DoneThat vs Daily

The short answer

Daily asks you what you are doing. DoneThat works it out.

That is the entire difference, and which one is better depends on a preference you probably already have: would you rather answer a small question a few times a day and get labels that are exactly right, or answer nothing and get complete coverage with the labels inferred?

The prompt model, fairly

Daily is unusually honest about what it is. There is no claim that a background agent will divine your intent. It asks, you answer, and the resulting record is correct because a person who knew the answer supplied it.

That has real advantages worth stating clearly:

The categorisation is right. No AI deciding that main.ts — project-b was project A. You said what it was.

The privacy surface is tiny. No screen capture, no cloud analysis of your activity, very little to worry about. For anyone who rules out capture-based tools on principle, Daily is in a different category of ask entirely.

Silent mode exists. You can apply an activity for a fixed duration so the prompt does not fire during deep work.

It is well liked. 4.6 out of 5 across more than 2,400 reviews is a strong signal for a small tool, and it is available through Setapp if you already subscribe.

Where the prompt model strains

It interrupts. That is the mechanism, not a bug — but the prompt does not know you are mid-thought, and the interruption cost of a small question during deep work is famously larger than the question.

It samples rather than records. You answer at intervals; the time between answers is inferred from your reply. If your afternoon fragmented across six things, a prompt-based record smooths it into whatever you said when asked.

Silent mode trades one problem for another. Setting an activity in advance to avoid the prompt means going back to telling the tool what you will be doing — which is a timer with extra steps.

It is macOS only, with iOS as a companion for viewing and editing. No Windows, no Linux.

What you get out of it is a timesheet dashboard you can view by day, week or month and edit after the fact, with activity scheduling and automatic overtime calculation. That is a reporting view rather than a billing system — enough to see and correct your own hours, not a client invoicing pipeline.

What automatic reconstruction changes

DoneThat has no prompt and no timer. Capture runs, the day is reconstructed, and you review and correct rather than answer questions. Coverage is complete, including the fragmented parts that a sampled record smooths over.

The trade is the ask, and it is a bigger one than Daily's: screenshots are never stored — processed in real time, deleted immediately, 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.

If that trade is unacceptable to you, Daily is a sensible place to land, and this comparison ends here in its favour.

Integrations

For a small tool Daily does well here: a web API, Zapier, and AppleScript automation, which is a genuinely useful escape hatch on macOS.

DoneThat offers CSV, an API, Zapier, n8n, and an MCP server at mcp.donethat.ai. The MCP piece is the substantive difference — it makes captured work history queryable by an AI assistant as long-term memory, which no amount of Zapier plumbing replicates.

Platform support

Where Daily is the better choice

• You are on macOS and want something small. Daily is focused and does not sprawl.
• You rule out screen capture. The privacy ask here is much smaller than ours.
• You want labels you control. Your answers, your categories, no inference.
• You already use Setapp. It may effectively be included.
• You want an iOS companion for reviewing timesheets away from the desk.

Where DoneThat is the better choice

• You are on Windows or Linux. Daily does not run there.
• Interruptions are expensive for you. Nothing asks you anything.
• Your day fragments. Sampled records smooth exactly the detail worth seeing.
• 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.
• You want AI tools to have your work history through MCP and an API.
• You want peer accountability. Follow others, share summaries, Slack and Discord, accountability groups — all opt-in.

How we compared these

Everything stated about Daily comes from its public product pages, read on 11 August 2026 and linked below. Daily does not publish pricing on its home page, so no figures are quoted here rather than guessed ones.

Claims about DoneThat are first-hand: we build it. Daily's privacy ask is genuinely smaller than ours and its categorisation is more reliable than any inference — both said plainly above.

Questions people ask about Daily

How does Daily track time?

It periodically asks what you are doing and you answer. There is no timer to start and no automatic capture of your screen — the record is built from your replies. You can also apply an activity silently for a set duration when you do not want to be interrupted.

Does Daily work on Windows or Linux?

No. Daily is a macOS app with an iOS companion for viewing and editing timesheets. DoneThat runs on Windows, macOS and Linux, but has no mobile app.

Is being prompted better than automatic tracking?

It is more accurate about categories and less accurate about coverage. You know exactly what you were doing, so the label is right — but the prompt only samples your day, and anything between prompts is inferred. Automatic capture inverts that: complete coverage, with the categorisation inferred instead.

Can I use Daily without being interrupted?

Partly. Silent mode applies an activity for a fixed duration so the prompt does not fire, which helps during deep work. It does mean going back to telling the tool what you will be doing, rather than it noticing.