Know what
matters.

Something arrives: a reel, a podcast, a newsletter, a link, a PDF. Bearing listens to it, reads it, and works out whether it has anything to do with what you are working on right now.

Whatever survives that ends up in the briefing. Whatever does not stays out, and you can look at it anyway whenever you want. No feed, no timeline, no unread counter. What you get is a text you read in a few minutes.

I listened to podcasts I could not say a single thing about the next day. They were good. Nothing in them pointed at anything I was actually working on.

The tools for this all solve the same half. Read-later apps store things, knowledge systems store them better. That something is not worth opening is what none of them will tell you.

That sentence is the entire product. An ad-funded product could never say it.

I picture context management as a triangle. Organisation management: what exists and where it sits. Information management: what arrives from outside, and whether it is worth your time. Knowledge management: what you already know.

The model is mine and it is in no textbook. It is the shortest way I have found to explain where Bearing works and where it stops.

Most tools sit in the third corner. What belongs in there is something you sort out beforehand, mostly in passing. I am building at the first corner too, it is called quay. That one is not far along.

A funding round in your market matters on Tuesday and is noise on Friday, depending on what you are deciding this week. Bearing keeps several profiles for that: product, market, team, private. The same article ranks differently in each.

That is also why there is no shared feed here. There is nothing for anything to be popular in.

If the same three perspectives keep reaching you, you will not notice. That is what a blindspot is: it does not announce itself.

The Bias Network puts the sources you actually read on an axis and shows how much of your time each one got. Where nothing reaches you, the gap turns up in the briefing.

There is an algorithm in Bearing. It learns which subjects concern you, never which position you enjoy. Both sides of a subject stay visible, whatever you click.

Fair question. A wrapper passes your question to a model and hands back the answer. Next session, empty desk.

Bearing depends on remembering. What you read, what you dismissed, what you already knew, where something caught up with you later. Those signals become the context the next signal gets judged against, and that context is worth more after a year than it is on day one.

There are models inside, several of them. They are components, the way a database is a component. The part that makes Bearing Bearing is what stays behind between the calls.

Bearing is built by Felix Winkel at SW labs. Product, design and code sit with one person. Decisions are made quickly, and you talk to the person who builds it.

Nothing has been released yet. There is a waitlist, and the people on it hear about prices and dates before anyone else.

I have large plans for this and a realistic sense of how long they take. Give me two years.