Y Combinator Company Data for AI Agents: Batch, Status and Founders

Y Combinator company data for AI agents means an agent can ask which batch a company joined, whether it is still active and who founded it, and get the answer from the directory as JSON instead of from the model's memory. One call returns the profile; two search sources find the slugs. The Y Combinator company source page is the reference for exact fields, request shape and prices.

The short version

• —What you get. Batch, status, team size, location, website, social links, founders with bios, news and Launch YC posts for one company, plus two search sources that list companies and founders by batch, industry, region and team size.
• —Who it is for. Teams and agents that need accelerator alumni lists, founder background before a call, or a batch and status column added to a company list.
• —What it does not do. No funding rounds, investors, valuation or contact details. The directory is a profile of the accelerator relationship, not of finances.
• —What it costs. A company lookup is a flat 5 points ($5 per 1,000 lookups). Searches cost 1 point per returned record ($1 per 1,000). Misses are not billed.
• —How you reach it. The same request works over REST and as an MCP tool, so a script and an agent use identical arguments.

The workflow, step by step

Y Combinator publishes a public directory of the companies it has funded and the founders behind them. Reading it by hand works for one company. It stops working when you want every company in a set of batches, or the same fields for two hundred names in a spreadsheet. The catalog has three Y Combinator sources that cover the directory from both ends: the company lookup, a companies search and a founders search. They share one request format and one response envelope, so the steps below use one client.

1. Find the companies

The companies search returns company records: name, link, one-liner, description, batch, status, stage, team size, industries, regions, tags, location, website, a hiring flag and a launch date. Filters are optional and combine: keyword, batch, industry, region and team size. Values that take several entries, such as batches, are comma-separated, for example "Winter 2024, Summer 2024". max_items caps the records returned, up to 250, and start_page continues past the first page.

In the source page's sample, a request with no filters and max_items set to 2 reported 6,283 companies found and returned the first two, DoorDash (Summer 2013, Public) and Airbnb (Winter 2009, Public). That count was what the directory reported when the sample was captured; it will not be the same number later.

2. Screen on the search fields

Keep the companies that match your screen: still active, hiring, inside a team-size range. This costs nothing extra, because the search already returned those fields. Filters are applied by the directory itself. The source does not rank, score or merge records, so any ordering beyond the directory's own is yours to add.

3. Look up each slug

The company lookup takes one field, company, set to a slug such as airbnb or to the link to the company page. It returns founders with their public titles, bios and profile links, the website, social links, news and Launch YC posts. This is the step that carries the detail, and the one an agent usually calls on its own when a user names a company.

4. Find people by role, if you need them

The founders search returns founder records: name, first and last name, title, company and company link, batches, industries, region and avatar. In its sample, the directory reported 13,943 founders found, and the first two returned were Andy Fang and Tony Xu, both at DoorDash in batch S13. One of them has a null primary title with "Founder" in the titles list, which is a reminder to read both fields. The founders search has no bios or social links; use the company lookup for those.

A worked example

Acme Oy, a Finnish company in this example, sells onboarding software to early-stage startups and wants a list of accelerator alumni to approach. This is illustrative, not a real customer.

Acme's analyst filters the companies search to two recent batches and a team-size range, and keeps the companies marked as hiring. For each result she looks up the slug, reads the founders and their public titles, and saves the company website and the status. A company whose status is no longer active is dropped before anyone writes to it. Nothing in the response gives an email address, so the team uses its own approved process for contact details, and it checks its handling of the founders' personal data against its own obligations first.

Why not a funding database, a scraper or a general vendor

A funding database tells you rounds, investors and valuation; this tells you batch, status and founders. They answer different questions, and the startup funding data API guide covers how the funding sources compare. A scraper of the directory works, but the layout, the output shape and the upkeep are yours, repeated for every platform you add; here the next source means changing the endpoint name and the body. A general company-data vendor covers far more than accelerator alumni. The Y Combinator sources are narrower on purpose, priced per lookup, and the right choice when the question is "who came through this accelerator, and what is their status now".

One real request and response

This is the company lookup for airbnb. The request is the source page's example call, and the response is a real captured response, not a hand-written one.

The request

curl -X POST https://api.neuralverge.ai/functions/v1/run-yc-companies \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"company":"airbnb"}'

The response

The session id is a placeholder, and the long text fields (description, founder bios and the photo link) are shortened in the captured sample itself, not by this guide. The news array holds two items in the sample while news_count reads 5, so the list is a shortened version of the full set.

{
  "session_id": "11111111-1111-1111-1111-111111111111",
  "kind": "yc_companies",
  "company": "airbnb",
  "company_id": "airbnb",
  "human": "# Y Combinator — company\n\n- **Id:** airbnb\n- **Pages opened:** 1\n- **Proxy traffic:** 114.7 KB\n- **One-liner:** Book accommodations around the world.\n- **Batch:** Winter 2009\n- **Status:** Public\n- **Team size:** 6132\n- **Location:** San Francisco",
  "machine": {
    "company": {
      "id": "airbnb",
      "alias": "airbnb",
      "yc_id": 271,
      "name": "Airbnb",
      "url": "https://www.ycombinator.com/companies/airbnb",
      "one_liner": "Book accommodations around the world.",
      "description": "Founded in August of 2008 and based in San Francisco, California, Airbnb is a trusted community marketplace for people to list, discover, and book unique accommodations around the world — online or fr",
      "batch": "Winter 2009",
      "batch_code": "W09",
      "status": "Public",
      "website": "http://airbnb.com",
      "logo": "https://bookface-images.s3.amazonaws.com/small_logos/3e9a0092bee2ccf926e650e59c06503ec6b9ee65.png",
      "tags": [
        "Marketplace",
        "Travel"
      ],
      "team_size": 6132,
      "year_founded": 2008,
      "location": "San Francisco",
      "city": "San Francisco",
      "country": "US",
      "linkedin_url": "https://www.linkedin.com/company/airbnb/",
      "twitter_url": "https://twitter.com/Airbnb",
      "facebook_url": "https://www.facebook.com/airbnb/",
      "crunchbase_url": "https://www.crunchbase.com/organization/airbnb",
      "github_url": null,
      "primary_group_partner": {
        "name": "Garry Tan",
        "url": "https://www.ycombinator.com/people/garry-tan"
      },
      "founders": [
        {
          "name": "Brian Chesky",
          "title": "Founder/CEO",
          "bio": "Brian Chesky is the co-founder,  Head of Community, and  CEO of Airbnb, which he started with Joe Gebbia and Nathan Blecharczyk in 2008. Brian sets the company’s strategy to connect people to unique t",
          "linkedin_url": "https://www.linkedin.com/in/brianchesky/",
          "twitter_url": "https://twitter.com/bchesky",
          "is_active": true,
          "user_id": 21981
        },
        {
          "name": "Nathan Blecharczyk",
          "title": "Founder/CTO",
          "bio": "Nathan Blecharczyk is the co-founder, Chief Strategy Officer, and Chairman of Airbnb China. Nathan plays a leading role in driving key strategic initiatives across the global business. Previously he o",
          "linkedin_url": "https://www.linkedin.com/in/blecharczyk/",
          "twitter_url": "https://twitter.com/nathanblec",
          "is_active": true,
          "user_id": 21988
        }
      ],
      "founder_names": [
        "Brian Chesky",
        "Nathan Blecharczyk"
      ],
      "photos": [
        "https://bookface-images.s3.us-west-2.amazonaws.com/attachments/8a7236c94b4d9b78b67ce66e02cbca497e632d99.png?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=ASIAQC4NIECAMCQGXKPC%2F20261009%2Fus-west-"
      ],
      "news_count": 5,
      "news": [
        {
          "title": "Airbnb CEO Brian Chesky on taking it back to basics: ‘I can’t make products just for 41-year-old tech founders’ - The Verge",
          "url": "https://www.theverge.com/2023/5/9/23716903/airbnb-ceo-brian-chesky-rooms-ai-travel-future-of-work-summer-2023",
          "date": "May 09, 2023"
        },
        {
          "title": "Airbnb launches Airbnb Rooms listing category for budget travel",
          "url": "https://www.usatoday.com/story/travel/news/2023/05/03/airbnb-rooms-listing-category-budget-travel/70178696007/",
          "date": "May 03, 2023"
        }
      ],
      "jobs_count": 0,
      "launches_count": 0,
      "launches": []
    },
    "blocked": false,
    "pages_opened": 1,
    "proxy_traffic_bytes": 117434
  },
  "total_points": 5
}

Reading the response

The envelope is the same for every source: a session_id you can trace, a kind naming the source, a human field with a readable summary, a machine object holding the structured result, and total_points, the points charged. The machine object also reports blocked, pages_opened and proxy_traffic_bytes, so you can see what the run did.

Start with total_points: the run cost 5 points, the flat price of a company lookup. Then the profile itself. Airbnb comes back as Winter 2009 (batch code W09), status Public, a team size of 6,132, based in San Francisco. Two founders are listed, each with a title, a bio and public profile links, and founder_names repeats just the names for when you only need those.

Three details show how to handle the output in code. github_url is null, so a missing link is an explicit null rather than an absent key. jobs_count and launches_count are 0 and launches is an empty array, so an empty result is still a well-formed one. And news_count is 5 while the list shows two items, so treat the list as a sample of the news, not a complete archive.

REST and MCP

The request above is the REST form. The same source is exposed as an MCP tool with the same name and the same arguments, so an agent can call it directly:

{ "tool": "run_yc_companies", "arguments": {"company":"airbnb"} }

Both routes reach the same source and return the same envelope; the choice is about who makes the call. REST suits pipelines and scheduled jobs. MCP suits an agent that decides mid-conversation to look a company up, for instance when a user asks who founded a startup and when it joined the accelerator. The search sources are available the same way. For connecting the hosted MCP server to a client, see the MCP setup guide.

Cost and limits

Pricing comes from the source pages and counts in points, which come out of your monthly plan allowance. The pricing page has the plans.

Searches are billed per returned record and capped by max_items, which can go up to 250. In the two search samples, two records returned cost 2 points, which matches the per-record rate. A request that returns nothing costs nothing, and a lookup that finds nothing is not billed.

What the sources do not do

• —No funding data. The directory page does not list funding rounds, investors or valuation. A profile that says "Public" does not tell you what the company raised.
• —Team size, status and news can lag. The directory is read at request time and nothing is served from a stored copy, but the numbers are as the directory lists them, which can be behind the company.
• —No scoring or de-duplication on search. The source does not rank, score or merge records.
• —No contact details. Founder records do not carry email addresses or phone numbers.

Pitfalls to plan for

A lookup that finds nothing returns no data, so your pipeline should treat a miss as a normal outcome, not an error. Parsers should expect null links and empty arrays, as in the sample. Founder names, titles and bios are personal data, so decide what you store, for how long and for what purpose.

On the trade-off itself: reading the directory by hand is free and slow, building your own scraper costs engineering time and a maintenance commitment, and the catalog costs points per lookup with the same request format as every other source. For a one-off check on a single company, the directory in a browser is the right tool. For a list, a recurring pull or an agent that needs the answer inside a workflow, the per-lookup price is usually cheaper than the hours.

How to decide

Check these before you build on any Y Combinator data source:

1. —Which fields do you need? List them against the response above. If you need funding, valuation or investors, this source is only part of the answer.
2. —How will you get slugs? Start from a list you already hold, from the companies search, or from the founders search.
3. —Is the use case a fit? It works well for alumni and portfolio research, batch lists for outreach, sector and region scans (fintech in Europe, say), founder background for meeting prep, adding website, batch and social links to a CRM, and agent tools that answer from a source.
4. —How fresh must the data be? If a decision depends on a company's current state, confirm with a second source.
5. —What is the volume? Multiply lookups by 5 points and search records by 1 point, and compare the total against your plan allowance.
6. —Which route do you need? REST for pipelines, MCP for agents; both are available.
7. —How will you handle personal data? Founder names, titles and bios are personal data, so settle storage and purpose first.

Frequently asked questions

How do I find the slug to look up a Y Combinator company?

Take it from the company page link, where it is the last part of the address, or use the Y Combinator companies search source, which returns the slug for every company it lists. The lookup also accepts the full link to the company page.

Does the Y Combinator company lookup include funding rounds or valuation?

No. The response covers the directory profile: batch, status, team size, location, links, founders, news and Launch YC posts. Funding rounds and valuation are not part of the directory page, so they need a different source.

What does a Y Combinator lookup cost?

A company lookup costs a flat 5 points, which is $5 per 1,000 lookups. Directory and founder searches cost 1 point per returned record, which is $1 per 1,000 records. A lookup that finds nothing is not billed, and points come out of your monthly plan allowance.

Can I use this through MCP as well as REST?

Yes. The source is available over REST and as an MCP tool named run_yc_companies on the hosted MCP server, with the same arguments. The REST request works the same way for every source.

How current is the Y Combinator data an agent gets back?

The directory is read when you make the call, so the answer is the directory as it stands that day. Status, team size and news are as the directory lists them, and the directory can lag the company, so confirm anything a decision depends on with a second source.

Do the founder records include email addresses or phone numbers?

No. Founder records carry a name, title, bio, company, batch, industries, region and the public profile links the directory shows. They contain no contact details, and you should handle any personal data in line with your own obligations.