How to Find Leads in Claude

To find leads in Claude, you need to connect Claude to a tool that has access to contact data. Claude is an excellent research and reasoning tool — it can analyze a market, describe your ideal customer profile, and help you write outreach — but Anthropic does not give Claude native access to company databases, contact records, or email finders. That means the fastest path to actual leads inside a Claude conversation is to connect ToolRouter and let Claude call the right tool automatically: Lead Finder for prospecting companies and people at scale, Lead Planner for building structured outreach and ad strategies, and People Search for researching specific individuals. One connector, not five separate apps.

This query keeps coming up because Claude is genuinely better at lead research than most people realize. It can synthesize targeting logic, interpret unusual company descriptions, generate hypotheses about ideal customer profiles, and reason across multiple signals at once. What it cannot do by itself is look anyone up. The combination of Claude's reasoning and real database access is what makes this actually useful rather than just "ask Claude to think about your ICP."

According to HubSpot, salespeople spend an average of two working days every week on prospecting and research. According to McKinsey, sales teams using AI for lead research see 50% more leads and appointments at 60-70% less cost. According to Gartner, 77% of B2B buyers say their last purchase was very complex or difficult. The implication is clear: better upfront research on who you are reaching, what they care about, and what problems they are trying to solve is worth the time. Claude is reasonably good at that synthesis. Connected tools are where the actual data comes from.

What Claude Can and Cannot Do Natively for Lead Generation#

Claude's built-in capabilities for sales work are real but bounded:

• Claude can help you define an ideal customer profile from first principles.
• Claude can write cold outreach emails, LinkedIn messages, and follow-up sequences.
• Claude can analyze a company's website, product, and positioning if you paste in the content.
• Claude can build persona frameworks and prioritization logic.
• Claude cannot search a live database of companies or contacts.
• Claude cannot find or verify email addresses.
• Claude cannot look up a specific person's professional history, social profiles, or contact details.
• Claude cannot generate lead lists from filters like funding stage, headcount, geography, or job title.

The practical split looks like this:

If your goal is to think about sales strategy, Claude alone is often enough. If your goal is to produce actual leads with real contact information, you need database access. That is what the ToolRouter connection provides.

How to Connect ToolRouter and Start Finding Leads#

The setup is the same two-step process as any connector.

1. In Claude, go to Settings → Connectors → Add custom connector.
2. Enter:

• Name: ToolRouter
• URL: https://api.toolrouter.com/mcp

1. Start chatting normally.

Once connected, you describe the outcome you want rather than memorizing tool names:

• "Find 30 Series A SaaS companies in the UK that are hiring a Head of Sales."
• "Get me the LinkedIn and email for the Head of Partnerships at Notion."
• "Find companies similar to Figma but smaller — I want to prospect their marketing teams."
• "Build me a cold outreach plan for my consulting business, starting with a diagnosis of what kind of leads I need."

Claude picks the right tool, decides what to pass it, and handles multi-step workflows without you manually moving data between applications.

Which ToolRouter Lead Tool to Use#

The three tools handle different phases of the prospecting workflow. Most users end up using all three in sequence.

The practical workflow for most people is: Lead Planner to figure out who to target → Lead Finder to find and enrich contacts → People Search for individual deep-dives → Lead Planner for building the outreach copy and plan.

Four Lead Generation Workflows That Are Better Than Cold Lists#

The most valuable approaches are usually chained conversations, not single-shot searches.

1. ICP-First Company Prospecting#

Use this when you have a vague sense of your target customer but need actual companies.

Workflow: Lead Planner (diagnose) → Lead Finder (find_companies) → Lead Finder (enrich_company)

Example:

I sell a B2B SaaS product for construction project management. Revenue is $30k MRR. I mostly close deals via warm intro and inbound. Help me diagnose my lead gen bottleneck, then find 25 construction companies in the US with 50-200 employees that are growing and enrich each domain to get contacts.

This is where Claude's reasoning adds real value on top of the database. A filter-only tool returns results that match criteria. Claude interprets "growing construction companies" into the right query logic, diagnoses whether your outreach or targeting is the real bottleneck, and then calls the lead finder with the right parameters. You get a list that reflects intent, not just metadata.

2. Contact Enrichment With Verified Emails#

Use this when you already have a company list and need contacts.

Workflow: Lead Finder (find_people) → Lead Finder (enrich_person)

Example:

Here's a list of 15 companies I want to reach: [paste domains]. For each, find the Head of Marketing or Director of Demand Generation, then enrich their contact info. I need verified work emails if possible.

Lead Finder pulls email candidates from multiple sources — Prospeo, GitHub, Hunter, and Exa — and returns a scored list with source provenance. The top-level email field is the Prospeo default. This is meaningfully more reliable than a single-source lookup because the tool cross-references rather than trusting one database.

What Claude adds here is the qualification pass. Before calling enrich, it can look at the company list, spot which ones are actually a good fit, and deprioritize the rest so you are not enriching 100 contacts when you only want the 20 most likely to convert.

3. Lookalike Expansion From a Competitor URL#

Use this when your best customers look like each other and you want to find more.

Workflow: Lead Finder (find_similar) → Lead Finder (enrich_company or find_people)

Example:

My best customers are similar to Intercom, Zendesk, and Freshdesk — SaaS companies with a support team of 20-50 people. Find me 40 lookalike companies using those three URLs as seeds. Then find the VP of Customer Success at each.

The find_similar skill takes a competitor or reference URL and returns companies with similar characteristics. It is a faster starting point than trying to define every filter manually. This works well when you know who your happy customers look like but not necessarily how to express that in a database query.

4. Individual Research Before a Big Reach-Out#

Use this when you are about to reach out to a specific person and want real context first.

Workflow: People Search (deep_search or osint_search) → Lead Planner (outreach_plan)

Example:

Before I reach out to [name], the Head of Revenue at [company], find everything publicly available: their LinkedIn, recent posts, podcast appearances, articles they've written, companies they've worked at. Then help me write an outreach email that references something specific.

People Search runs across 500+ sites and can take a name, email, username, or phone as the starting point. The osint_search skill runs specialized Google dork queries — LinkedIn, resumes, news, forum posts, academic work, documents — and extracts structured links and profiles.

The result is real pre-call research in two minutes instead of 20. Claude reads back the profile and writes outreach that references an actual post they wrote, a company they used to work at, or a specific pain point they have publicly mentioned. That is a meaningfully different email than a templated sequence.

Best Prompts to Use in Claude#

These prompt patterns work well once ToolRouter is connected.

Company prospecting with specific filters#

Find 30 e-commerce companies in the US that are:
- Funded between $1M and $20M (Seed to Series A)
- 20-100 employees
- Hiring for a marketing or growth role
- Selling a physical product, not software

Then enrich each domain to get headcount, tech stack, and main contacts.
I want to reach out to whoever owns the paid ads budget.

Contact find with email enrichment#

I have this list of companies: [paste list of domains].
For each, find the person responsible for procurement or operations — 
VP, Director, or Head of level. Enrich with their work email and LinkedIn URL.
Flag which emails are verified vs. guessed so I know where to spend follow-up effort.

Lookalike prospecting from existing customers#

My three best customers are [company A], [company B], [company C].
Find 50 companies that look like them based on industry, size, and product category.
Then find the Head of [relevant role] at each and compile a contact list.

Personal research before outreach#

I'm reaching out to [full name] at [company] tomorrow.
Find what you can about them publicly — LinkedIn, recent content, career history,
any interviews or podcast appearances. Then write a personalized opening line
for a cold email that I could realistically have noticed without being creepy.

Lead generation strategy diagnosis#

Here is my business: [brief description].
Current monthly revenue: [amount]. Lead sources: [how you currently get leads].
Team: [size and structure]. Goal: [what you're trying to achieve].
Diagnose where my lead generation is actually breaking down and give me
one specific plan to fix the most important bottleneck first.

Why One Connector Beats Stitching Together Separate Lead Tools#

Manual lead research workflows are common. They usually look like this: use one tool to get a company list, export it to a spreadsheet, paste company names into a separate email finder, verify results in another tool, move everything into a CRM, then open Claude separately to write outreach. That works. It is also slow and loses context constantly.

Here is what the comparison looks like in practice:

The specific thing that disappears when you use a single conversation is re-explanation. If you find a company list, enrich it, and then ask Claude to write tailored outreach in the same thread, Claude already knows the company names, the contact details, the target role, and the tone you have been using. You do not have to paste anything back in. You do not have to explain what you already told it. That continuity is worth more than the feature comparison suggests.