What is an MCP server for B2B data (and why sales teams want one)?
An MCP server for B2B data lets AI agents like Claude and ChatGPT search and enrich people and company records inside a chat. Here is how it works and why.
Sales teams already use AI assistants to write emails, summarise calls and research accounts. The missing piece has been reliable contact and company data inside that same conversation. An MCP server for B2B data fills that gap: it gives the agent tools to find and enrich prospects, so you can go from "who should we call?" to a usable list without leaving the chat.
What is an MCP server?
MCP, the Model Context Protocol, is an open standard that lets AI clients call external tools. An MCP server publishes a set of tools with clear inputs and outputs. When you connect it, your assistant can decide when to use those tools and pass them structured arguments.
Clients that support MCP include Claude, ChatGPT, Claude Code, Cursor and VS Code, along with many other agent frameworks. You connect once, and the tools become available in every conversation where the connector is turned on.
What does an MCP server for B2B data do?
A B2B data MCP server exposes people and company records as tools. With Enricho, there are two main tools:
- person_search finds, looks up and enriches people.
- company_search finds, looks up and enriches companies.
There are also account tools: account_usage to check credits, account_upgrade and account_buy_credits to get a secure checkout link, and account_billing_portal to manage billing.
The agent translates your request into filters. Filters take lists, which match any value within a list and combine with AND across filters. For example:
- Job title, seniority (founder, owner, c_suite, vp, head, director), role and skills
- Industry and company size bands such as 11-50, 51-200 or 201-500
- Country and city
- Contact requirements: has email, has phone, has LinkedIn
- Exclusions to leave out contacts or companies you already have
Full records can include name, job title, seniority, role, company, industry, company size, location, skills, emails, phone numbers where available, and LinkedIn and other social profiles. Company records include website, address and size. An empty field means the value is not on file.
How is this different from a traditional prospect database?
Traditional prospect databases are built around a search screen. You learn the filters, build a list, export it and then move it somewhere else. An MCP server moves that work into the agent you already use.
| Traditional prospect database | MCP server for B2B data | |
|---|---|---|
| Where you search | A dedicated web app | Your AI client (Claude, ChatGPT, Cursor and others) |
| How you search | Manual filters and saved lists | Plain-language requests the agent turns into filters |
| Reviewing results | Scroll and export | The agent shows a sample and summarises it |
| Next steps | Copy into other tools | Draft outreach, score fit or write code in the same chat |
| Cost control | Varies by vendor | Free preview, then you approve the spend |
The practical difference is context. The agent already knows your product, your ideal customer and the conversation so far, so it can choose sensible filters and explain its assumptions.
Why do sales teams want one?
There are four common reasons.
- Speed. A single request replaces several rounds of filter clicking and exporting.
- Context. The agent can research an account, then find the right people there, then draft a first message, all in one thread.
- Control over spend. With Enricho, previews are free and the agent asks before spending credits on full records.
- Developer access. Engineers can use the same tools from Claude Code, Cursor or VS Code, or call the REST API from scripts and headless agents.
What does a request look like?
Here is a typical first prompt. Mentioning Enricho by name the first time helps the assistant pick the right tools.
Use Enricho to find heads of sales at software companies in Sydney
with 51-200 employees. Only people with an email.
The agent runs a free preview with filters like seniority "head", job title "sales", industry "software", city "Sydney", size band "51-200" and has_email true. It shows you the targeting and a sample of up to 10 records without contact details, then asks whether to fetch full records.
You can follow up in plain language:
Leave out anyone at the companies in this list, then get 20 full records.
For more on this workflow, see how to build a B2B lead list in Claude.
How do you connect an MCP server for B2B data?
Every client uses the same server URL: https://mcp.enricho.co/mcp.
- In Claude or ChatGPT, add Enricho as a custom connector with that URL.
- Sign in with your email and the 6-digit code Enricho sends. There is no password.
- In Claude Code, Cursor, VS Code or another MCP client, add the server URL and sign in when prompted.
- For scripts and headless agents, create an API key and use the REST API.
Exact steps for each client are in the setup docs. New accounts start on the Free plan with 100 credits a month, and no card is needed.
How much does it cost?
Enricho uses credits. Previews, empty results and usage checks are free. Each full record returned costs 1 credit, and unmatched enrichment inputs cost nothing. Plans run from Free (100 credits a month) to Scale ($199 for 500,000 credits a month), and you can buy credit packs that never expire. See pricing or read how credit-based pricing works for AI agents.
What data is included, and how is it handled?
Enricho has 200+ million data points of professional and business information. It does not sell customer data or use it to train AI. GDPR and CCPA rights are honoured, and anyone can request removal of their data at /privacy/request, with requests answered within 30 days. You can read more about Enricho or contact the team with questions.