API Reference
Data CatalogGet a DemoStart Free
Start typing to search…

AI Assistant Integrations

Enhance your AI assistants with real-time business and contact data through AgentSource MCP integration.

Overview

AgentSource MCP seamlessly integrates with popular AI assistants, enabling them to access comprehensive business intelligence directly within chat interfaces. This integration transforms your AI assistant into a powerful research tool capable of finding companies, identifying contacts, and providing detailed insights on demand.

Supported AI Assistants

Claude Desktop

Claude by Anthropic - The most popular MCP-compatible AI assistant

Claude Desktop provides native support for MCP servers, making it the ideal choice for accessing AgentSource data through natural conversation.

Setup Instructions

  1. Install Claude Desktop from claude.ai

  2. Configure MCP Server

    Add this configuration to your claude_desktop_config.json:

    {
  "mcpServers": {
    "Explorium": {
      "command": "python",
      "args": ["-m", "explorium_mcp_server"],
      "env": {
        "EXPLORIUM_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

  3. Restart Claude Desktop to load the configuration

  4. Verify Integration

    Ask Claude: "What tools do you have available from Explorium?"

Example Conversations

Finding Companies:

You: Find me 10 SaaS companies in San Francisco with 50-200 employees

Claude: I'll search for SaaS companies in San Francisco with 50-200 employees.

[Searches using fetch_businesses tool]

I found 10 SaaS companies matching your criteria:

1. TechFlow Analytics
   - 125 employees
   - Revenue: $10M-$25M
   - Technologies: AWS, React, PostgreSQL
   
2. DataSync Solutions
   - 87 employees
   - Revenue: $5M-$10M
   - Technologies: Google Cloud, Python, MongoDB
   
[... continues with detailed results]

Researching Contacts:

You: Who is the CTO of Snowflake and what's their background?

Claude: I'll look up information about Snowflake's CTO.

[Uses match_businesses, then enrich_businesses_financial_metrics]

The CTO of Snowflake is Benoit Dageville. Here's his background:
- Co-founder of Snowflake
- Previously at Oracle for 16 years
- PhD in Computer Science from University of Paris
- Expertise in database systems and query optimization

Cursor IDE

Cursor - The AI-powered code editor with MCP support

Cursor's built-in AI features can leverage AgentSource data for research while coding, making it perfect for developers building data-driven applications.

Setup Instructions

  1. Open Cursor Settings

    • Navigate to: Cursor > Settings > Cursor Settings > MCP

  2. Add Explorium Integration

    • Name: Explorium
    • Command: python -m explorium_mcp_server
    • Environment Variables: 
      EXPLORIUM_API_KEY=your_api_key_here

  3. Enable Yolo Mode (Optional)

    • Location: Cursor > Settings > Features > Chat
    • Toggle: "Enable Yolo mode"
    • Benefit: Tools execute automatically without confirmation

  4. Test the Integration

    • Open chat (Cmd+K or Ctrl+K)
    • Ask: "Using Explorium, find tech companies using Python"

Use Cases in Cursor

  • Research companies while building CRM integrations
  • Get real data for testing and development
  • Generate mock data based on real company information
  • Validate business logic with actual market data

Common Patterns

Research Assistant Pattern

Transform your AI into a business research assistant:

1. User asks about a company or industry
2. AI uses match_businesses or fetch_businesses
3. AI enriches with firmographics, technographics, financials
4. AI summarizes findings in natural language

Sales Intelligence Pattern

Enable your AI to support sales activities:

1. User provides target company name
2. AI finds company with match_businesses
3. AI fetches key contacts with fetch_prospects
4. AI enriches contact information
5. AI provides actionable insights

Market Analysis Pattern

Use AI for competitive intelligence:

1. User asks about market segment
2. AI searches companies with filters
3. AI aggregates statistics
4. AI identifies trends and patterns
5. AI generates market report

Best Practices

🎯 Prompt Engineering

Be Specific

Good: "Find B2B SaaS companies in NYC with 50-200 employees using AWS"
Bad: "Find some tech companies"

Chain Requests

"First find Microsoft's business ID, then show me all VPs in their engineering department"

Request Formats

"Show me the results as a table with company name, size, and revenue"

⚡ Performance Tips

  • Batch Operations: Ask for multiple companies at once
  • Specific Filters: Use precise criteria to reduce results
  • Caching: AI assistants often cache recent results
  • Pagination: Request specific page numbers for large datasets

🔒 Security Considerations

  • Store API keys in secure configuration files
  • Never share API keys in conversations
  • Use read-only keys when possible
  • Monitor usage through the admin panel

Advanced Features

Tool Chaining

Most AI assistants can automatically chain multiple tools:

"Find the top 5 competitors of Salesforce and compare their employee counts and technologies"

This triggers:
1. match_businesses (Salesforce)
2. enrich_businesses_competitive_landscape
3. match_businesses (for each competitor)
4. enrich_businesses_firmographics
5. enrich_businesses_technographics

Conversational Memory

AI assistants maintain context across messages:

You: "Find info about Uber"
AI: [Provides Uber details]
You: "Now show me their main competitors"
AI: [Understands "their" refers to Uber]

Natural Language Filters

Convert natural language to API filters:

"Companies founded in the last 3 years" → company_age: ["0-3"]
"Mid-size businesses" → company_size: ["51-200", "201-500"]
"Tech companies" → linkedin_category: ["Technology"]

Troubleshooting

Common Issues

"MCP server not found"

  • Ensure explorium_mcp_server is installed
  • Check Python path configuration
  • Verify command syntax in config

"Authentication failed"

  • Verify API key is correct
  • Check environment variable name
  • Ensure key has proper permissions

"No results found"

  • Try broader search criteria
  • Check filter compatibility
  • Verify data availability

Debug Mode

Enable verbose logging in most AI assistants:

Claude Desktop:

{
  "mcpServers": {
    "Explorium": {
      "command": "python",
      "args": ["-m", "explorium_mcp_server", "--verbose"],
      "env": {
        "EXPLORIUM_API_KEY": "YOUR_API_KEY",
        "LOG_LEVEL": "DEBUG"
      }
    }
  }
}

Examples Gallery

Company Research

"Give me a complete profile of Tesla including their technology stack, 
financial metrics, recent events, and key executives"

Competitive Analysis

"Compare the top 5 CRM companies by employee count, revenue, 
and technology stack. Show the results in a table."

Lead Generation

"Find marketing directors at fintech companies in London 
with 100-500 employees. Include their email addresses."

Market Trends

"What percentage of SaaS companies with $10M-$50M revenue 
are using AWS vs Google Cloud vs Azure?"

Get Support

Resources

Contact

Ready to enhance your AI assistant? Install theMCP server and follow the setup guide for your platform above.