Twelve revenue data tools, plugged into Claude, Cursor, n8n or the agent you built yourself. Your teams describe what they need in plain English, the agent calls the tools, and you only pay for what is found.
Your teams already know how to write the prompt that segments, prioritizes and drafts. The chain always breaks at the same point: when it is time to fetch a fact about a real person, the model has nothing to work with.
With no source, a model outputs firstname.lastname@domain.com and lets you find out the bounce rate mid-campaign. A plausible pattern is not a verified address.
The cost: your sender reputation.A browser driven by the agent gets blocked within a few dozen pages, and job title, tenure and headcount are exactly the data that decides who you target.
The cost: the automation never gets past the demo.One provider for email, another for mobile, a third for scraping: three API keys, three invoices, and coverage that stays partial on each of them. How the provider cascade enriches your CRM →
The cost: three licenses for one use case.Each tool does one thing and returns a structured result, not a sentence. The agent reads their descriptions and picks which ones to call and in what order: no chain to wire up, no schema to write.
A first name, a last name and a company, and you get back the matching LinkedIn profile URL, with a confidence score when several people share the same name.
A company name or a domain, and you have the company page. It is the key that then unlocks headcount, industry and the right people to talk to.
The full profile: current role, tenure, experience, location, education. Enough to decide whether the person is in your target before you write a word.
The company profile: actual headcount, industry, headquarters, description, hiring pace. The context that tells a target account apart from a company with the same name.
Official legal data for companies registered in France: SIREN, SIRET, NAF industry code, legal form, officers, incorporation date. Pulled from the public French registry, 1 credit when found.
Compares what you know about a contact with what is published today, and flags both the departure and the arrival: previous company, new company, new title, estimated date. The buying signals you can draw from it →
The work email, verified before it is returned. Providers are queried one after another until the address is found, and you only see a single call.
The direct mobile number, in international format. It is the scarcest data on the market: the tool tells you plainly when it exists nowhere instead of handing you the switchboard.
An address on its own, and you know who is on the other end: identity, company, title, LinkedIn profile. For sequence replies, sign-ups and export rows where only an email is left.
A number on its own, and you know who called. For missed calls, half-filled forms and databases where the phone number outlived everything else.
Criteria instead of a list: job title, seniority, department, country, company size. You get back identities ready to go through the enrichment tools.
The same idea at account level: industry, headcount, country, tech stack, hiring pace. Enough to sketch an ICP in one sentence and refine it in the next.
The server is remote: nothing to run on your side, nothing to update. Your MCP client connects to it over HTTP with your key and discovers the twelve tools on its own.
It draws on your workspace credits. You can create one per tool or per team, and revoke one without breaking anything else.
From the command line:
Or in the configuration file:
Nothing left to wire up: the agent reads the tool descriptions, picks which ones to call and in what order, chains the results and hands you the list. Your teams write a sentence, not a pipeline.
"Give me 200 US SaaS companies with 50 to 200 employees, the VP of Sales at each one, their LinkedIn and their work email."
"This number called during my meeting and this address just replied to the sequence: who are these two people?"
"Out of my 3,000 contacts, pull the ones who changed companies this quarter and find their new email."
MCP calls use the same credits as your Jipup agents, billed on results. A profile that can't be found, a mobile that exists nowhere, an address nobody publishes: none of it is charged, so an agent that fumbles around won't drain your balance. The one exception is the job-change scan, billed per scan. A credit costs €0.0120 down to €0.0090 excl. VAT, depending on your plan.
Prices in credits, per result found (the job-change scan is per scan; a contact's LinkedIn URL is only searched for once). Unused credits roll over. See every plan and the full rate card →
Behind each tool, 15+ data providers are queried in order until something is found. No Kaspr, Lusha or Apollo license to pay for on your side.
What your agent calls over MCP is exactly what the Jipup workflows (Duplicate detection and merging, CRM enrichment, Job change tracking and New hire detection) write into your CRM. Same data, same cascade, same billing.
Each call is logged in your workspace: which tool, which input, which result, what it cost. An agent going off the rails shows up right away.
What you send to the server is used to answer your call, not to feed a shared dataset. Hosted in Europe, with contractually bound subprocessors.
The MCP server is included in your Jipup workspace: the key is available as soon as you create your account, and the credits are the same ones your CRM agents use, alongside the native integration with HubSpot and Salesforce (OAuth connection to your CRM →). Plans start at €120 a month (about $139) for 10,000 credits, with no commitment. Compare plans and pricing →