Use cases
The common thread: a company has already decided something, and written it down in public, and nobody has gone and read it. At 6 credits a scan, reading it is cheap.
Sales and go-to-market
A company that just opened its first data engineer role has funded a warehouse project. A first solutions engineer means they started selling upmarket. You are not guessing at intent — you are reading the hire they are already paying a recruiter to fill.
How: Scan your target accounts on a cadence. The delta on the second scan is the trigger: opened roles, and any function that appeared for the first time.
newFunctions[] is the loudest field. A capability they did not have is a capability they are about to buy tools for.
Recruiting and talent
Open reqs are the only public, dated, unspun statement a company makes about where it is growing. Watch a set of companies and you see the freezes and the pushes weeks before anyone announces either.
How: Scan a list weekly. Closed roles across a whole function usually mean a freeze; a doubling in one team means a push.
closedFunctions[] and a negative netChange are the freeze signal.
Competitive intelligence
Job descriptions leak the plan. A team hiring for a named database, a named CRM, or a compliance certification is telling you what they are building and what they have already bought, in their own words.
How: Scan with include_descriptions and read summary.toolsMentioned — extracted from the text itself, never inferred from a title.
Tools are only ever named when we actually read a description. Empty means we did not look, not that they use nothing.
Investors and analysts
Hiring shows up in a job board months before it shows up in a filing or a press release. A team that quietly doubled, or a whole function that vanished, is a fact you can date.
How: Scan a portfolio on a schedule and keep the history. Every scan is free to re-read, so the series builds itself.
The diff key is the platform own job id, so a retitled req is not counted as a new one.
Data teams
Greenhouse, Lever and Ashby agree on almost nothing — different field names, different team models, different ideas of what a location is. Normalizing that is a week of work and a permanent maintenance tax.
How: One role object, one function taxonomy, and an explicit capabilities block saying which fields that platform can populate.
Where a platform publishes nothing, the field is null and we say so. Nothing is inferred to make the shape look complete.
Autonomous agents
An agent qualifying an account has nowhere to put an API key. The x402 endpoint takes USDC on Base, resolves the board before quoting a price, and refuses free when there is no board to read.
How: POST the x402 endpoint, read the free preview in the 402, sign, retry. No signup, no key, no wallet beyond USDC.
Verification happens before settlement, because there is no refund on chain.
This reads a company’s public job board. It is not a people database, it holds no contacts, and it will not tell you who works there. It cannot see private reqs, agency searches, or a role filled before it was ever posted. And it reaches about four companies in five — the rest come back board not found, which is an honest blank, never a zero, and is never billed.