HProfile-to-profile matching›in private beta

Matched as a person.Not your title. Not your CV. Not a keyword.

You drop a CV. Claude builds a richer profile — how you work, what you value, the kind of people you do your best work with. Employers describe the human they need, not just the role. We match profile to profile.

Build my profileHow matching worksEncrypted on upload · no credit card
9.2
Profile matchhuman ↔ human
14:02 · scored just now
← you
Pablo G.E.
8 yrs · EU · lead-IC
deep-focusasync-defaultcraft > speedB2B SaaS
them →
Sourcegraph · Cody team
staff IC · EU · 6-person crew
deep-focusasync-defaultcraft > speedcustomer-facing
why a 9.2
Same depth-of-work pattern. Both default to async. Both ship to B2B users at 50–500 person teams. The role title is incidental.
person ↔ personnot title ↔ title
01 / profile

Built from your CV, deeper

Skills are the table stakes. Values, working style, scope, team-shape — those are the signal.

02 / match

Profile to profile

Employers describe the human they need. We compare attribute-by-attribute, not keyword-to-keyword.

03 / snapshot

Tailored, proof-checked

CV + cover letter drafted on demand. Every claim traces back to your real profile.

04 / private

Server can't read it

Your profile, hunts, and snapshots stay ciphertext. We physically can't peek.

profile.match(you, them)person ↔ person, not title ↔ titlematch_attribute_by_attribute()CV → encrypted profilescan(linkedin · rss · custom)generate_application_snapshot()proof_read · claim-traceDEK · AES-256-GCM · PBKDF2 · two modesapply · audit · doneprofile.match(you, them)person ↔ person, not title ↔ titlematch_attribute_by_attribute()CV → encrypted profilescan(linkedin · rss · custom)generate_application_snapshot()proof_read · claim-trace
01 Why hunt.work

Most job tools match titles to titles.
We match people to people.

A role title and a keyword list aren't who you are. They aren't what an employer actually needs either. We model both sides as humans — then match attribute by attribute.

01 / matching

Person to person. Not keyword to keyword.

Claude builds a richer profile of you — working style, values, scope, the teams you thrive in. Employers describe the human they need. We compare those, not job titles. The fit score reflects that.

you.styledeep-focus, async
them.styledeep-focus, async
you.scopelead-IC, 50–500
them.scopestaff IC, 50–500
match()9.2 · 7 / 8 attrs
02 / privacy

Encrypted before the network sees it.

Your email is the key. Your profile, hunts, and snapshots are encrypted on your device. We store ciphertext, orchestrate the work, and can't read a thing — even if we wanted to.

profile.v34   7b3a92f1c4e8a05d…
hunt.eu      a09c41dd6e7b3a92…
snapshot.175 f1c4e8a07b3a92dd…
→ server sees ciphertext only
03 / craft

Snapshots, not spam.

One tailored CV + cover letter, generated only when you say go. A proof-reader cross-checks every claim against your encrypted profile before you submit — no fabricated achievements, no "honored to apply" filler.

SOURCEGRAPH · STAFF AI ENG · COVER LETTER
Shipped LLM-backed product to 200k+ B2B users at BetaCorp, …
Led the architecture for EU GDPR-compliant inference across …
"helped over 500 enterprises" ← no trace
02 How it works

Four steps. The rest is in the background.

You bring the CV and the criteria. Claude handles the boring surface area — scanning, scoring, drafting, proof-reading — so you can spend your energy on the conversations that matter.

step 01

Drop a CV.
Get a structured profile.

Claude parses your CV into a typed, encrypted profile. Never stored as a master document — this becomes the source of truth that every future application traces back to.

pablo-cv.pdfencrypted · v34
// parsed profile, encrypted at rest
{
  "display_name": "Pablo G.E.",
  "work_history": [11 entries],
  "languages": ["EN/C2", "ES/N", "DE/B2", …],
  "skills": [18 entries],
  "education": [4 entries],
}
step 02

Describe the person you'd thrive with.
Not just the role.

What kind of team do you do your best work with? What working style? What scope of problems? Claude turns the conversation into a structured profile of your ideal match — the role title is just one slot among many.

Claude
What kind of team do you do your best work with?
Small, async, craft-obsessed. Shipping to real users, not internal stakeholders.
Claude
Got it. What working style is a no-go?
Your match profile
styledeep-focus, async, craft > speed
teamsmall · ships to users
scopelead-IC, 50–500 person co.
avoidconsensus-by-meeting
step 03

Pick sources.
Claude scans on cadence.

LinkedIn search URLs, RSS feeds, company boards, or any URL with a {keyword} slot. Claude scans on schedule, scores every listing, and surfaces matches in your pipeline.

LinkedIn — AI roles EU
linkedin.com/jobs/search/?keywords={keyword}
Anthropic careers RSS
anthropic.com/jobs.rss
WeWorkRemotely — Product
weworkremotely.com/categories/remote-product-jobs
step 04

Review snapshots.
Apply with confidence.

Claude drafts a tailored CV + cover letter for each match you green-light. A proof-reader pass cross-checks every claim against your profile — flagged claims show up red. No fabricated achievements, no "honored to apply" filler.

Cover letter — Sourcegraph
tailored from profile.v34

Shipped LLM-backed product to 200k+ B2B users at BetaCorp, leading a team of 6 engineers through a full GDPR-compliant migration.

At EventTech I built a multi-tenant inference platform for Series-B SaaS customers, anchored on…

Proof-reader · 8 claims · 8 verified
"200k+ B2B users" — profile.work[0].metric_3
"team of 6 engineers" — profile.work[0].team_size
03 Inside the app

One pipeline. Every hunt, every source, scored and sorted.

The whole product is built around a single triage surface. Open a row, read the AI's case, generate a snapshot when it earns the time.

Preview · private beta · public sign-ups coming soon

hunt.work
Pipeline 13Hunts 2Sources 3
P Pablo G.E. · seeker
Encrypted at rest · audit-chain verified · server can't bypass row-level isolationlast audit · 14:02

Pipeline

13 listings · last scan 8m ago · next in 3h 52m
Role · CompanyHuntLoc · CompStatusUpd
9
Staff AI Engineer — Implementation
Sourcegraph
AI Impl. Lead — EU
EU remote USD 180–240k
new5h
9
Head of AI Solutions — EMEA
Vercel
AI Impl. Lead — EU
Valencia ok EUR 130–170k
reviewing1d
8
Manager, Applied AI Architecture
Anthropic
AI Impl. Lead — EU
London undisclosed
new3h
7
Lead PM · LLM Platform
Linear
AI Impl. Lead — EU
EU remote EUR 110–145k
reviewing6h
6
Director of Product, AI Workflows
Notion
PM Director — B2B
NYC hybrid USD 220–280k
applied9h
4
Engineering Manager · ML Platform
Mistral
AI Impl. Lead — EU
Paris onsite EUR 140–180k
passed3d
01 / triage surface

One table, every hunt

Filter by status, hunt, source, or fit floor. Mono numerals on the right for instant scan. New listings get a periwinkle rail.

02 / ai co-pilot

Live queue, no chat window

Claude's tool calls stream into the side rail in real time. No hidden agent loop. No surprise emails sent on your behalf.

03 / trust chrome

Encryption banner, always

The DEK lease bar is permanent, not buried in settings. Last audit timestamp visible on every screen.

04 Security model

The job hunt should be private. So we built it that way from key one.

Private by policy by default. Zero-knowledge when you opt in.

  • Standard mode. AES-256-GCM on every sensitive field at rest. TLS in transit. Per-account row-level isolation enforced by the database itself.
  • Max-Privacy mode. Pick a passphrase during onboarding — your browser derives a DEK (PBKDF2-SHA256, 200k iterations) and the server holds ciphertext only. Switching modes re-encrypts in a single transaction.
  • Hash-chained audit log. Every mutation is written by a Postgres trigger and chained to the previous row. Tampering breaks the chain at the exact row. Visible to you, exportable on demand.
  • GDPR-native. Delete is one click and the deletion is real, not a soft flag. EU data residency from day one.
data flow · today
Your device
CV uploaded · profile assembled · TLS in transit
on-device
↓ TLS · authenticated session
hunt.work server
Encrypted at rest · row-level isolation · audited
eu-west · audited
↓ scoped request · per-call
Claude inference
Plaintext within a scoped, audited call only
audited
05 Private beta · since Mar 2026

Tiny on purpose. Aimed at signal, not volume.

312k
Listings scored across active hunts
4.2k
Snapshots generated & reviewed
91%
Snapshots applied to without edits
0
Profiles, hunts, or snapshots leaked across tenants
06 Pricing

Free to start. Pay when you're actually looking.

The free tier covers casual browsing. Premium unlocks the cadence and proof-reader that make the search feel quiet.

Free
€0 forever

For casual lookers. Get a feel for how it scores against your criteria.

  • 10 snapshots / month
  • Bring-your-own-key bypass (free unlimited via your API key)
  • Unlimited hunts & sources
  • Profile-to-profile matching
  • Encrypted at rest · audited
Start free
Beta
Premium
Coming Q3 2026

Private beta now. Reach out for access — we're onboarding a handful of seekers ahead of the wider opening.

  • Unlimited tailored snapshots
  • Multilingual CVs (EN, ES, DE, FR + on request)
  • Priority match queue
  • Premium CV templates
  • Full CV history retention
Request beta access

Free tier ships now. Premium pricing lands when Stripe goes live in Q3.

07 FAQ

Questions we get — mostly about the encryption.

If you've got more, drop us a line — [email protected].

01

Wait — what do you mean "profile to profile"?

Most tools score your CV title against a job title and call it a match. We don't. Claude builds a richer profile of you — how you work, what you value, the kind of team you do your best work with. Employers describe the human they need, not just the role. We compare those two profiles attribute-by-attribute. A role title is one slot. Your working style, scope, and values are the rest.

02

How is matching actually done?

Matching happens inside a scoped, audited Claude call — your profile slice + the listing's profile, just long enough for the call. Claude returns the score and reasoning, the slice is discarded. Everything is logged in your audit log.

03

Can I use my own Claude API key?

Yes — Free tier supports bring-your-own-key. Paste it into Settings, your inference calls route through your account at Anthropic instead of ours, and the monthly snapshot cap doesn't apply. Useful if you're a heavy user before Premium lands.

04

Does this work with LinkedIn / Indeed / etc?

Today: LinkedIn search URLs, generic RSS, public job boards (WeWorkRemotely, Remotive), and any custom URL with a {keyword} placeholder. Indeed and Glassdoor are on the roadmap pending TOS-compliant integration. We don't auto-apply on your behalf — you always click submit.

05

Is the snapshot just AI slop with my name on it?

It's tailored from your actual profile, with a proof-reader pass that flags every sentence that doesn't trace back to a parsed CV claim. You see the citations side-by-side. You can edit anything. No "honored to apply" filler.

06

Will employers see "made with hunt.work" on my application?

No. Snapshots are plain PDFs, no watermark, no metadata trail. The point is to make your search invisible to everyone except you.

Stop sending your CV. Send your profile.

Three minutes to seed it. Then matches surface in the background — on cadence, in private, scored as you, not as your title.

Get a key — freeRead the security model
Encrypted at rest. No credit card. No mass-apply.
hunt.work — matched as a person, not your title or CV