◆ Why we exist

Any LLM ranks the best CV.
Huntchy knows who this client says yes to.

Same role, same candidates. Watch what a raw LLM gives you — then watch the layer that only exists because Huntchy learns from every real outcome. That layer is the whole company.

The match under test illustrative case

Senior Java Engineer — Banco Atlântico

€420–480/day · banking · 5+ yrs · Lisbon 2d/wk on-site

Candidate A

Best engineer. Clean Kafka/microservices, 7y. No banking. Wants €520/d. Remote-only.

Candidate B

6y Java/Spring. Two banking builds (SEPA, fraud). Asks €450/d. Fine with on-site.

Candidate C

Strong profile, 3 roles in 4y. Wants €500/d.

A raw LLM

reads the text

Huntchy

reads the outcomes

◆ What ChatGPT / any LLM tells you

Candidate A — best match

Strongest technical profile: 7 years, cleanest architecture work, deepest Kafka. Matches the senior bar and the core stack better than the others. B is solid but a step below technically; C has a job-hopping pattern.

What it read: the job description + the three CVs. That's all it has. It ranks the best engineer against the words on the page — exactly what any recruiter gets free in 30 seconds.

◆ What Huntchy tells you

Candidate B — wins the submission

A is the better engineer. B is the one this client accepts. Submit A and you likely burn the slot; submit B and you close. Not because of the CV — because of what Banco Atlântico has actually done before.

▾ The layer beyond the LLM — three things only Huntchy has

◆

Revealed client preference

Banco Atlântico rejected the last 3 sector-outsiders — even ones stronger on paper than A. Their JD says "Java"; their behaviour says "banking domain, non-negotiable." A loses here, and the JD never told you that.

from outcomes · not the CV

€

The rate that actually closes

This client closes Senior Java at €420–480. A wants €520 — above the line on every placement they've made. B at €450 sits inside it. The best engineer dies on rate before the interview.

from 11 past placements

⇄

Whether the candidate accepts the client

A is remote-only; this role is 2 days on-site. A has declined two on-site roles already. Even if the client said yes, A would likely say no. A match needs both sides — the LLM only checked one.

from candidate signal

◆ What actually happened

The client took B. They never interviewed A — passed at CV stage, "not enough banking." Exactly the call Huntchy made, and the exact opposite of the LLM. The layer was right because it learned from the last 11 times.

A raw LLM ranks the candidate against the job description.
Huntchy ranks the submission against what this client has actually accepted.
That gap is the product. It's the only thing a competitor can't copy with a better prompt.

Why it compounds (and a prompt never will)

A recruiter closes a match — and confirms the outcome

Both sides give 2-tap feedback — client accepted? candidate fit?

The graph learns what this client really wants — revealed, not stated

The next match for that client is sharper than any LLM can be

↻ Every outcome makes the layer deeper. The data is the moat.

Any LLM ranks the best CV.
Huntchy knows who this client says yes to.

Why it compounds (and a prompt never will)

Product

Company

Legal

Any LLM ranks the best CV.Huntchy knows who this client says yes to.

Why it compounds (and a prompt never will)

Product

Company

Legal

Any LLM ranks the best CV.
Huntchy knows who this client says yes to.