Catch the keyword.
Read the window.
Up to five newborn search opportunities every morning, each with a window state that tells you to build, hurry, or skip.
Full refund before the first email.
Everything you need to decide, on one card.
Every pick links to a full card — and the window keeps updating after we send it.
API-only release with no official web UI. Tool-shaped queries ("online", "free", "4k") appeared within hours; positions 7–10 are still forums and thin pages.
Your feed shows you the launch. We find the keyword it left behind.
Long-tail discovery
Small-vendor changelogs, fresh GitHub repos, Wikipedia spikes — not the headlines everyone saw.
Daily window check
A standardized SERP snapshot every day. Only real tools count, and only after two sightings in a row.
Sealed every day
A salted hash of the picks goes public the same day. Proof we never back-date a call.
Every call stays on the record. Misses too.
Two hit rates — did demand arrive, and was there still room to rank — each against a baseline of every candidate we saw.
Open the full record →Only 100 founding seats.
Everyone gets the same keywords. Fewer seats keep the windows open longer.
- Daily picks, up to 5
- Every opportunity card
- Window tracking on each pick
- Everything in Pro
- A say in the scoring rules
- Full refund before the first email
No live track record yet — you're joining before it exists.
If they're this good, why not build them yourself?
We do — one a week, by a public rule, with rankings published. One person can't build them all.
What happens on a slow day?
Zero picks, plus what we rejected and why. Five is a ceiling, not a quota.
How do you know a keyword is new?
Every candidate gets a birth record the first time any source mentions it, and that timestamp is never rewritten.
Can I check you are not back-dating picks?
Yes. Each issue is sealed with a salted SHA-256 the day it goes out. The salt and payload are revealed later on the Verify page so anyone can recompute it.