# Crustdata review (2026): live B2B data built for agents

By [Florian Martens](https://pipe0.com/authors/florian-martens) · Published Oct 6, 2026

Source: https://pipe0.com/reviews/crustdata
Scoring method: https://pipe0.com/reviews#methodology
Disclosure: pipe0 buys data from Crustdata and uses it in its waterfalls. The scores come from our own benchmark, not from that relationship.

> Crustdata is an API for company and people data that agents can call directly: headcount, hiring, funding, web traffic, and contacts, with live lookups when cached data isn't fresh enough.

**Verdict: 4.3/5.** The best company and people data API we have used for agents: broad, fresh, and fast, with one of the most complete MCP setups in the category. Opaque pricing and tight rate limits are the catch.

| Criterion | Score | Why |
| --- | --- | --- |
| Ease of use | 4.1/5 | An API, not an app. Through its MCP server, though, anyone with Claude or ChatGPT can query it in plain language in minutes. |
| Data quality | 4.5/5 | Deep company data (headcount, hiring, funding, web traffic, tech stack) with live lookups on request, and 78% mobile coverage in our benchmark. |
| Pricing value | 4.0/5 | Searches cost 0.03 credits a result and misses are cheap, but plan prices aren't published and contact data needs an Enterprise plan. |
| API and agent access | 4.7/5 | The most agent-ready data API we have used: MCP for every major client, a dry-run mode that spends nothing, watchers with webhooks, and agent-readable docs. Default rate limits are low. |
| **Overall** (mean) | **4.3/5** | |

**Best for:**

- Teams building agents or products on top of B2B data
- Account research that needs hiring, headcount, and funding signals
- Signal-based outbound that watches companies for changes

**Skip it if:**

- You want a prospecting app with a UI for reps
- You need thousands of lookups a minute on a self-serve plan
- You only need emails and phone numbers

**Price:** from $95/mo for the lowest self-serve tier, according to Crustdata's blog. No free plan.

## Crustdata at a glance

- **Founders:** Abhilash Chowdhary, Manmohit Grewal, and Chris Pisarski ([source](https://crustdata.com/about))
- **Headquarters:** San Francisco ([source](https://crustdata.com/about))
- **Accelerator:** Y Combinator, Fall 2024 ([source](https://www.ycombinator.com/companies/crustdata))
- **Funding:** $6M seed led by Y Combinator and A Capital, October 2025 ([source](https://crustdata.com/blog/crustdata-closes-6m-seed-round))
- **Dataset:** 1B+ people and 60M+ companies, according to Crustdata ([source](https://crustdata.com))
- **Freshness:** Refreshed every 14 to 28 days, with live lookups on request ([source](https://crustdata.com/about))
- **Signals:** Watchers for companies, people, jobs, and posts, delivered by webhook ([source](https://docs.crustdata.com))
- **MCP server:** OAuth or API key; Claude, ChatGPT, Cursor, VS Code, and Codex ([source](https://docs.crustdata.com/for-agents/mcp))

## Crustdata in our benchmark

pipe0 ran the same records through every provider it tests (August 2026).

| Test | Records | Coverage | Agreement | Median time | Run |
| --- | --- | --- | --- | --- | --- |
| LinkedIn URL to mobile (LinkedIn profiles) | 150 | 78% | – | 0.9 s | 20260804T202803Z-237d |

### How fast Crustdata answers

LinkedIn URL to mobile, median response time per lookup (run 20260804T202803Z-237d).

- Crustdata: 0.9 s
- Surfe: 4.4 s
- Amplemarket (premium): 18.6 s

- Of the three providers above 75% mobile coverage, Crustdata answered fastest: a median of 0.9 seconds, against 4.4 and 18.6 seconds for the other two.
- This benchmark measures contact data only. Crustdata's company and people data, which is what most teams buy it for, is not part of it.
- Samples are small, 150 records per test, so treat a few points of difference as noise.
- The LinkedIn profile set passed one vendor's resolvability filter, so absolute rates run higher than on a typical list. Comparisons between providers stay fair.

## What is Crustdata? [#what-is-crustdata]

Crustdata is a B2B data API, and it was designed for software to read rather
than for people to click. You give it a company or a person and get back a
structured record: for a company, headcount and how it is trending, funding
rounds, open jobs, web traffic and where it comes from, the tech stack,
news, reviews, and the leadership team; for a person, their profile, career,
and on Enterprise plans their email and phone number.

Abhilash Chowdhary, Manmohit Grewal, and Chris Pisarski built it in San
Francisco and went through Y Combinator in fall 2024. It raised a $6 million
seed in October 2025, and its customers include YC itself.

## Company data that stays current [#company-data-that-stays-current]

The difference between Crustdata and most data providers is what a record
contains and how old it is. A company lookup does not stop at industry and
size. It shows whether headcount grew over the last six months, which roles
the company is hiring for and where, and whether its web traffic jumped
after a launch. Those are the facts outbound actually runs on.

Its dataset refreshes every 14 to 28 days. When that is too old, live
endpoints fetch a person or company at request time. Watchers go one step
further: you register the companies or people you care about, and Crustdata
sends a webhook when something changes, such as a new job posting, a
funding round, or a leadership move. For signal-based outbound, that
replaces a pile of scrapers.

## Contact data [#contact-data]

Crustdata also sells emails and phone numbers, through a contact endpoint on
Enterprise plans. We benchmarked it: across 150 LinkedIn profiles, Crustdata
found a mobile number for 78%, within two points of the best single premium
source we tested. It was also the fastest of the high-coverage providers,
with a median response of 0.9 seconds against 4.4 and 18.6 seconds for the
other two. Speed like that matters when an agent is waiting on the answer.

## Built for agents [#built-for-agents]

Crustdata describes itself as a data provider for AI agents, and the tooling
backs that up. The MCP server installs in Claude, ChatGPT, Cursor, VS Code,
and Codex and signs in with OAuth. A dry-run mode shows what a call would
cost before it spends anything, which is exactly what you want when an agent
is deciding what to look up. The docs publish an `llms.txt` index, and your
account's own prices can be read from the API for free.

The limits are on throughput. By default, enrichment allows 15 requests a
minute and search 30, and contact data needs an Enterprise plan. Larger
jobs go through batch endpoints that run asynchronously. For an agent doing
account research, that is plenty; for bulk enrichment of a million rows, talk
to Crustdata first.

## Crustdata and pipe0 [#crustdata-and-pipe0]

We buy data from Crustdata. It runs in pipe0's default mobile and work email
waterfalls, and it powers pipe0's people and company searches, job search,
and company matching. We chose it for the freshness and depth of its company
data and for the speed of its contact lookups.

If you want Crustdata's data without a separate contract, pipe0 is one way
to get it: the same searches and enrichments sit next to 50+ other
providers, behind one API and MCP server, billed per result.

## Who should buy Crustdata [#who-should-buy-crustdata]

Buy Crustdata if you are building an agent or a product that needs current
company and people data, or if your outbound runs on signals like hiring,
headcount, and funding. It is the most agent-ready B2B data API we have
worked with.

Look elsewhere if your team wants a prospecting app with a UI, or if you
only need contact details. For mobile numbers at the highest coverage,
[Surfe](/reviews/surfe) led our benchmark.

## Crustdata pros and cons

### Pros

- **Company data with real depth.** A company record carries headcount and its trend, funding, job postings, web traffic and its sources, technographics, news, reviews, and the leadership team. Few APIs return that much from one call.
- **Fresh when it matters.** The standard dataset refreshes every 14 to 28 days, and live endpoints fetch a person or company on request when that isn't fresh enough.
- **Fast, high-coverage mobile numbers.** In our benchmark Crustdata found a mobile number for 78% of 150 LinkedIn profiles, within two points of the best single premium source, with a median response of 0.9 seconds.
- **Built for agents from the start.** Its MCP server works in Claude, ChatGPT, Cursor, VS Code, and Codex, signs in with OAuth, and has a dry-run mode that shows what a call would cost without spending credits.

### Cons

- **Pricing you have to ask about.** Crustdata publishes credit costs per endpoint but not what a credit costs. Its own blog puts the lowest tier at $95 a month.
- **Contact data is Enterprise-only.** Emails and phone numbers come from the contact enrichment endpoint, which self-serve keys can't call.
- **Tight rate limits.** Person and company enrichment allow 15 requests a minute and search 30, by default. Large jobs go through batch endpoints instead.
- **No app for reps.** There is no prospecting UI, Chrome extension, or sequencer. Sales teams reach it through an agent, a spreadsheet tool, or their own code.

## Crustdata pricing

| Plan | Price | What you get |
| --- | --- | --- |
| Self-serve API | From $95/mo | Credits for search, enrichment, web, and watcher endpoints. One-time or automatic top-ups by card. Free trial on request. |
| Enterprise | Custom | Contact enrichment (emails and phones), live endpoints, and higher rate limits. |
| Datasets | Custom | Flat files of company and people data, refreshed monthly. |

- People and company search cost 0.03 credits per result. A person enrichment costs 1 to 7 credits depending on what comes back, a company enrichment 2 to 4.
- Contact enrichment costs 1 credit per work email, 2 per personal email and 2 per phone, up to 5.5 per person with verification.
- Watchers that track companies or people for changes cost 0.5 to 5 credits per record on a 30-day cycle, or 150 on a one-day cycle.
- Your account's own prices are free to read from the API, and the MCP server can dry-run any call at zero cost.

Prices as of Oct 6, 2026, from https://docs.crustdata.com/general/pricing.

## Crustdata alternatives

- [pipe0](https://pipe0.com/enrichment-api): You want Crustdata's data alongside 50+ other providers through one API and MCP server, without a separate contract.
- [Surfe](https://pipe0.com/reviews/surfe): Contact coverage is the priority, especially mobile numbers, and your reps work from LinkedIn into a CRM.
- [Clay](https://pipe0.com/reviews/clay): You want company research and enrichment in a table people can edit, and someone on the team will build it.
- [Apollo](https://pipe0.com/reviews/apollo): You want a prospecting app with sequences rather than an API.

## FAQ

### What is Crustdata?

Crustdata is a B2B data API for company and people data, built for AI agents and products. It returns firmographics, headcount trends, hiring, funding, web traffic, and contact details, with live lookups when cached data isn't fresh enough.

### How much does Crustdata cost?

Crustdata sells credits. Searches cost 0.03 credits per result, person enrichment 1 to 7 credits, and company enrichment 2 to 4. It doesn't publish what a credit costs; its blog puts the lowest tier at $95 a month. Contact data needs an Enterprise plan.

### Does Crustdata have email and phone data?

Yes, through its contact enrichment endpoint on Enterprise plans. In our benchmark it found a mobile number for 78% of 150 LinkedIn profiles, with a median response of 0.9 seconds.

### Does Crustdata have an MCP server?

Yes. It works in Claude, ChatGPT, Cursor, VS Code, and Codex, signs in with OAuth or an API key, and bills at normal credit prices. A dry-run mode shows what a call would cost without spending credits.

### How fresh is Crustdata's data?

Crustdata refreshes its dataset every 14 to 28 days. Its live endpoints fetch a person or company on request, for cases where that isn't recent enough.

### What are Crustdata's rate limits?

By default, person and company enrichment allow 15 requests a minute, search 30, and live endpoints 10. Batch endpoints handle larger jobs asynchronously.