Company Intelligence & Reviews API and MCP: Gartner, Capterra and Trustpilot in One Call

A company intelligence API returns what is publicly known about a company as structured data a program can use: what customers say about it, how it ranks against competitors and how much it has raised. NeuralVerge covers this with Gartner, Capterra, Trustpilot and Crunchbase lookups, reached over REST or MCP and billed in points. "One call" in the title means one account and one response shape, not one request.

The short version

• —What you get: four sources, Gartner Peer Insights, Capterra, Trustpilot and Crunchbase, each read as structured JSON. They are separate lookups with their own input and price, sharing one account, one authentication scheme, one point balance and one response envelope.
• —Who it is for: analysts, sales and procurement teams and agents looking at companies that are not their own, who want reputation and funding signals next to firmographics.
• —What it costs: searches and review lookups cost 1 point per returned record, profile lookups 5 points, Crunchbase 15 points. The full table is below, and the source catalog holds the reference for every lookup.
• —How to reach it: every lookup is available over REST and as an MCP tool, and the response tells you what it cost.

What a company intelligence API is

The platform is the product: you ask a question about a company and get structured evidence back. REST and MCP are two ways to ask. Three contrasts mark the boundary.

Versus firmographic enrichment

Firmographic enrichment answers "what is this company?": employee band, industry, headquarters, domain. It is the backbone of lead scoring and CRM hygiene, and the posts on data enrichment and the registry sources cover that side. Company intelligence is the other half: the reputation and market signals that firmographics do not carry. Two companies can look identical on paper and differ in review profile, category ranking and funding.

Versus a review-monitoring tool

Review-monitoring products watch one company's reviews and alert its owner. A company intelligence API is built for the reader on the other side: the analyst, sales team, procurement lead or agent looking at many companies that are not their own. The unit of work is "read this company's profile", not "watch my brand".

Versus reading the sites by hand

Opening each platform and reading works for one vendor before a call and does not survive a list of fifty. Pulling a few hundred reviews as data lets you count, group and search them.

The four sources

The full source catalog lists every lookup. They split into three review and directory platforms and one funding profile, and each platform has a different population of reviewers.

Gartner Peer Insights

Gartner Peer Insights reviews come from IT buyers reviewing enterprise software for other IT buyers. That skews the population toward larger, more considered purchases than a broad catalog of small-business tools. The sources cover the full hierarchy: a vendor is the company, a product is one thing it sells, and a market is the category the product competes in. There are seven lookups, from a keyword search over products or markets, through product, vendor and market profiles, to the ranked products inside a market and the full category taxonomy.

Capterra

Capterra is a software directory where buyers compare vendors in a category. The sources let you search by keyword, list the products in a category, pull a product's profile (vendor, pricing, ratings, features) and read its reviews with pros, cons, sub-ratings and author details. Capterra review text is where the actual wording of software buyers lives, which makes it the most useful of the three for voice-of-customer analysis.

Trustpilot

Trustpilot is a consumer-facing review platform, and a business profile there carries signals beyond the star rating: the rating distribution, whether the business has claimed its profile, and how it replies to reviews. The profile lookup accepts a plain business domain, a Trustpilot URL or the company's own website URL, so you do not have to find the exact page first. Separate sources search businesses by keyword and return a business's reviews.

Crunchbase

The Crunchbase source takes a Crunchbase company URL and returns the profile as structured data: name, website, location, founding year, employee band, industries, total funding raised and the company description. Total funding is usually the field a go-to-market team is after, as a public proxy for whether a company can afford what you sell. The Crunchbase API alternative post covers this source in more depth; here it is the funding leg of the picture.

How a lookup works

1. You send an identifier or a keyword

Searches take a keyword. Profiles take an alias, an ID, a URL or, for Trustpilot, a domain.

2. The lookup runs inside the request

These are synchronous: the lookup runs while you wait and the result comes back in the same response, with no job to poll. If a lookup hits its time budget before finishing, the response is flagged "partial": true and carries whatever was collected so far. Re-running with a smaller max_items gives a complete result.

3. The response comes back in one envelope

Every response has the same top level: a session_id you can trace, a kind naming the source, the inputs echoed back, a short human summary string, a machine object with the structured data, and the total_points charged. A pipeline can branch on kind and read machine, regardless of which source answered.

4. You are billed in points

Billing is per record returned or per resolved profile, as the cost table below shows.

5. You join the results

The last step is yours: match on a domain, a name or an alias and put the review evidence next to the funding figure. When you only have a company name, a search source or the AI research pipeline can resolve the identifier first.

A worked example: sizing up a vendor for Acme Oy

Imagine Acme Oy, a Finnish company, evaluating a handful of vendors in a software category. This illustrates the order of lookups and is not data about any real company.

1. —A Gartner markets search or a Capterra products search finds the category and the candidate products. This is the discovery step, billed per result returned.
2. —For each candidate, the Gartner product or Capterra product lookup returns the vendor, ratings and profile.
3. —A Trustpilot profile lookup on the vendor's domain shows whether the company has claimed its profile and how it handles replies, which says something about how it treats customers after the sale.
4. —Capterra reviews, limited to a sensible max_items, give the pros and cons in buyers' own words.
5. —A Crunchbase profile lookup, where the vendor has one, shows what it has raised, a useful input when the question is whether the vendor will be around in three years.

A vendor with no Trustpilot listing simply produces no record for step 3. Per the billing lines on the source pages, profile lookups are charged only when the profile resolves.

REST and MCP: one real request

The clearest way to see the shape is a small, real request. This one uses Capterra products search, the discovery step in most review workflows, since it gives you the product IDs and URLs the deeper lookups need. Every source in this post is reachable both ways: the REST path uses hyphens, the MCP tool uses underscores, and both return the same response. AI research is the one exception across the platform, as it runs over REST only. For connecting an MCP client, see the MCP server guide.

The request over REST

curl -X POST https://api.neuralverge.ai/functions/v1/run-capterra-products-search \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "query": "crm software", "max_items": 2 }'

On the source page the example uses "max_items": 25; here it is lowered to 2 so the call matches the captured response below. query is required and max_items accepts 1 to 250, with a default of 25.

The same request over MCP

The hosted server at https://mcp.neuralverge.ai/mcp exposes the lookup as the tool run_capterra_products_search:

{ "tool": "run_capterra_products_search", "arguments": { "query": "crm software", "max_items": 25 } }

The response

As captured and published on the source page:

{
  "session_id": "11111111-1111-1111-1111-111111111111",
  "kind": "capterra_products_search",
  "query": "crm software",
  "max_items": 2,
  "partial": false,
  "human": "# Capterra products search\n\n- **Query:** crm software\n- **Max items:** 2\n- **Records returned:** 2\n- **Response size:** 428.0 KB",
  "machine": {
    "total_items": 2,
    "items": [
      {
        "@type": "CapterraSearchProduct",
        "id": "199936",
        "seo_id": "199936",
        "alias": "CRM-Software-App",
        "name": "CRM Software App",
        "url": "https://www.capterra.com/p/199936/CRM-Software-App/",
        "image": "https://gdm-catalog-fmapi-prod.imgix.net/ProductLogo/7a92c169-ec9e-42b7-b6f5-6e96326526f8.png",
        "short_description": "CRM solution that assists businesses with task lists, collaboration, Gantt charts, timesheets, presentations, and more.",
        "rating": 4.5,
        "review_count": 4,
        "position": 1
      },
      {
        "@type": "CapterraSearchProduct",
        "id": "10091063",
        "seo_id": "10091063",
        "alias": "Ultimatix-Payroll-Software-Human-Resource-and-payroll-software",
        "name": "Ultimatix Payroll Software  - Human Resource and payroll software",
        "url": "https://www.capterra.com/p/10091063/Ultimatix-Payroll-Software-Human-Resource-and-payroll-software/",
        "image": null,
        "short_description": null,
        "rating": null,
        "review_count": 0,
        "position": 2
      }
    ],
    "response_bytes": 438234
  },
  "total_points": 2
}

Reading the response

• —The cost is in the response. Two records came back and total_points is 2. Nothing is estimated after the fact.
• —Fields are nullable on purpose. The second product has no rating, image or description, so rating is null. Do not assume a number.
• —The ordering is Capterra's. position is the rank in this search, which reflects Capterra's own algorithm. The first result being a small product with four reviews is a reminder that rank here is not a market-share ranking, and the second result matching a "crm software" search while being a payroll product shows that a keyword can pull in loosely related products.
• —The id and url are the handles. They are what you pass to the Capterra product lookup for the full profile or to the Capterra reviews lookup for what buyers wrote.

The full parameter list and the other response shapes are in the documentation. They are not restated here, because they are the part most likely to change.

Cost and limits

Pricing is by point, and the numbers below come from the source pages. Check the pricing page for current plans.

The table lists all 17 company-intelligence entries from the source pages, counted at the time of writing. On the source pages, 1 point is stated as $1 per 1,000, 5 points as $5 per 1,000 and 15 points as $15 per 1,000.

Two budgeting habits follow from that table. First, the cheap per-record lookups can still add up when a request returns the maximum, so set max_items deliberately instead of leaving it at the default. Second, the profile lookups cost more per call than a search result because they return a whole structured profile, so run searches first and profile lookups only for the shortlisted companies.

Limits to plan for

• —Reviews are self-selected. People who feel strongly are more likely to write one, so extremes are over-represented. A rating is a snapshot of who wrote, not a measurement of every customer.
• —Platforms differ. Gartner Peer Insights reflects IT buyers on enterprise purchases. Capterra is a software directory. Trustpilot is a consumer review site. The same company can read differently on each, and that is information, not a bug.
• —Coverage is uneven. A small or new company may have no listing on a given platform, in which case the lookup returns nothing and, for profile lookups, you are not charged. No coverage figure is published for any of these platforms, because none has been measured in a way we can point to.
• —Funding is a floor. A profile reflects rounds once they have been announced, so total funding is not a guaranteed complete picture.
• —Snapshots, not feeds. Lookups read the profile at request time. For change over time, store each response and compare.

How to decide

Work through these before you build on it:

1. —Does the identifier you have match what the source takes? Crunchbase wants a company URL, Gartner wants an alias or URL, Trustpilot accepts a domain. If you only have names, plan a resolution step.
2. —Which platform's reviewers match your question? An enterprise IT question points at Gartner, a software-buyer question at Capterra, a customer-treatment question at Trustpilot.
3. —What happens on a miss? Confirm in the source page's billing line that a lookup with no result is not charged, and treat "no record" as a valid answer.
4. —Have you set max_items? The default is a convenience, not a budget.
5. —Do you handle partial: true? Decide whether to retry with a smaller max_items or accept the partial set.
6. —Do you store snapshots? If trends matter, save each response with its session_id and date.
7. —Will you show reviews as opinion, not fact? If the output goes to a human decision-maker, label ratings as what reviewers wrote on a given platform.

The jobs it fits best:

• —Vendor due diligence. Procurement and security reviewers compare candidates' ratings, how they respond to reviews and how much they have raised before a contract.
• —Competitive analysis. Product marketers track how a competitor ranks in its Gartner market or Capterra category and read the language its reviewers use.
• —Voice of customer. Researchers pull a few hundred reviews for a product and analyze the pros, cons and sub-ratings, rather than skimming the first page.
• —Sales qualification. Pairing a funding profile with review signals separates a funded, well-reviewed prospect from one that is neither.

Frequently asked questions

Is this really one call for all four sources?

No, and it is worth being precise about it. Each source is its own lookup with its own input and its own price: a Capterra search takes a keyword, a Trustpilot profile takes a domain, a Gartner product takes an alias, and a Crunchbase profile takes a company URL. What is shared is the account, the authentication, the billing in points and the response envelope, so a pipeline that reads all four is writing one integration, not four.

What does a company intelligence API return that a company database does not?

Opinion and momentum. A firmographic database tells you what a company is: size, location, industry. Review platforms tell you what buyers and customers say about it, how the company responds, and how its products rank against peers, and a funding profile tells you what it has raised. These are separate signals, and the useful part is reading them side by side for the same company.

Are the review sources available over MCP?

Yes. Each source is available over REST and as an MCP tool, for example run_capterra_products_search, so an MCP client can run the same lookups.

What happens if a company has no Trustpilot or Capterra listing?

You are not charged. Profile lookups are billed only when a profile resolves, and search and review lookups are billed per record returned, so a lookup that finds nothing costs nothing. The practical consequence is a coverage gap you should plan for, since small or new companies often have no listing on a given platform.

Can I treat review ratings as a measure of product quality?

Treat them as a snapshot of what reviewers chose to write, not a verdict. Reviews are self-selected, so strong opinions in both directions are over-represented, and each platform has its own reviewer population. A rating from IT buyers on one platform and a rating from small-business buyers on another can differ for the same product without either being wrong.

How do I get from a company name to a Crunchbase or Gartner lookup?

Those lookups take an identifier, such as a Crunchbase company URL or a Gartner alias, not a free-text name. When you only have a name, use a search step first, either the Capterra, Trustpilot or Gartner search sources or AI research, to resolve the right identifier, then run the profile lookup.