The log
Every hunt, graded honestly
Flint reads a business from a collision, then a separate grader decides whether it fuses, whether it’s operable, and whether anyone has already taken it. Most readings come back tame. The ones that survive are published here — with the count of the ones that didn’t.
Counted since 4 August 2026. Every reading, including the ones that went nowhere.
The ideas
How these are graded
Every idea here came out of a collision — a resource thrown against a business move, or a resource thrown at a pain. The engine reads a business from that collision. It does not decide whether the business is any good.
That job belongs to a separate grader that never sees the engine’s reasoning, only its output. It runs three gates in order. Fuses: are the two halves one instrument, or two things sold side by side? Operable: could someone actually run this? Untaken: is anyone in this industry already doing it? The verdict is the first gate that fails.
An idea that clears all three is a flicker. One that clears the first two but is already being done is tame — sound, but the surprise is used up. One that fails to fuse, or that nobody bleeds for, is a dud.
The counts above are the whole record, not the highlights. If the grader is wrong about any idea here — if you know a company already doing it — write in, and the correction gets published alongside it.
Run it on something only you know about.
Your asset, your market, a pain you keep running into. Same grader, same three gates, no flattery.
Hunt your own idea — $1