Wellfound Startups Scraper

Build a list of startups that are hiring. Every company in Wellfound's directory becomes one row with its name, its own website, its one-line pitch, its longer description, its headcount band, the industries it files itself under, the offices it lists, the badges Wellfound awards it, how many roles it has open and the breakdown of those roles by function.

That is a prospect list with a buying signal attached: a company appearing here is advertising that it is hiring, and open_jobs_count says how hard.

What a row looks like

{
  "company_id": "28804",
  "company_name": "HomeLight",
  "company_slug": "homelight",
  "company_url": "https://wellfound.com/company/homelight",
  "company_website": "https://www.homelight.com",
  "company_pitch": "Our vision is a world where every real estate transaction is simple",
  "company_description": "We're the essential technology platform used by hundreds of thousands…",
  "company_size": "SIZE_501_1000",
  "company_markets": ["Real Estate", "Marketplaces"],
  "company_locations": ["San Francisco", "Scottsdale"],
  "company_badges": ["Actively Hiring", "Responds within two weeks"],
  "open_jobs_count": 16,
  "open_roles": [
    { "role": "Marketing", "count": 2 },
    { "role": "Operations", "count": 2 },
    { "role": "Product", "count": 1 }
  ]
}

company_website is the company's own domain rather than its Wellfound page, which is what makes these rows usable as a prospect list without a second enrichment step somewhere else. company_size is Wellfound's own coding — SIZE_1_10 through SIZE_1001_5000 — and it is self-reported, so it is a band rather than a headcount.

Accepted input

{
  "maxItems": 100,
  "remote": false
}

maxItems is how many companies to save, up to 500. remote narrows the directory to companies hiring remotely; unlike the job search, the directory does publish a remote variant, so it is always valid here.

Funding, team and perks

enrichCompanyProfile reads each company's own profile and adds total_raised, funding_rounds with the amount raised, the valuation, the close date and the press article behind each one, team with titles, classification and tenure, perks, the company's own open_jobs, and its LinkedIn, X and blog links. sections chooses which tabs are read — overview, funding, jobs, people, culture — and it defaults to overview and funding.

It is one read per company and it is billed per company enriched, so it is off by default. Runs that use it take noticeably longer than runs that do not.

Questions

Why does a round have no amount on it? Because funding was not among the requested sections. The overview tab lists a company's rounds with their type and close date and omits the amount raised and the press link entirely — measured on HomeLight, whose eleven rounds come back with raised_amount: null from overview and with $60M, $55M, $263M and $100M from the funding tab. Request ["overview", "funding"], which is the default, and the two are merged by round so nothing is lost.

Why do some enriched companies have an empty perks or team? Either the tab was not requested or the company hid it. hidden_sections on each row says which, and it is the reason an empty section is readable at all: ["SALARIES"] means the company withheld its salary tab, not that the read failed. Salaries are not offered as a section here — Wellfound redirects that page and most companies hide it, so offering it would promise data the site does not serve.

open_jobs is shorter than open_jobs_count. Which is right? Both, and they answer different questions. open_jobs_count is the number the company's page states; open_jobs is the postings actually listed on the tab that was read, and only the ones belonging to that company. The overview tab renders a strip of recommended postings from other companies in the same block — OpenAI, which has no open roles at all, comes back with ten postings on its overview and every one belongs to somebody else — so postings are matched to their own company and the rest are dropped rather than reported as this company's.

How many companies can one run collect? Up to 500. The directory serves about twenty companies per read, so 500 is roughly twenty-five reads, and a walk stops early when the directory stops producing new companies. There is no filter for industry or city on this surface: the directory is one ordered list plus the remote variant, so narrowing happens after the export, on company_markets and company_locations.

Is this the same company record the jobs scraper attaches to a posting? The columns are the same names, so the two datasets join on company_slug, but the directory row is fuller: a posting carries the company's name, pitch, size, logo and badges, while the directory adds the website, the long description, industry tags, offices and open-role counts.

Are these companies ranked by anything? They arrive in the directory's own order, which Wellfound does not explain and which is not a ranking this Actor can reproduce or sort by. Sorting on open_jobs_count or on the badges is the honest way to prioritise the export.

Related Actors

Wellfound Company Scraper takes a list of company_slug values and returns full profiles, which is the cheaper order when the companies are already known.

Wellfound Jobs Scraper approaches the same market from the roles rather than the companies, and every one of its rows carries the hiring company.