Every competitor tool tells you something happened. That's the easy half. Track the handful of companies you really race against and see which of their moves the market cared about — and which ones went out to silence.
A funding round announced into a skeptical room and a quiet feature ship developers adopted immediately look identical in a changelog feed. They are not remotely the same event. Sorting by recency is how you over-react to noise and miss the move that counts.

Add the companies you actually compete with. The engine watches their releases, pricing, hiring and public repos on a schedule, diffs every observation against the last one, and scores what changed. Nothing is inferred — a move exists because two snapshots disagree.

When a competitor ships something genuinely significant and the market says nothing, that gap is the single most actionable signal in the product. It means the need wasn't felt, the framing missed, or the lane is still wide open. You only ever see it if reception is measured — a feed reports that launch as a win.

The companies you actually race against — not a 400-row market map.
The reaction each launch earned, coupled to that specific launch.
What shipped, what landed, what it means — one forwardable page.
Pushed only when a move is genuinely worth interrupting you.
Most of these are things a competitor-intelligence tool could ship tomorrow and charge for. Each one would cost the thing that makes the output worth reading.
We don't tell you to build it or kill it. You get evidence and the call stays yours.
The number comes from the diff and the measured reaction. Same inputs, same number, every time.
Every line carries its source URL and timestamp. If we can't cite it, we don't show it.
Hard constraint. We never hold your credentials for anything.
Nobody is told you're watching them. Read-only by architecture, not by policy.
Your tracked set is yours. It is not a dataset we resell.