People.ai MCP Setup Guide

The People.ai Model Context Protocol (MCP) server is the core data source for every skill in this library. It gives AI assistants direct access to account activity, engagement data, opportunity status, and contact intelligence.

What You Can Do

• Build AI skills that analyze accounts, opportunities, and stakeholder engagement in real time
• Create interactive dashboards from live CRM and communication data
• Generate meeting prep briefs, competitive battle cards, and executive summaries on demand
• Score deal health, renewal risk, and pipeline quality using AI-powered analysis
• Map relationships, detect engagement gaps, and recommend next best actions

Available MCP Tools

The People.ai MCP server exposes these tools to connected AI assistants:

1. Obtain MCP Credentials

1. Contact your People.ai administrator or CSM to request MCP API access
2. You'll receive two credentials:
   • Client ID — your organization's MCP client identifier
   • Client Secret — authentication secret
3. Store these credentials securely — you'll need them for platform configuration

2. Configure in Claude.ai

Claude.ai Projects support MCP integrations natively:

1. Open Claude.ai and go to Settings → Integrations
2. Find People.ai MCP in the available integrations
3. Click Connect and enter your Client ID and Client Secret
4. Enable the integration — it will now be available in all your Projects
5. Create a new Project, paste skill instructions, and start chatting

3. Configure in Claude Code

Add People.ai MCP to your Claude Code settings:

# In your Claude Code MCP settings (settings.json or .mcp.json):
{
  "mcpServers": {
    "peopleai": {
      "url": "https://mcp.people.ai/mcp",
      "transport": "sse",
      "headers": {
        "PAI-Client-Id": "your-client-id",
        "PAI-Client-Secret": "your-client-secret",
        "Content-Type": "application/json"
      }
    }
  }
}

Copy skill instructions into your CLAUDE.md file or invoke them as slash commands.

4. Configure in ChatGPT

ChatGPT Custom GPTs use Actions to connect to external APIs:

1. Go to ChatGPT → Explore GPTs → Create
2. Paste the skill instructions into the Instructions field
3. Under Actions, click Create new action
4. Configure API endpoints for each People.ai MCP tool the skill uses
5. Set authentication headers: PAI-Client-Id and PAI-Client-Secret
6. Save and publish the GPT

5. Configure in Microsoft Copilot

Use Copilot Studio to create agents with People.ai data:

1. Open Microsoft Copilot Studio and create a new agent
2. Paste the skill instructions into the agent's instructions field
3. Under Plugins, add a new connector for People.ai
4. Configure the MCP endpoint: https://mcp.people.ai/mcp
5. Add authentication headers: PAI-Client-Id and PAI-Client-Secret
6. Publish and test the agent

6. Configure in Google Gemini

7. Developer Platforms

For programmatic access, use these SDK examples:

LangGraph (Python)

pip install langchain-mcp-adapters

from langchain_mcp_adapters.client import MultiServerMCPClient

client = MultiServerMCPClient({
    "peopleai": {
        "url": "https://mcp.people.ai/mcp",
        "transport": "sse",
        "headers": {
            "PAI-Client-Id": os.environ["PEOPLEAI_CLIENT_ID"],
            "PAI-Client-Secret": os.environ["PEOPLEAI_CLIENT_SECRET"],
            "Content-Type": "application/json",
        }
    }
})
tools = await client.get_tools()

Claude Agent SDK (Python)

pip install claude-agent-sdk anthropic

from claude_agent_sdk import Agent
import anthropic

agent = Agent(
    client=anthropic.Anthropic(),
    model="claude-sonnet-4-20250514",
    system="You are a sales intelligence assistant...",
    mcp_servers=[{
        "name": "peopleai",
        "url": "https://mcp.people.ai/mcp",
        "headers": {
            "PAI-Client-Id": os.environ["PEOPLEAI_CLIENT_ID"],
            "PAI-Client-Secret": os.environ["PEOPLEAI_CLIENT_SECRET"],
            "Content-Type": "application/json",
        }
    }],
)

OpenAI Agents SDK (Python)

pip install openai-agents

from agents import Agent
from agents.mcp import MCPServerSse

peopleai_mcp = MCPServerSse(
    name="peopleai",
    url="https://mcp.people.ai/mcp",
    headers={
        "PAI-Client-Id": os.environ["PEOPLEAI_CLIENT_ID"],
        "PAI-Client-Secret": os.environ["PEOPLEAI_CLIENT_SECRET"],
        "Content-Type": "application/json",
    },
)

agent = Agent(
    name="Sales Intelligence",
    instructions="You are a sales intelligence assistant...",
    mcp_servers=[peopleai_mcp],
)

8. Environment Variables

Set these for Python-based integrations:

export PEOPLEAI_MCP_URL="https://mcp.people.ai/mcp"
export PEOPLEAI_CLIENT_ID="your-client-id"
export PEOPLEAI_CLIENT_SECRET="your-client-secret"

# Also needed for the LLM:
export ANTHROPIC_API_KEY="sk-ant-..."   # for Claude-based integrations
export OPENAI_API_KEY="sk-..."          # for OpenAI-based integrations

9. Test Your Connection

Try these example queries to verify the connection works:

• "Find the account for Acme Corp"
• "What are the risks for the Acme Enterprise deal?"
• "Show me recent activity for Globex Industries"

If the AI responds with real account data, you're connected. If it says it can't access People.ai tools, double-check your credentials and integration setup.

10. Troubleshooting

Setup Guides

Step-by-step instructions for connecting People.ai MCP to your preferred AI platform.

Salesforce Agentforce + People.ai MCP

Connect People.ai's MCP server to Salesforce Agentforce to bring real-time sales intelligence directly into your Salesforce environment. Agentforce agents can use People.ai data to analyze accounts, score opportunities, and recommend next best actions — all within Salesforce.

Why Agentforce + People.ai?

• Embedded in Salesforce — agents surface insights where reps already work, no context switching
• Dual data access — combine native Salesforce objects with People.ai communication intelligence
• Event-driven triggers — fire agents automatically on stage changes, close date shifts, or activity gaps
• Write-back capabilities — agents can update Salesforce fields, create tasks, and log activities
• Governed by Salesforce Trust Layer — data masking, audit trails, and permission-based access

1. Register People.ai as an External Service

1. In Salesforce Setup, search for Named Credentials
2. Click New Named Credential and configure:

   Label	People_ai_MCP
   URL	https://mcp.people.ai/mcp
   Identity Type	Named Principal
   Authentication Protocol	Custom Header

3. Add custom headers:

   Header Name	Value
   PAI-Client-Id	[Your Client ID]
   PAI-Client-Secret	[Your Client Secret]
   Content-Type	application/json

4. Save the Named Credential

2. Create Agentforce Actions for MCP Tools

Each People.ai MCP tool becomes an Agentforce Action. Create these as External Service Actions or Apex Invocable Actions:

@InvocableMethod(label='Get Account Status from People.ai'
    description='Retrieves account health and engagement data')
public static List<AccountStatusResult> getAccountStatus(
    List<AccountStatusRequest> requests
) {
    HttpRequest req = new HttpRequest();
    req.setEndpoint('callout:People_ai_MCP/get_account_status');
    req.setMethod('POST');
    req.setBody(JSON.serialize(new Map<String, Object>{
        'account_id' => requests[0].accountId
    }));
    HttpResponse res = new Http().send(req);
    // Parse and return results
}

3. Create an Agentforce Agent

1. In Setup, go to Agentforce Agents (or search "Agent Builder")
2. Click New Agent
3. Give it a name (e.g., Sales Intelligence Agent)
4. Add a Topic for each skill category:

   Topic	Description	Actions to Assign
   Account Intelligence	Analyze account health, engagement, and relationships	Find Account, Get Account Health, Get Engaged Contacts, Ask Sales AI
   Deal Management	Opportunity analysis, win probability, and deal coaching	Get Opportunity Status, Ask Sales AI About Deal, Get Scorecard
   Pipeline Review	Pipeline health, forecasting, and territory analysis	Get Top Records, Get Opportunity Status, Ask Sales AI

5. Paste the skill instructions from this library into each Topic's Instructions field
6. Assign the People.ai Actions you created in Step 2 to each Topic
7. Click Activate to deploy the agent

4. Configure Triggers (Optional)

Agentforce agents can run automatically based on Salesforce events:

• Stage Change — Run opportunity analysis when a deal moves to a new stage
• Close Date Shift — Trigger risk assessment when a close date is pushed
• Activity Gap — Alert when no meetings or emails logged for 14+ days
• New Account Assignment — Auto-generate account plan when an AE is assigned a new account

Configure triggers in Setup → Process Builder or Flow Builder, invoking your Agentforce agent as the action.

5. Test Your Agent

Try these queries in the Agentforce test console:

• "Analyze the Acme Corp account — what's the engagement health?"
• "What's the win probability for the Globex Enterprise deal?"
• "Who are the key stakeholders at Initech and when did we last meet?"

If the agent responds with People.ai data, you're connected. Check the agent's execution log for tool call details.

6. Troubleshooting