How to use Cursor or Claude to find LinkedIn leads

Connect AI clients like Cursor or Claude to LinkedIn via MCP servers and REST APIs to find high-fit leads and draft personalized connection requests.

Get started

Short answer

You can find LinkedIn leads using Cursor or Claude by connecting them to a tool with a Model Context Protocol (MCP) server or a REST API. This lets the AI search LinkedIn based on your customer profile, score the fit, and draft personalized requests.

Define your ideal customer profile

AI clients can't find high-quality leads if you give them generic instructions. If you just ask for marketing managers, you'll get a list of people who might not have the budget or the specific problem your product solves. You'll waste your daily invite limits on people who will never buy.

You need a profile with specific markers. Include the exact job title, company size, industry, and the triggers that make them a buyer today. These triggers could be a recent job change, a specific post they wrote, or a certification they earned. This specificity allows the AI to distinguish between a casual user and a high-intent buyer.

The goal is to create rules the AI can use to score a profile. If a lead meets all must-have criteria and several nice-to-have signals, they get a high score. If they miss a must-have, they're discarded. This prevents your pipeline from filling up with low-quality contacts that look right on paper but don't fit your actual needs.

Consider the 'negative' profile as well. Tell the AI who to avoid. For example, if you sell a high-end service, you might want to exclude freelancers or people at companies with fewer than 10 employees. This saves the AI from processing irrelevant data and keeps your lead groups clean.

Vague profile

  • Find me tech founders in the US.
  • Look for people interested in AI.
  • Find marketing managers at B2B companies.

Specific profile

  • Founders of B2B SaaS companies with 10 to 50 employees who have posted about churn in the last 30 days.
  • Head of Demand Gen at logistics firms in Chicago with 200 to 500 employees.
  • VPs of Sales who recently started a new role and mention pipeline gaps in their profile summary.

Connect your AI client to a LinkedIn data source

Cursor and Claude can't browse LinkedIn in real time through a standard chat window because of platform restrictions. To get them to see lead data, you have to connect them to an external tool using the Model Context Protocol (MCP) or a REST API.

MCP lets the AI call specific functions, like searching for a profile or reading a comment, as if the AI had a built-in tool for that task. This means you don't have to manually copy and paste data from LinkedIn into the chat window, which is slow and prone to errors.

Once the connection is set up, the AI client can pull live data from LinkedIn and process it based on your instructions. You can ask the AI to check if a person is still at the company they listed three years ago or if they've recently changed their title.

If you are a developer, you can use a REST Agent API to build your own scripts. This allows you to automate the flow between finding a lead and adding them to a specific group for outreach. This is a more stable way to manage large volumes of leads than relying on a chat interface alone.

  1. 1

    Open AI client settings

    Go to the MCP settings in Cursor or the tool configuration menu in Claude.

  2. 2

    Add the MCP server

    Enter the server URL or install the specific MCP package provided by your lead generation tool.

  3. 3

    Authenticate the connection

    Use an API key or OAuth flow to grant the AI permission to execute search and retrieval functions.

  4. 4

    Test the connection

    Ask the AI to perform a simple task, like finding one profile based on a name, to ensure the server is responding.

Draft personalized connection requests

Generic messages get ignored. To get a high acceptance rate, the AI must use a specific observation from the lead's profile or a recent post they wrote. If the message looks like a template, the user will likely delete it without reading.

The AI should focus on why that person became a lead. If they were found because they commented on a competitor's post, the message should reference that specific conversation. This shows you are paying attention to their actual thoughts, not just their job title.

Keep the request short. LinkedIn connection notes have a strict character limit. Avoid sales pitches in the first message and focus on a shared interest or a specific observation. The goal of the connection request is to get the request accepted, not to close the deal.

Once the connection is accepted, you can move into a conversation. Use the AI to draft follow-ups that build on the initial connection. The AI can use your own sales playbook to ensure the tone remains consistent with your brand while still feeling personal.

Example

  1. Suggested reply

    Hi Sarah, I saw your comment on the post about pipeline gaps in logistics. I'm also seeing that trend in Chicago and would love to connect.

  2. Prospect

    Thanks for reaching out! Yes, it's been a struggle lately. Happy to connect.

  3. You

    Glad to be connected, Sarah. Out of curiosity, are you using any specific tools to track those gaps or just spreadsheets?

Avoid words like 'synergy', 'game-changing', or 'revolutionary' in your drafts. They are immediate markers of AI-generated spam.

Manage the outreach process

Automating the search is only half the battle. You need a system to manage the actual sending of messages. If you send too many requests too quickly, LinkedIn will flag your account. You must operate at a human pace.

A human pace means spacing out actions. Sending one action every few minutes is safer than sending fifty in a single second. You should also set a send window that matches the lead's time zone so you aren't sending messages at 3 AM their time.

Handle replies based on the intent. Some people will ask for a price immediately, while others will ask about your experience. You can let an AI handle the initial qualifying questions until there is real interest, then jump in to book the meeting yourself.

Keep track of your pending invitations. If you have too many outstanding requests that haven't been accepted, your account health drops. It is a good practice to withdraw old, stale invitations every few weeks to keep your pool healthy.

  1. 1

    Set daily limits

    Decide on a maximum number of invites and messages to send per day to avoid account restrictions.

  2. 2

    Define the send window

    Set your outreach to happen during business hours in the prospect's local time zone.

  3. 3

    Choose a reply mode

    Decide if the AI should book the meeting via a link or hand the conversation to you at the first reply.

  4. 4

    Monitor the inbox

    Check your unified inbox daily to prioritize interested leads and handle objections.

Doing this with Kinlead

Kinlead provides the MCP server and REST Agent API needed to let Claude, ChatGPT, and Cursor find leads and check replies. It finds people who match your description or website and scores them from 0 to 100, discarding any that fall below your fit line.

The system handles outreach from your own account via Unipile. It sends at a human pace, with at most one action every 5 minutes, and stays within the daily limits and send windows you set for each lead's time zone. Kinlead never stores your LinkedIn password.

You can choose how replies are handled. The AI can handle the conversation until a meeting is booked using your scheduling link, or it can hand the conversation to you at the first reply. All messages land in a unified inbox where interested replies are prioritized.

The Steal Customers agent specifically watches competitor company pages and profiles, reading comments to turn active commenters into leads. Each lead keeps the post URL and the content of their comment for personalized outreach.

  1. 1

    Connect your account

    Connect your LinkedIn account through Unipile in the settings.

  2. 2

    Configure the AI agent

    Set your goal, tone, and sales playbook to guide how the AI writes messages.

  3. 3

    Launch the campaign

    Point the campaign at a lead group and set your daily invite limits.

Frequently asked questions

Related

Get started with Kinlead

Prefer markdown? How to use Cursor or Claude to find LinkedIn leads.md