Loom

Loom is designed to make work tracking light: a single keystroke captures what is being worked on without pulling attention away from the work itself, and the data stays on the user’s machine because a personal work log can contain unfinished thinking, client context, and small private details that should not have to leave one's device.

For me, building Loom also meant deciding what kind of information was worth keeping: a complete record of every task or minute quickly would be too heavy, while a vague summary would not be so useful later. Loom sits between those extremes, preserving enough context to be useful without becoming another system to maintain.

The aim is for Loom to turn a work log into a weekly account of the things that are easy to forget. I already had a good tracking system for my weekly tracking; but it was not as good at logging the less important but as consequential items such as a bug fixed, methodological choice made, a tool, or even an idea that I would like to go back to later. Loom is intended to preserve those details before they fade.

The local AI layer was designed to produce the weekly account without sending the log elsewhere. I designed the onboarding as a conversation with an LLM rather than through a settings file.

The project is available on GitHub.