Use case
It's 6pm. You were busy all day, you barely stopped, and you can't account for any of it. Here is an app that answers the question for you - automatically, every day.
The question is almost never rhetorical. The day went somewhere: the morning that dissolved into fourteen small context switches, the "quick look" that became forty real minutes of research, the hour helping a teammate that never made it into memory. But human memory keeps highlights, not hours. By evening you have three fragments and a vague guilt, and no way to check either.
Dayflow gives you a truthful record of where your time went with zero manual timers or tags. It records your screen and has AI write a readable summary of your day. Full recordings stay on your Mac, and Local AI keeps analysis on-device.
Dayflow records your Mac's screen at one frame every 10 seconds and, every few minutes, has AI summarize what you were actually doing into a timeline written in plain sentences:
Note the third entry. An honest answer to "where did my day go" has to include the detours, and a tracker you might judge yourself in front of has to be private (more on that below). The point is not surveillance of yourself - it is that you cannot fix a day you cannot see, and most days are kinder in the record than in the fog.
Manual time trackers assume the discipline the question proves you don't have that day. Starting and stopping timers is a prospective-memory task; forgetting it is exactly how the hours got lost in the first place. This is doubly true for anyone with ADHD-style time blindness, where the tool that demands time awareness to operate cannot be the tool that provides it.
App-usage charts (Screen Time, RescueTime) are automatic but answer a different question: which apps were open. "Chrome: 5 hours 12 minutes" is not where your day went - it is where your windows were. The useful unit is the task, in words: what were you doing, when, for how long. That's what a work journal gives you that a chart can't. (Here's the full Screen Time comparison.)
A record of your entire working day is intimate. Full recordings and the local timeline stay on your Mac. Local AI through Ollama or LM Studio keeps screenshot analysis on-device; BYO cloud providers and Dayflow Pro can receive selected screenshots for analysis. The app is open source under the MIT license with 6,900+ stars on GitHub, so the privacy behavior is visible in the code. The recording pipeline runs on ~100MB of RAM and under 1% CPU.
Install Dayflow in the morning, work a normal day, and open the timeline at 6pm. You will get the answer to "where did my day go" - a real one, with times. Most people find two surprises in the first week: their actual focus window (usually not when they thought), and their actual top time sink (almost never the one they would have named).
Yes. Dayflow is a free, open-source Mac app that records your screen at one frame every 10 seconds and uses AI to write a readable timeline and daily summary, with no timers or tags. Full recordings stay on your Mac.
Dayflow does exactly this on the Mac: it observes your screen, understands what you were working on, and writes a daily summary you can read in a minute. It is free and open source, and Local AI keeps screenshot analysis on-device.
Screen Time and RescueTime count app usage and show charts. Dayflow writes a narrative: what you actually did, in sentences, with times. Charts tell you Chrome was open for 5 hours; Dayflow tells you what happened in those 5 hours.
Full recordings stay on your Mac. Local AI through Ollama or LM Studio keeps screenshot analysis on-device. If you choose a BYO cloud provider or Dayflow Pro, selected screenshots can be sent to that service for analysis. The open-source Mac client shows local storage and outbound network behavior; the privacy policy covers cloud retention.
Dayflow records about one frame every 10 seconds - not video - so it runs on roughly 100MB of RAM and under 1% CPU. Most people cannot tell it is running.
Dayflow is $0 and open source. Full recordings stay on your Mac, with Local AI available for on-device analysis.
Download for MacOS macOS 14+ · MIT licensed · no account required