Hey Product Hunt 👋 The idea came from a frustrating pattern I kept seeing — consultants and freelancers who were genuinely good at their work but chronically bad at logging their hours. Not because they were lazy, but because stopping to hit a timer every time you switch tasks is just not how focused work actually happens. By end of day, half the billable time was gone from memory. The obvious fix is automation. But every automated tool I looked at had the same problem: they sent your screen activity to the cloud for processing. For anyone working with confidential client data — lawyers, accountants, developers under NDA — that's a non-starter. You can't guarantee your client's data stayed on your machine if it was uploaded to a vendor's server, even briefly. So I started asking a different question: what if the AI ran locally? The first version was rough — slow inference, clunky classification, a lot of manual cleanup still required. What changed everything was the new generation of Apple Silicon chips. Suddenly a quantized 8B model could run fast enough on a MacBook Air to be genuinely useful, not just a demo. That's when it clicked that this could actually be a product. The hardest part wasn't the AI — it was making it invisible. A background process that takes screenshots and runs inference can absolutely destroy your battery and make your fans spin constantly. Getting that to run quietly enough that you forget it's there took most of the engineering time. Would love to hear from anyone who's tried other automatic time trackers and hit the privacy wall — curious what workarounds you've been using.

TimesheetMe

Timesheets that write themselves. Privately, on your device

Timesheets that write themselves. Privately, on your device