Startup & Funding Data API and MCP: Dealroom, PitchBook, Tracxn and 90 More

A startup funding data API turns "who raised what, from whom, and when" into a request that returns JSON instead of a page to read. NeuralVerge's Startup & Funding Data category has 93 sources from 18 providers, including Dealroom, PitchBook and Tracxn, each reachable over REST or MCP and priced in points. This guide maps the category, shows one real response, and covers cost, limits and how to choose. Crunchbase has its own Crunchbase alternative write-up; this post covers the rest.

The short version

• —What it is: 93 sources from 18 providers: funding databases (Dealroom, Tracxn, Owler), European directories (Tech.eu, EU-Startups), accelerator and launch platforms (Y Combinator, Product Hunt) and crowdfunding sites (Kickstarter, Seedrs, WeFunder). One request format covers all of them.
• —Who it is for: investors, sales and account researchers, analysts and agent builders who want public funding, ownership and launch data as JSON per lookup, not as a seat in a subscription.
• —What it costs: a lookup is 5 points, a search is 1 point per returned record, and Owler's company lookup is 10. A request that finds nothing is not billed. The Base plan is $50 for 50,000 points a month.
• —What it is not: public information only. The three PitchBook sources return the free public profile preview, not PitchBook's paid deal data.

What a startup funding data API covers

"Startup data" is not one dataset. It is a pile of overlapping databases that each answer a different question, and most confusion about which one to use comes from treating them as interchangeable. The useful split is by the question each source was built to answer.

Funding databases versus launch and community platforms

Funding databases such as Dealroom, PitchBook and Tracxn are organised around the company and its financing: tier, sector, headquarters, investors by round, valuation figures, ownership, and the list of rounds with dates and amounts. They are the right place to ask "who invested" and "how much has this company raised".

Launch and community platforms such as Product Hunt, BetaList, Indie Hackers and There's An AI For That are organised around the product: what launched, when, who made it, what people said in the comments and what it is compared against. They say little about money, but they are often the earliest public trace of a company that no funding database lists yet.

Accelerator and investor directories

Y Combinator, EU-Startups, Tech.eu and Wellfound sit between the two. Y Combinator lists companies and founders by batch, industry and region. EU-Startups and Tech.eu focus on Europe. Wellfound adds jobs and locations. Reach for these when the question is regional or about a cohort rather than one company.

Crowdfunding and equity campaigns

Kickstarter, Indiegogo, Seedrs (Republic Europe) and WeFunder publish campaigns with goals, pledged amounts, backers or investor counts and, for the equity platforms, valuation terms. They answer a different question again: how much did the public put in, and on what terms.

Data API versus data subscription

A subscription to a funding database gives you a user interface, an export quota and a contract. A startup funding data API gives you the same kind of record as JSON, billed per lookup. The trade runs both ways: a subscription can include sections that are not public, while a pay-per-lookup source returns what the platform shows publicly at the moment you ask. The AI research product is a different tool again, for open-ended questions where you do not yet know which source holds the answer.

How a request works

Whichever platform is behind a source, the mechanics are the same, so you learn one request format and reuse it across all 93.

You name the record or describe the search. A lookup takes an id, slug or profile link: a Tracxn profile link, a PitchBook profile id, a Seedrs campaign slug. A search takes a keyword, optional filters and a max_items cap. The source pages list the exact parameters and the reference lives at docs.neuralverge.ai, so this post does not restate them.

The platform is read at request time. The response reflects the platform as it stands when you ask, so you do not inherit a stale snapshot. The limit is that a platform can only show what it has recorded, and funding platforms record rounds some time after they happen.

The result comes back in one envelope. Every source returns the same outer shape: a session id you can trace, a readable summary in human, the structured result in machine, and the points charged in total_points. Inside machine, searches return an items array and a total_found count when the platform reports one; lookups return the record's own sections.

You pay in points. A lookup of one record costs 5 points. A search costs 1 point per returned record. A point is a tenth of a cent on the per-1,000 prices shown on the source pages, and points come out of the monthly allowance of your plan.

Every provider in the category

The tables below are built from the source pages, counted at the time of writing: 93 entries across 18 providers. Lookups and references are 5 points per successful run, except the Owler company lookup at 10. Searches and lists are 1 point per returned record, and a request that returns nothing is not billed. The three PitchBook entries are the free public profile preview only, not PitchBook's paid deal data.

Startup and funding databases

European and regional directories

Accelerators, launches and communities

Crowdfunding and equity campaigns

Notes on the larger providers

• —Tracxn company funding returns every round of one company with date, amount, round name and investors. A profile link that finds nothing is free.
• —Dealroom company and investor lookups return ownership, investors by round and funding events, and on the investor side portfolio, exits and similar investors.
• —Tech.eu funding round search returns European rounds with the amount in EUR next to the original currency, stage, investors and a confidence label per row, filterable by country, stage, sector, minimum amount and date. Tech.eu exits return acquisitions with the acquirer and, where known, the value.
• —Owler returns an estimated profile with revenue and employee estimates, rounds, competitors and news. At 10 points the company lookup is the priciest single record here, so use it for the companies that matter.
• —Y Combinator companies search filters by batch, industry, region, team size or hiring status; founders have their own search.

The PitchBook sources, and what they are

PitchBook has three sources: a company profile preview (5 points), an investor profile preview (5 points) and a profile search by keyword (1 point per record). The word "preview" is deliberate. These return the free public preview of a profile: description, website, founding year, status, employees, industries, parent company and address. Valuation and funding sections are paywalled by PitchBook and come back with empty tables. If you came here expecting PitchBook's deal data, read this section twice: the sources identify and describe an entity, and they do not replace the subscription.

One real request: searching Dealroom

To keep this concrete, here is one request and the response that came back, taken from the Dealroom company search page. The request asks Dealroom for companies matching the keyword fintech, capped at two records.

curl -X POST https://api.neuralverge.ai/functions/v1/run-dealroom-companies-search \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"keyword":"fintech","max_items":2}'

This is the response, copied from the source page. The session_id is a placeholder in the page's sample.

{
  "session_id": "11111111-1111-1111-1111-111111111111",
  "kind": "dealroom_companies_search",
  "keyword": "fintech",
  "max_items": 2,
  "start_page": 1,
  "partial": false,
  "human": "# Dealroom — companies search\n\n- **Query:** keyword=fintech\n- **Results:** 2 of 7913 found\n- **Proxy traffic:** 17.4 KB",
  "machine": {
    "total_items": 2,
    "first_item": {
      "id": "berkshire-hathaway-corp-",
      "name": "Berkshire Hathaway",
      "url": "https://dealroom.co/companies/berkshire-hathaway-corp-/",
      "alias": "berkshire-hathaway-corp-",
      "tier": "Listed",
      "sector": "Fintech",
      "hq": "Omaha, United States",
      "country": "United States",
      "founded_year": 1839,
      "valuation": 1180000000000,
      "valuation_range": "$1.18T",
      "exit_note": "IPO 1983",
      "logo": "https://storage.googleapis.com/dealroom-images-production/2026/09/26/a1b40123091158d6c35609cf6ac1cfb5.png"
    },
    "items": [
      {
        "id": "berkshire-hathaway-corp-",
        "name": "Berkshire Hathaway",
        "url": "https://dealroom.co/companies/berkshire-hathaway-corp-/",
        "alias": "berkshire-hathaway-corp-",
        "tier": "Listed",
        "sector": "Fintech",
        "hq": "Omaha, United States",
        "country": "United States",
        "founded_year": 1839,
        "valuation": 1180000000000,
        "valuation_range": "$1.18T",
        "exit_note": "IPO 1983",
        "logo": "https://storage.googleapis.com/dealroom-images-production/2026/09/26/a1b40123091158d6c35609cf6ac1cfb5.png"
      },
      {
        "id": "jpmorgan",
        "name": "JP Morgan",
        "url": "https://dealroom.co/companies/jpmorgan/",
        "alias": "jpmorgan",
        "tier": "Listed",
        "sector": "Fintech",
        "hq": "New York City, United States",
        "country": "United States",
        "founded_year": 1799,
        "valuation": 883528000000,
        "valuation_range": "$884B",
        "exit_note": "IPO 1983",
        "logo": "https://storage.googleapis.com/dealroom-images-production/2023/01/19/9726f27cce6233556b25070db8642dd3.png"
      }
    ],
    "total_found": 7913,
    "page_count": 1,
    "blocked": false,
    "proxy_traffic_bytes": 17803
  },
  "total_points": 2
}

The cost is on the last line. total_points is 2: one point for each of the two returned records. On the source page that is $1 per 1,000 records, so this request cost a fifth of a cent.

total_found is 7913, while only two records came back, because the request set max_items to 2. That cap limits both the list and the bill. The count is also a cheap way to size a segment: the source page describes using total_found with a filter to see how many companies Dealroom lists for it.

The labels belong to Dealroom. The keyword fintech matched Dealroom's sector label, and the two top results are large listed companies, not early-stage startups. To get the startups you want, add a tier, sector or country filter, or sort by valuation or founding year. At least one of keyword, tier, sector or country is required.

This is a summary row. Investors, funding events and ownership live in the separate Dealroom company lookup, which costs 5 points per resolved record. Searching to find candidates and then looking up the few that matter is the cheap way to use the pair. The full field list for this search is on the Dealroom company search page.

REST and MCP

The request above is the REST call. Over MCP the same call is a tool named run_dealroom_companies_search with the same arguments, {"keyword": "fintech", "max_items": 2}. Every startup and funding source is available both ways: the REST path uses hyphens (run-dealroom-companies-search) and the MCP tool name uses underscores. The tools live on the hosted MCP server at https://mcp.neuralverge.ai/mcp, so an agent can pick a source on its own, choose between a search and a lookup, and pay the same points per call. The MCP setup tutorial covers connecting a client.

A worked example

Consider a hypothetical analyst at a venture fund in Helsinki who is tracking Acme Oy, an illustrative Finnish startup. The example is invented to show a workflow, not to describe a real company.

She starts with a search. A Tech.eu funding round search filtered to Finland and a recent date range tells her whether any rounds involving companies like Acme Oy have been recorded lately, and which investors were in them. The cost is 1 point per returned record, and max_items keeps it bounded.

Then she looks up the company itself. A Dealroom company lookup, at 5 points, adds investors by round and funding events, and a Tracxn company funding lookup gives a second view of the same history to compare.

She notices the two platforms disagree on one round. That is the useful outcome: it tells her which figure to verify with the company before it goes in a memo. Finally, a Product Hunt product search shows whether Acme Oy ever launched there, and a Wellfound lookup shows whether it is hiring. Each call is the same request format with a different endpoint name.

Cost and limits

Pricing is in points, and the plans on the pricing page are monthly allowances. At the time of writing the Base plan is $50 for 50,000 points per month and the Core plan is $100 for 100,000, with larger plans above them. Check the pricing page for current terms.

From the source pages: a single-record lookup is 5 points ($5 per 1,000 lookups), a search is 1 point per returned record ($1 per 1,000 records, capped by max_items), Owler's company lookup is 10 points, and a request that finds nothing is not billed. The Dealroom search above is the working example: two records, two points. The limits are as important as the prices, and several are specific to this category.

• —Public information only. Each source returns what the platform shows publicly. Sections behind a login or a paywall, such as PitchBook's valuation and funding tables, come back empty.
• —Labels are the platform's. Tier, sector, stage and valuation ranges are each platform's own classification, returned as given. Two platforms can label the same company differently.
• —Fields vary by company. Several fields come back empty or null for a given record. Parse defensively.
• —Funding data trails events. A platform lists a round after it has recorded it. If the question is "did this company raise last week", an absent round is not proof that none happened.
• —Estimates are estimates. Revenue and headcount on Owler and Dealroom are ranges or estimates, not financial statements.

How to decide

Before you commit to a source or a mix of sources, check these:

1. —What question does it answer? Funding, product launch, region or crowdfunding. Pick the platform built for that question.
2. —Is the section you need public? Open the source page and read its known gaps. If the field you need is paywalled, a pay-per-lookup source will not return it.
3. —How is it billed? Per resolved lookup or per returned record, and whether misses are free.
4. —Can you cap spend? Look for max_items on searches, and budget lookups at 5 or 10 points each.
5. —What does an unresolved record look like? Test a company that does not exist and confirm your code handles the empty response.
6. —Does the response shape suit your parser? The real sample on each source page shows the exact fields. Check for nulls.
7. —Do two sources agree? Run the same company through two platforms before you rely on either.
8. —How much lag can you tolerate? Funding data trails events.

Where these sources fit in practice:

• —Investor sourcing. Search a region or sector on Dealroom, Tracxn or Tech.eu, shortlist, then look up the shortlist for investors and rounds.
• —Sales and account research. Use a recent round as a trigger, then enrich the account with headcount from Wellfound and competitors from Owler.
• —Competitive and market maps. Pull a sector from several platforms and compare, using Product Hunt alternatives and Tracxn sector data.
• —Due diligence screening. Check the public funding history and ownership of a counterparty before a deeper review; public sources are a first pass, not a substitute for a data room.

Frequently asked questions

Is the PitchBook data the full PitchBook database?

No. The PitchBook sources return the free public profile preview of a company or an investor. Sections such as valuation and funding are paywalled by PitchBook and come back with empty tables. The preview is useful for identifying an entity and reading its public description, but it is not a substitute for a PitchBook subscription.

Do I pay when a lookup finds nothing?

The source pages state that a request that finds nothing is not billed. Single-record lookups are charged only when the record resolves, and searches are charged per record actually returned, so a search with no matches costs nothing. Check the page of the source you use for the exact billing line.

Can an agent call these sources over MCP today?

Every startup and funding source is reachable over REST and as an MCP tool on the hosted MCP server. The tool name is the REST function name with underscores, for example run_dealroom_companies_search, and the arguments match the REST body.

Why would I use more than one startup database for the same company?

Each platform has its own coverage, labels and update rhythm, so the same company can look different in each. Reading two sources and comparing them surfaces disagreements and gaps that a single source would hide. The sources use the same request format, so adding a second one is a small change.

How fresh is the funding data?

Each record is read from the platform at request time, so the answer reflects the platform as it stands when you ask. That is not the same as the round being current: a platform only lists a round once it has recorded it, which can trail the real event.

Which source should I start with?

Start with the question you need answered. For who invested and in which rounds, use Dealroom, Tracxn or Tech.eu. For what launched, use Product Hunt or BetaList. For a region or a cohort, use Tech.eu, EU-Startups or Y Combinator. A cheap pattern is to search to find candidates, then look up only the few that matter.