DEV Community

Diego Costa
Diego Costa

Posted on

Eliminating Hallucinations in Sales Agents with B2B Lead Enrichment MCP

#mcp #ai #automation

Eliminating Hallucinations in Sales Agents with B2B Lead Enrichment MCP

To stop LLMs from hallucinating company metadata during lead research, developers must use a B2B lead enrichment MCP server that integrates strict Zod annotations for parameter enforcement. This native Model Context Protocol implementation ensures that agents query real-time firmographics and intent data with precise, validated inputs instead of guessing company details.

Core Features

The B2B Lead Enrichment MCP API provides a robust data layer for AI agents (Cursor, Claude Desktop, VS Code Copilot) to retrieve verified intelligence without leaving the IDE or chat interface. By leveraging the Model Context Protocol, this server exposes tools that use rigid schema definitions to prevent the "fuzzy input" problem common in standard REST calls.

  • Firmographic Intelligence: Real-time access to revenue ranges, employee counts, and HQ locations.
  • Technographic Detection: Identify the specific software stack (SaaS, Cloud, Infrastructure) a target company uses.
  • Buying Intent Signals: Surface verified B2B intent data to prioritize high-value leads.
  • Hallucination Prevention: All tool parameters (domain, company_name) are guarded by Zod-typed schemas, forcing the LLM to provide clean, valid strings before the request is even dispatched.

Claude Desktop & Cursor Configuration

To give your local AI agent native access to this live data, add the following configuration to your claude_desktop_config.json or Cursor's MCP settings:

{
  "mcpServers": {
    "b2b-enrichment": {
      "command": "curl",
      "args": [
        "-s",
        "https://lead-enrichment-mcp.agent-infra.workers.dev/mcp"
      ],
      "env": {
        "API_KEY": "YOUR_ENRICHMENT_API_KEY"
      }
    }
  }
}
Enter fullscreen mode Exit fullscreen mode

JSON-RPC Enrichment Payload Example

When the LLM invokes the enrich_lead tool, the MCP server returns a structured JSON-RPC response. This ensures the agent processes clean data rather than parsing messy HTML or unverified search results:

{
  "jsonrpc": "2.0",
  "result": {
    "companyName": "Example Tech Corp",
    "industry": "Cloud Infrastructure",
    "technographics": ["AWS", "Kubernetes", "Snowflake"],
    "intentSignals": ["Increasing spend on cybersecurity", "Hiring DevOps"],
    "firmographics": {
      "revenue": "$50M-$100M",
      "employees": "250-500"
    },
    "confidenceScore": 0.94
  }
}
Enter fullscreen mode Exit fullscreen mode

Risk-Free Metered Billing

Traditional B2B data providers charge for every API call, regardless of quality. This MCP server implements a Risk-Free Metered Billing model optimized for AI workflows. You are only billed for successful enrichments that return a Confidence Score > 0.6. If the data is low-confidence or the query fails to find a match, the cost is $0, making it the most cost-effective solution for building autonomous SDR swarms and research agents.

Get Your Free B2B Enrichment API Key

Ready to upgrade your AI agent with real-time firmographic data? Visit lead-enrichment-mcp.agent-infra.workers.dev to generate your API key on the free tier and start building hallucination-free sales workflows today.

Top comments (0)