ATS Jobs Scraper & API: Greenhouse, Lever, Ashby, Workday and 5 more

Give this actor company names or domains. It finds each company's official job board across nine ATS platforms, Greenhouse, Lever, Ashby, Workday, SmartRecruiters, Workable, Breezy, Personio, Rippling, then returns the live postings as one clean, normalized schema, or just what opened, closed and changed since your last run.

The data comes from the ATS vendors' own public job-board APIs, the same endpoints each employer's careers page is built on. Not LinkedIn, not Indeed, no login, no cookies, no proxies, nothing that breaks when a social network changes its markup. If the employer publishes it, you get it, with a source URL you can click.

Why this beats a job-site scraper

	Job-site scrapers	This actor
Source	LinkedIn/Indeed HTML, changes without notice	The employer's own ATS API
Coverage per company	Whatever the site indexed, often days late	The complete live board, every run
Salary	Occasional text blob	Employer-published ranges as structured min/max/currency/period, plus careful text parsing, each value labeled with its source
Company identification	You search and hope	find_board mode tells you which ATS a company uses, with evidence and a confidence score
Change tracking	Diff it yourself	watch_changes emits opened / closed / changed events with the fields that changed
Anti-bot risk	Constant	None. These endpoints exist to be read

When you should use something else instead

There is a second kind of tool on this Store, and it is not a job-site scraper: an aggregated ATS index. Those crawl a very large number of boards ahead of time, hold a few million postings, and let you search them by keyword, location, skill or salary. They cover far more platforms than the nine here.

If your question is "find me senior Go roles in Berlin", buy one of those, not this. Searching an index is what they are for, and this actor does not do it.

Buy this one when the question is about named companies:

• "What is open at these 60 accounts right now?" This reads each employer's own board at request time, so a role closed an hour ago is closed here. An index refreshes on its own schedule.
• "Which of my accounts started hiring this week?" watch_changes emits opened, closed and changed events per company. An index gives you a different result set and leaves the diffing to you.
• "Does this company even have a public board, and where?" find_board answers that for a company that no index has crawled, with evidence and a confidence score.
• "Is that salary real?" Every derived field says whether the employer published it or we parsed it. Tools that infer skills and salaries with a model rarely tell you which numbers were stated and which were guessed, and on compensation that difference matters.

Recruiters, sales teams and competitive researchers are usually asking the second kind of question. Job seekers are usually asking the first.

The three modes

1. fetch_jobs: live postings for your companies

Input companies as names ("Ramp"), domains ("ramp.com", more precise), or board URLs (exact, and the only way to reach Workday tenants):

{
  "mode": "fetch_jobs",
  "companies": ["stripe.com", "duolingo.com", "https://jobs.ashbyhq.com/ramp"],
  "max_jobs_per_company": 100
}

Every job comes back in the same shape regardless of ATS. A real record, fetched 2026-08-10:

{
  "job_id": "ashby:ramp:34413f8d-26bf-4bbc-8ade-eb309a0e2245",
  "company_name": "ramp",
  "ats_platform": "ashby",
  "title": "Security Engineer, Cloud",
  "department": "Engineering",
  "team": "Backend",
  "location": {"raw": "New York, NY (HQ)", "city": "New York",
               "region": "NY", "country": "US", "is_remote": true},
  "workplace_type": "hybrid",
  "workplace_type_source": "published",
  "employment_type": "full_time",
  "seniority": "mid",
  "seniority_source": "inferred",
  "salary": {"min": 211400.0, "max": 290600.0, "currency": "USD",
             "period": "year", "source": "published"},
  "salary_disclosed": true,
  "posted_at": "2026-04-07T17:12:35Z",
  "age_days": 124,
  "job_url": "https://jobs.ashbyhq.com/ramp/34413f8d-26bf-4bbc-8ade-eb309a0e2245",
  "description_text": "...",
  "fingerprint": "60402780...",
  "fetched_at": "2026-08-10T09:52:00Z"
}

2. watch_changes: only what changed

The mode the others cannot do. Give it the same list on a schedule (daily is typical) with a stable watch_id:

{
  "mode": "watch_changes",
  "watch_id": "competitor-watch",
  "companies": ["stripe.com", "ramp.com", "duolingo.com"]
}

The first run baselines silently. Every later run emits only events:

{"event": "opened",  "board": "ramp.com", "title": "Staff Engineer, Payments", "job": {"...": "full record"}}
{"event": "closed",  "board": "stripe.com", "title": "Head of EMEA Sales", "last_seen": {"...": ""}}
{"event": "changed", "board": "ramp.com", "changed_fields": ["salary"],
 "before": {"salary": "150000-190000 USD/year"},
 "after":  {"salary": "165000-210000 USD/year"}}

What people build on this: competitor hiring intel (a burst of sales roles in Berlin is a market-entry signal), account-based sales triggers ("they just opened three data-engineering roles, call them about your data product"), recruiting intelligence (a closed role means a hire or a freeze), salary-band monitoring on specific companies.

Honesty rule built in: if an ATS is down or rate-limits during a run, that board emits no events and its baseline is kept, and the failure is named in fetch_errors. A vendor 503 will never be reported to you as "all their jobs closed". Deleting a company from your list drops its baseline without emitting closes, either.

3. find_board: which ATS does this company use?

{"mode": "find_board", "companies": ["Ramp", "NVIDIA", "stripe.com"]}

{"company_input": "Ramp",
 "best": {"platform": "ashby", "token": "ramp", "open_jobs": 122,
          "confidence": 0.99,
          "evidence": "slug probe 'ramp'; linked from the company's own careers page; token matches the company domain"},
 "boards": ["...every candidate found, ranked..."]}

Useful on its own (ATS market research, lead routing by tech stack) and as the precision step before a big fetch_jobs or watch_changes run.

What we will tell you that others will not

• Every derived field names its source. workplace_type_source, seniority_source and salary.source are published (the employer's own structured field), parsed (read out of the employer's own text), or inferred (our guess from the title). Filter on published-only if your use case demands it.
• Absent means absent. No salary published, salary: null and salary_disclosed: false. We never fabricate a "market estimate".
• Discovery shows its evidence. Every board comes with how it was found, its live job count, and a confidence score. When we cannot confirm a board (NVIDIA hides its Workday tenant behind scripts, for example), you get best: null plus candidates and the reason, never a guess dressed as an answer. For Workday, paste the board URL and it resolves exactly.
• A failed lookup is named, not swallowed. Rate limits and outages per platform appear in failed_platforms / fetch_errors, so "not on Personio" and "Personio was busy" stay different answers.

Coverage and limits, stated plainly

• Platforms: Greenhouse, Lever, Ashby, Workday (by board URL), SmartRecruiters, Workable, Breezy, Personio, Rippling. That covers most venture-backed and mid-market tech employers and a large share of the enterprise. Not yet covered: iCIMS, Taleo, SuccessFactors, BambooHR (no public board JSON), Teamtailor (requires a per-company key).
• Workday needs the board URL (e.g. https://acme.wd5.myworkdayjobs.com/External): tenant hosts are not guessable from a company name, and we say so rather than pretend.
• Companies per run: 500 for find_board and watch_changes, 200 for fetch_jobs.
• watch_changes needs descriptions consistent: keep include_descriptions the same between runs of one watch, or the run after the flip reports every description as changed. Default is off for watches.

Pricing

Pay per event, no subscription:

Event	When
board_lookup	Per company successfully resolved in find_board (unresolved companies are free)
job	Per job returned in fetch_jobs. No rounding. $1.30 per 1,000 jobs on the free tier, falling to $0.90
board_watched	Per board checked in a watch_changes run
change_event	Per opened/closed/changed event emitted

A daily watch on 50 companies costs 50 board_watched per day plus the handful of real events; a full pull of a 500-job enterprise is 500 job.

Quick starts

Sales triggers, daily, into your CRM:

{"mode": "watch_changes", "watch_id": "abm-accounts",
 "companies": ["snowflake.com", "databricks.com", "confluent.io"]}

Recruiting market map, one shot:

{"mode": "fetch_jobs",
 "companies": ["stripe.com", "ramp.com", "brex.com", "mercury.com"],
 "include_descriptions": true}

Which ATS do my 200 target accounts use?

{"mode": "find_board", "companies": ["...200 domains..."]}

Not affiliated with or endorsed by any ATS vendor; all trademarks belong to their owners. Reads only public, unauthenticated job-board endpoints that employers publish for exactly this purpose.