Let Claude, OpenAI GPT, Cursor, and any AI agent query 950M+ profiles directly — social, behavioral, contact — with a single natural language command. Set up in 30 seconds.
MCP (Model Context Protocol) is an open standard that allows AI models — Claude, GPT, Cursor, and other agents — to call external tools and retrieve data directly during reasoning.
Instead of users copy-pasting data into prompts, the AI agent automatically calls SocialEnrich MCP Server to fetch social profiles, behavioral signals, and verified contacts — then uses that context to complete the task.
SocialEnrich is the first enrichment platform to provide a native MCP server — turning 950M+ profiles into ready-made context for any AI agent.
SocialEnrich supports both Anthropic Claude (MCP native) and OpenAI GPT (function calling) — choose the protocol that fits your stack.
Claude supports MCP protocol natively — SocialEnrich MCP Server registers as a tool and Claude can call it directly in conversation. Supports Claude Desktop, Claude API, and Claude in Cursor.
npm install -g @socialenrich/mcp-server
Add to claude_desktop_config.json:
{
"mcpServers": {
"socialenrich": {
"command": "socialenrich-mcp",
"env": {
"SE_API_KEY": "se_live_..."
}
}
}
}Ask Claude: "Enrich jane@stripe.com and write a personalized email for her"
OpenAI GPT-4o and GPT-4.1 support function calling — SocialEnrich registers as a function tool. Works with Chat Completions API, Assistants API, and custom GPT Actions.
pip install socialenrich openai
tools = [{ "type": "function", "function": { "name": "enrich_contact", "description": "Enrich a contact...", "parameters": { ... } } }]
GPT receives tool_call → calls SocialEnrich API → gets profile → continues reasoning with full context.
MCP Server exposes 6 tools to AI agents — each tool maps to a core SocialEnrich capability.
Full contact enrichment from email address — returns social, behavioral, contact, insights.
enrich_by_email("jane@stripe.com")
Enrich from phone number — supports international formats, auto-normalization.
enrich_by_phone("+84-xxx-xxx")
Enrich from LinkedIn URL, X/Twitter handle, or any social profile URL.
enrich_by_social_url("linkedin.com/in/...")
Search contacts by keyword, domain, title, or industry — returns a list of matches.
search_contacts("VP Sales fintech")
Retrieve behavioral data — intent signals, buying stage, cohorts — for a known contact.
get_behavioral_signals("jane@stripe.com")
Check current credit balance — so agents know when to notify users about usage.
check_credits()
Here's a real workflow: A sales rep asks Claude to enrich a prospect, write a personalized email based on social insights and behavioral signals — all in 1 conversation turn.
Claude calls SocialEnrich MCP → receives complete profile → uses personal insights to craft an outreach message. Time: 5 seconds instead of 30 minutes of manual research.
// User prompt in Claude Desktop: // "Enrich jane@stripe.com and write // a personalized outbound email" // Claude automatically calls MCP tool: Tool call: enrich_by_email Input: { "email": "jane@stripe.com" } // SocialEnrich returns: { "contact": { "name": "Jane Doe", "title": "VP of Sales", "company": "Stripe" }, "insights": { "interests": ["AI sales tools"], "communities": ["Pavilion"], "podcast": "Revenue Architects" }, "behavioral": { "intent": "evaluating_crm", "buying_stage": "evaluation" } } // → Claude writes email using above insights // → Mention podcast, Pavilion community // → Address CRM evaluation intent
How revenue teams are using MCP Server in practice.
Agent receives a list of prospects → calls SocialEnrich for each contact → writes personalized emails based on social insights, interests, and intent signals. Reply rate increases 3–5x compared to generic templates.
Agent runs before every meeting: enriches all attendees → compiles social profiles, mutual connections, behavioral signals → outputs a 1-page briefing for the AE.
Agent runs daily: checks behavioral signals for target account list → when an intent signal fires ("evaluating CRM", "hiring SDRs") → alerts SDR via Slack with full context.
Agent receives target account → search_contacts finds all decision makers → enriches each person → identifies warm intro paths via relationship mapping → suggests multi-threading strategy.
Both access 950M+ profiles. They differ in integration approach — choose based on your current stack.
| Criteria | Claude (MCP) | OpenAI GPT (Functions) |
|---|---|---|
| Protocol | MCP (Model Context Protocol) | Function Calling / Tool Use |
| Setup | npm install + config JSON | Register function schema in API call |
| Setup time | ~30 seconds | ~5 minutes |
| Desktop App | ✓ Claude Desktop native | ✗ Needs custom app |
| IDE Integration | ✓ Cursor native | Via extension |
| Conversation memory | ✓ Auto context window | ✓ Via messages history |
| Custom GPT / Actions | N/A | ✓ GPT Actions support |
| Assistants API | N/A | ✓ Native tool |
| Data access | 950M+ profiles | 950M+ profiles |
| Behavioral signals | ✓ Full | ✓ Full |
| Credits | Shared pool | Shared pool |
| Best for | Claude power users, Cursor devs, MCP ecosystem | GPT ecosystem, custom agents, Assistants API |
Conclusion: Use Claude MCP if you're in the Claude/Cursor ecosystem. Use OpenAI functions if you build on the GPT API. Both access the same data layer.
950M+ profiles. One API. First-party data, no middlemen.
No credit card required · 100 free enrichments · Cancel anytime