🧠 LinkedIn Lead Generation with Claude and Exa.ai: A Step-by-Step Guide

TL;DR — Too Long Didn’t Read

Skip the guide and do this:

• Exa.ai is a semantic search engine built for AI — it understands the meaning behind a query, not just keywords, which makes it far more effective than LinkedIn's native search for ICP-based prospecting.

• Connecting Exa to Claude via MCP (Model Context Protocol) removes the copy-paste step. Search and analysis happen in the same session — you go from ICP to outreach-ready pipeline without switching tools.

• The anchor-based approach works better than broad industry searches: give Exa 3–5 existing customers and ask it to find 50 similar companies. It matches on structural patterns, not just keywords.

• The MCP session can run parallel searches — company discovery, decision-maker lookup, and buying signal detection simultaneously.

• Roles outperform names for durability: titles like "Head of CX" or "VP of AI" stay accurate longer than specific individuals whose LinkedIn data may be months out of date.

• Claude handles ranking and outreach angles in the same session — 50 raw companies becomes a top-10 prioritized list with one outreach hook per account.

• Installation takes one command for Claude Code users. Claude Desktop users need a config file edit and a restart.

1. What Is Exa.ai for LinkedIn Prospecting?

Most search tools return links. Exa returns structured intelligence.

Exa.ai is a search engine built specifically for AI applications — not for human browsing. It uses neural embeddings, the same underlying technology that powers large language models, to understand the meaning and intent behind a query rather than matching keywords. For LinkedIn lead generation, this distinction matters a lot: you can describe your ICP in plain English and get back companies, roles, and decision-makers that actually match the pattern — not just companies that mention certain words in their LinkedIn descriptions.

Traditional keyword search returns results based on what appears in text. Exa's semantic search returns results based on what something means. Ask for "companies similar to Walmart and Verizon" and it understands industry profile, size, complexity, and structural characteristics — not just names that appear alongside those keywords.

What this means for prospecting:

Exa raised $85M in Series B funding at a $700M valuation in September 2025, led by Benchmark with participation from Lightspeed and NVentures (NVIDIA's venture fund). It powers AI-native applications across thousands of companies, including Cursor and Notion AI.

2. Exa.ai vs Claude MCP: What’s the Difference?

This distinction is important because the same underlying search capability produces very different output depending on where you access it.

Exa's web interface (exa.ai and websets.exa.ai) is genuinely useful for prospecting. You can run natural language ICP queries and get real results. The bottleneck is what comes after: the responses are large and raw. You get a wall of text, then you have to copy it into another AI tool to clean it up, analyze it, build a table, or turn it into ranked pipeline. That extra step breaks the workflow and adds significant manual time.

Exa via MCP inside Claude removes that step entirely. When Exa is installed as an MCP connector inside Claude, retrieval and analysis happen in the same session. Claude gets the search results directly and immediately processes them — ranking accounts, identifying decision-makers, flagging buying signals, and building outreach hooks — without you doing anything in between.

The retrieval is identical. What changes is everything that comes after it.

There is also a practical cost advantage. Enrichment, structured output, deep research, and table-building are paid features inside Exa's web UI. When Exa is connected via MCP inside Claude, Claude handles the analysis layer — you get equivalent output without paying for Exa's premium tier on top of your Claude subscription.

3. How to Install Exa MCP in Claude (Step-by-Step)

Installation is simpler than most MCP setups. Exa provides a hosted MCP server, so you don't need to run anything locally.

You need:

• A Claude Desktop or Claude Code installation

• An Exa API key (available at exa.ai — new accounts get $10 in free credits, no credit card required)

Installing in Claude Code (one command)

Run this in your terminal to enable the full lead generation tool set:

claude mcp add --transport http exa \
"https://mcp.exa.ai/mcp?tools=web_search_exa,linkedin_search_exa,\
company_research_exa,deep_researcher_start,crawling_exa"

This connects Claude Code to Exa's hosted MCP server with all five lead generation tools enabled.

Here’s a quick overview of what that looks like:

Installing in Claude Desktop

Exa's documentation at docs.exa.ai/reference/exa-mcp generates a tailored install prompt based on the specific tool you're using. The general process:

1. Go to docs.exa.ai/reference/exa-mcp and select Claude Desktop

2. Copy the generated configuration JSON

3. Paste it into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or the equivalent path on Windows

4. Add your Exa API key where indicated

5. Restart Claude Desktop

The MCP endpoint is: https://mcp.exa.ai/mcp

By default, only web_search_exa and get_code_context_exa are enabled. For the full LinkedIn lead generation workflow, you need to explicitly enable the additional tools in your configuration (see the tool list in the next section).

This is where you get real AI workflows, prompts, and systems you can use to automate your work. If you're serious about using tools like Claude to grow your business, this is for you.

Already a paying subscriber? Sign In.