Scout
Builds the automated daily search you run with an assistant, and refuses to call anything a job until it has been confirmed on the employer’s own site.
Three things, and it says how it did each one.
Ten numbered hard gates
Your ranked titles, industries, pay floor and distance limit go into the brief as numbered gates rather than a paragraph of preferences. An assistant that skips one has to say which one and why.
An evidence rule
A search results page is not proof a job exists. Every aggregator find gets checked against the employer’s own careers page before it counts, and a coverage count has to be written before any narrative.
A stop rule
Three empty runs and it stops rather than padding the list to look productive. The self check prints defects only, so you read what went wrong instead of a summary of what went right.
Three runs. One of them is lying to you.
A run is Valid at five of seven Priority 1 sources. Below that the report says so on its own line, because a short list from a blocked reader looks exactly like an empty market.
These three runs are made up, so nobody's real search is sitting on this page. The station itself works exactly like this and prints the gate behind every elimination. The live demo runs the whole app on its own, with nothing to fill in.
Everything it hands you.
- ✓A ranked shortlist that loads onto your board in one paste.
- ✓Your whole board rides inside the prompt with closed roles marked declined, so a dead requisition does not come back next week.
- ✓Nothing to export. The brief is the product.
It works with ChatGPT, Perplexity, Grok or Claude, which means it borrows the assistant subscription you probably already pay for instead of charging you for a second one.
Where this sits in the search.
Free, no account, and nothing you enter is uploaded.
It installs to your phone from the browser with no app store, and every station keeps running with no signal.