How to Build a LinkedIn Lead List with Work Emails in Claude or ChatGPT

Short answer: add a LinkedIn MCP server, such as LinkMCP, to Claude or ChatGPT. Then paste LinkedIn profile links and describe a good lead. The assistant reads each profile and company and rates the fit. It also finds each person's work email, with a status that says if the address can receive mail.

This job works in the free trial. It reads profiles and companies from their links and looks up emails, so you do not connect your own LinkedIn account.

We make LinkMCP.

Step by step

1. Add LinkMCP to your AI app. Claude: Is there a LinkedIn connector for Claude?. ChatGPT: How to Connect ChatGPT to LinkedIn. When the app asks, sign in to LinkMCP. A new account starts a 7-day free trial, and you do not need a card.
2. Paste the profile links and describe a good lead: what you sell, and which job titles, company sizes and regions fit.
3. The assistant reads the profiles with linkedin_bulk_get_profiles, up to 10 in one call. It gets each person's name, headline, location and current job.
4. It reads the companies with linkedin_bulk_get_companies, also up to 10 in one call: the website, the size range, the headquarters and the tagline.
5. It rates each person High, Medium or Low against your description, with one reason from the data. In our test, the reasons were things like a company above the size limit or a role outside sales.
6. It finds the work emails with find_email. The tool takes the first name, the last name and the company's website domain. Each email comes back with a status: valid, catch-all, unknown or invalid. When the tool finds no email, it says so.
7. You get one table, best fit first. Every tool in this job only reads, so nothing is sent to anyone.

To check an email that you already have, ask the assistant to use validate_email. It returns the same statuses and the mail provider of the domain.

Copy this prompt

Here are LinkedIn profile links of people I want to contact: [paste the links]

We sell [what you sell] to [who buys it]. A good lead is [job titles, company size, region].

Use LinkMCP. For each person, read the profile and the company, rate the fit (High, Medium or Low) with one reason from the data, and find the work email. Then give me one table, best fit first: name, title, company, company size, location, fit, reason, work email, email status, profile link. Do not send anything.

After the table, you can ask for the same table as CSV, or for a first message to one person as a draft.

If the assistant answers without LinkedIn data, name the tool in your prompt, for example: "Use the LinkMCP tool linkedin_bulk_get_profiles."

What happened when we ran it

We ran this prompt on 2026-10-23 in Claude Code (Claude Sonnet 5.5) with LinkMCP connected. We pasted the profile links of 5 well-known tech executives. A good lead was "a founder, CEO or head of sales at a B2B software company with 50 to 5,000 employees in North America".

• The assistant made 7 LinkMCP calls: 1 profile call for all 5 people, 1 company call for their 5 companies and 5 email lookups.
• It found all 5 profiles and all 5 companies. It found a work email for each person, and all 5 had the status "valid".
• It rated 2 people Medium and 3 Low, each with a reason. Nobody was a High fit. Each person missed a part of the description: the company size, the type of company or the role.
• The run took under a minute and sent nothing.

In a second run, the assistant first guessed a company's domain, and the lookup found no email. With the website from the company's LinkedIn page, the next lookup found it.

Free trial and paid plans

• Works in the free trial: every step on this page. Profile, company and email lookups do not use your LinkedIn account.
• The trial has one lookup allowance for the whole trial. Each profile, company and email lookup uses part of it. When it is used up, lookups stop until you choose a plan. Paid plans have a monthly lookup allowance.
• Needs a paid plan with your LinkedIn account connected: people search, mutual connections, connection requests and messages. To find the people with a search: How to Find Prospects on LinkedIn by Job Title and Company.

Limits

• One call reads up to 10 profiles or 10 companies. For a longer list, the assistant makes more calls.
• The email finder looks for the address at the company domain that it gets. Use the person's current company, and let the assistant read the company first, so that it has the real website.
• The finder does not have an email for every person.
• LinkMCP does not read Sales Navigator lead lists. Paste the profile links instead.

Where other tools fit better

• Very large lists. For thousands of rows a month, a contact data platform with its own database and file upload fits better. With LinkMCP, every profile, company and email is a lookup that your assistant makes in the chat.
• Leads that live in Sales Navigator. LinkMCP does not read or change lead lists. Copy the profile links, or run a saved Sales Navigator search on a paid plan: LinkedIn Sales Navigator MCP for Claude and ChatGPT.

FAQ

Related

• Find Work Emails and Mobile Numbers with AI (docs)
• How to Find Prospects on LinkedIn by Job Title and Company
• LinkedIn Candidate Search with AI: Claude, ChatGPT