Dayflow

Neat idea keeping it free with a "bring your own AI model"

Cool! But is it safe to install this on my work laptop?

the screen data approach is what i'd want for passive journaling, no manual tagging. curious about the data model though - does it store actual screen captures locally or just extracted text/events? open source helps with trust but i'd want to know what actually persists on disk before installing something that watches my screen all day

The brag-doc pain is painfully real, "what did you ship this quarter" and the mind goes blank is a universal developer experience :)My question is about the boundary, because Dayflow doesn't see your work, it sees your screen. And my screen on a workday also has the 2am guitar tab I hunted down, a medical search, personal DMs, the apartment listings I shouldn't have been browsing at 3pm. For a brag doc, none of that should ever surface. So how does the separation actually work, an app/domain allowlist I define up front, AI classification of "work vs personal", a manual scrubbing pass? The honest record is only an asset if I trust it won't quietly file my private life into my performance review.But here's the twist I can't stop thinking about: that same personal stream is a latent memory index. "When did I find that guitar tab again?" is exactly the kind of question I'd kill to answer, and Dayflow technically saw the moment. I build in the private-memories space myself, so I'm genuinely curious, is the timeline queryable by activity type, or grouped beyond the workday view? Because the thing that's a privacy liability for the brag doc (it captures everything) is the exact thing that could make it a personal recall tool. Two products hiding in one capture stream.Local-first + MIT is the right foundation for either. Congrats on the launch! :)

The screen-capture approach to auto-journaling is smart — most brag-doc tools fail because the manual logging burden kills the habit. Curious how Dayflow handles sensitive content on screen (passwords, private chats) when it's continuously capturing?

The no-timer workflow is what makes this stand out. Work often happens in small scattered moments, so having Dayflow turn that into a clean record for standups and reviews feels practical.

A lot of the discussion is about whether the timeline is accurate. The part I'm more curious about is what happens when it actually gets used. Does the summary go straight into a review or brag doc, or is there always a review/edit step first? That handoff feels just as important as the capture itself. Congrats on the launch!

The brag doc angle is clever, but the thing I keep thinking about is how it tells real focus from a tab you left open, like if I'm heads down in my editor for two hours with a Jira ticket parked in the background, does Dayflow credit the editor work or the ticket that was just sitting there?

The evaporated-memory problem is real, but the part I'd want pinned down is how Dayflow captures the work without me logging anything by hand. Does it passively ingest from sources like git, Slack, or calendar, and since it's open source, do those integrations run locally/self-hosted or does the raw activity get sent to a server? And where does the accumulated work history actually live — a local DB I own, or your cloud?

It would be better to add an X handle field to the website.

This one is crazy - never thought about that, but it makes a lot of sense. Like you already said, "...and your mind goes blank."
It happens to me too. I can't even recall what I did one or two weeks ago and need to check my handwritten notes first before I can tell anyone anything.

Local-first plus run-any-model is the right call for screen data. One thing I'd push on from doing vision-over-screenshots myself: a frame tells you what's visible, not what you actually worked on. Our summaries would confidently claim 'worked on JIRA-1234 for two hours' when the ticket was just a tab left open while the person was heads-down in the editor. Does Dayflow weight by real keyboard or window-focus signal, or infer effort from frame content? That's the line between a believable brag doc and one your manager flags as inflated.

the framing is spot on. the people who get promoted are the ones who can articulate what they did, not necessarily the ones who did the most. i've lost track of genuinely important work just because i didn't write it down in the moment. the local-first approach matters a lot here too because screen data is about as personal as it gets. does the AI summarize at the end of the day automatically or do you review a timeline and pick what's worth keeping?

The tagline says Dayflow helps you get promoted, which is a very specific promise for an open source tool. Is the main workflow more about capturing accomplishments over time, turning work into status updates, or guiding people on what to focus on next? The AI Agents / AI Dictation Apps hints make me wonder how much is automated versus manually curated.

Remembering what you shipped months later is a real pain. The screen capture side is the only thing I’d be careful with: can Dayflow auto-exclude certain apps or websites before they’re analyzed, or is privacy mostly handled by pausing and deleting entries after?