25 ATS portals · one canonical ontology · applicant-first

Every job.
One ontology.
Answer once.

Fetch ten thousand jobs and something surprising happens: every screening form in the world collapses into a small mapping. Jobley grabs everything, normalizes it into one ontology, and keeps that ontology honest with you.

No account needed. The playground runs as Demo User · demo@gmail.com on a real seeded workspace.

Ontology collapsenormalizing
21,196 questions seen 20 : 1 1,053 canonical fields
Same market · different pace
25ATS adapters
9,400+Postings fetched
541Company boards
21,196Screening questions
1,053Canonical fields
How it works

Grab. Normalize. Cross-maintain.

21,196 screening questions from the wild reduce to 1,053 canonical fields. That 20:1 collapse is the whole product: past ~10k jobs, the space of possible form fields is finite — and ownable.

01
GRAB

All postings, direct from the ATS

Adapters hit the public APIs of 25 ATS portals and pull every posting and every application form — fields, options, required flags — into one common shape.

GET greenhouse/boards/stripe
→ 214 postings · 1,908 fields · ok
02
NORMALIZE

Everything maps to the ontology

Every question is classified against priority-ordered mappings into a canonical profile key. Twenty ways of asking “are you authorized to work?” become one field.

"Are you legally authorized…"
work_authorization · conf 0.98
03
REDO CALLBACKS

You keep the ontology honest

Hit something unmapped? Answer it once. Jobley learns a new mapping, fires a redo pass over everything already crawled, and that question resolves everywhere, forever.

"Describe your sommelier cert"
cert_sommelier · redo 3 postings
The AI layer

Ontologies, leads, and prep-books — generated for you.

Claude-powered classification runs over every form we crawl. But the model isn’t just filling fields — it’s building applicant-side intelligence.

  • Custom screening ontologies — your own canonical field set, generated from the questions you actually face and refined by your redo callbacks.
  • Lead mapping — recruiter threads from LinkedIn and Gmail parsed into structured leads: contacts, companies, and upcoming calls flagged automatically.
  • Pre-interview prep-books — before every call, a generated brief: the role, the form you filled, the thread history, and what to say next.
Normalization feedlive
“Are you legally authorized to work in the US?”
greenhouse · seen 214×
work_authorization
“What are your pronouns?”
ashby · seen 96×
pronouns
“Desired compensation range?”
lever · seen 143×
comp_expectation
“Describe your sommelier certification”
workable · unseen
redo → learned
Recruiter reply · call Thu 2:30pm
gmail thread · prep-book ready
lead
Applicant-first, not company-first

Engagement recommendations built around you.

Every other tool organizes the world by company. Jobley organizes it around the applicant: leads emerge across every surface — ATS portals, company boards, LinkedIn threads, Gmail — and get routed to you as one prioritized queue of what to engage next.

Lead emergence, all surfaces

A recruiter DM, a reply buried in Gmail, a fresh posting on a board you starred — the moment a lead surfaces anywhere, it lands in your queue with full context attached.

Recommendations, not dashboards

Not “here are 500 companies” — but “these three threads are warm, this call is Thursday, this application is one answer from ready.” Ranked by your momentum, not their org chart.

Your profile compounds

Every answer, document, and redo callback enriches one applicant profile. The more you engage, the less the system ever asks of you again.

Discovery crawlbfs · dfs
BFS sweep · Greenhouse
208 boards · 3,914 postings
live
DFS probe · Workday tenant
nested board · 3 levels deep
found 47
Liveness check · Ashby
1,120 postings probed
31 expired
Board migration detected
lever → ashby · trail preserved
re-linked
Schema drift · new field
registered, sync unbroken
absorbed
Deep discovery

BFS/DFS over the whole hiring graph.

Job boards decay. Links rot. Companies switch ATSes mid-search. Jobley runs deep breadth-first and depth-first topology searches across ATS portals and company boards, so discoverability never silently degrades.

  • Full-topology crawls fan out across all 25 ATS portals and 541 company boards
  • Liveness probes tell live postings from ghosts — expired roles clear themselves out
  • Board migrations tracked: when a company moves ATS, the trail follows
  • Schema drift absorbed: new fields register instead of breaking the pipeline
Coverage

25 ATS portals, an adapter for each.

Built by people who job-hunt too

“When will you have a job?”
You’ll have an answer.

The forms are finite. The ontology is small. Your time should go to interviews, not to typing your work authorization for the two-hundredth time.