enough as a memory system?
For decisions and reasoning, yes — those only exist if you write them down. For what you actually did with your time, no. Notes record the fraction of your work you stopped to describe, and the busiest periods, when the record matters most, are exactly when the habit lapses.
What does automatic capture add to a notes habit?
The timeline underneath the notes. Capture answers what you worked on, for how long, and in what order — the context your notes assume but never state. It also covers the days you wrote nothing, which is usually when something went wrong.
Should I stop taking notes if I use DoneThat?
No. They record different things and neither substitutes for the other. Capture cannot see why you chose an approach or what you concluded; only you can write that. Keep the notes, and let capture handle the record of what happened around them.
Can an AI assistant use both?
Yes, and that is the strongest version of this setup. DoneThat exposes captured work history through an API and an MCP server, so an assistant can combine your written notes with an actual record of the work — rather than relying on you to remember and retype the context.

Comparison last verified against Keep notes's public materials on . DoneThat makes one of the products compared here; competitor claims are drawn from public sources and never from private data.

Ready to try DoneThat?

Fully automated AI time tracking and coaching. 14-day free trial.