Head to head
ActivityWatch vs Dayflow: the honest comparison
They are the only two time trackers that are free, open source, and local-first. They still answer different questions: ActivityWatch tells you the numbers, Dayflow tells you the story.
If privacy brought you here, good news: you cannot really lose. ActivityWatch and Dayflow are both open source, both store everything on your own machine, and both cost $0. The difference is what they capture and what they give back.
The core difference in one paragraph
ActivityWatch logs metadata: which app and window title was active, second by second, as queryable raw data. Dayflow captures context: it records your screen at one frame every 10 seconds and has AI write a plain-language timeline of what you actually did - "researching flight prices," not "Chrome, 47 minutes."
Side by side
| ActivityWatch | Dayflow | |
|---|---|---|
| Price | $0 | $0 (optional Pro, $16/mo annual) |
| License | Open source (MPL-2.0) | Open source (MIT) |
| Platforms | Mac, Windows, Linux (Android too) | Mac (14+); Windows waitlist |
| What it captures | Active app, window title, browser tab (with extension), AFK status | Screen at one frame every 10 seconds |
| What you get back | Raw activity buckets, charts, queryable local API | AI-written timeline with summaries every ~15 minutes |
| AI | None built in | Local (Ollama, LM Studio) or bring-your-own cloud key, or Pro |
| Works fully offline | Yes | Yes, with a local model |
| Footprint | Light | ~100MB RAM, under 1% CPU |
| Best for | Raw data, cross-platform, tinkering | A readable record of your workday |
Where ActivityWatch is better
- Cross-platform. Mac, Windows, Linux, even Android. Dayflow is Mac-only today.
- Raw data and extensibility. Everything lands in queryable local buckets with a REST API, watchers, and community extensions. If you want to build your own dashboards, this is the platform.
- A decade of maturity. It has been the reference open-source tracker since 2016, with docs, forums, and a large community.
- Minimal capture surface. It never records your screen, only app and window metadata - some people prefer that boundary on principle.
Where Dayflow is better
- You can read the output. The end of a Dayflow day is a timeline in sentences and a daily summary of where your time actually went. The end of an ActivityWatch day is a bar chart of window titles you now have to interpret.
- Context, not labels. Window metadata cannot tell "editing the client proposal" from "reformatting your resume" in the same app. Screen-level AI can, and does.
- Zero-config insight. No categories to define, no rules to write, no dashboards to assemble. Install it and read your day.
- Polish. A native SwiftUI app designed as a product, not a toolkit.
The honest trade-offs
Dayflow's approach costs more disk (it stores compressed screen recordings locally) and, if you choose a cloud model for analysis, frames are sent to that provider - the fully-private path is a local model, which uses your GPU. ActivityWatch's approach costs your attention: the data is only as useful as the time you spend categorizing and interpreting it. Neither app blocks distractions, tracks billable rates, or manages teams - for that, see our full Mac time-tracker roundup.
The verdict
Choose ActivityWatch if you want raw, cross-platform activity data you control completely and enjoy building your own views on top. Choose Dayflow if you want the same privacy guarantees but the computer to do the interpreting - an automatic work journal instead of a dataset. Running both is legitimate too; they measure different layers of the same day.
Frequently asked questions
Are ActivityWatch and Dayflow both really free?
Yes. ActivityWatch is open source under MPL-2.0 and free on Mac, Windows, and Linux. Dayflow is open source under MIT and free on Mac; you can optionally pay for Dayflow Pro ($16/month billed annually) which handles the AI for you so you do not need to set up a local model or bring an API key. Neither app requires an account.
Which is more private, ActivityWatch or Dayflow?
Both are genuinely local-first and open source, which puts them ahead of every commercial tracker. ActivityWatch stores app and window-title logs on your machine. Dayflow stores its screen recordings (one frame every 10 seconds) and their AI summaries on your Mac; with a local model (Ollama or LM Studio) nothing ever leaves the device, while using a cloud model (Gemini, ChatGPT, Claude) sends frames to that provider for analysis - your choice, clearly disclosed.
Can I run ActivityWatch and Dayflow at the same time?
Yes, they do not conflict. ActivityWatch logs app and window metadata; Dayflow records the screen and writes summaries. Some people run both: ActivityWatch for precise per-app stats and long-term raw data, Dayflow for the readable daily story.
Does Dayflow work on Windows or Linux like ActivityWatch?
Not yet. Dayflow is a native macOS app (macOS 14 or later) with a Windows waitlist. ActivityWatch runs on Mac, Windows, and Linux today, which makes it the default choice if you need cross-platform tracking now.
Read your day, don't decode it
Dayflow is $0, open source, and everything stays on your Mac.
Download for MacOS macOS 14+ · MIT licensed · no account required