Connect Autobound to Claude: Track Buyer Intent & Sales Signals

Integrating sales intelligence data into large language models transforms them from generic text generators into highly contextual sales assistants. By using Truto's Model Context Protocol (MCP) servers, you can connect Autobound directly to Claude, giving it real-time access to buyer intent signals, contact enrichment, and hyper-personalized content generation.

This article provides a step-by-step engineering guide to configuring an Autobound MCP server and connecting it to Claude. If your team uses ChatGPT, check out our guide on connecting Autobound to ChatGPT. Building custom infrastructure? See how to connect Autobound to AI Agents.

The Engineering Reality of the Autobound API

When exposing the Autobound API to an LLM, you are dealing with more than standard CRUD operations. The API has specific constraints that you and your agent orchestrator must account for.

1. Credit-Based Pricing and Execution Costs Autobound's API is heavily metered via credits. Different endpoints consume different amounts. For instance, generating a sequence (contentType: 'sequence') costs 2 credits per email generated. Submitting an intent export job costs 10 credits per topic. Claude can easily exhaust an account's monthly limit if left unchecked in a looping reasoning state. You must explicitly prompt Claude to check the balance (get_single_autobound_credit_by_id, which is free) before executing heavy operations.

2. Asynchronous Polling for Large Exports Buyer intent exports do not return synchronously. When extracting intent data, you must first submit a job, which returns an export_id. Claude must be instructed to poll the status of this job until it resolves and returns a download URL. The download URL is a JSONL file that expires after 24 hours, meaning Claude needs to act on the output quickly or hand it off to a secondary processor.

3. Strict Rate Limit Pass-Through Truto does not retry, throttle, or apply backoff on rate limit errors. When the Autobound API returns an HTTP 429, Truto passes that error directly back to Claude. Truto normalizes the upstream rate limit information into standardized headers (ratelimit-limit, ratelimit-remaining, ratelimit-reset) per the IETF specification. This means your Claude client or LangChain wrapper is strictly responsible for interpreting these headers and executing its own backoff strategy.

Step 1: Create the Autobound MCP Server

Truto generates MCP tools dynamically. It reads the integration's resource definitions and documentation records, meaning tools only appear if they are well-documented. Each server is scoped to a single integrated tenant account.

You can create the server via the Truto UI or programmatically via the API.

Method A: Via the Truto UI

1. Log into your Truto dashboard and navigate to the integrated Autobound account.
2. Click the MCP Servers tab.
3. Click Create MCP Server.
4. Name the server (e.g., "Autobound Claude Server").
5. Select specific methods or tags if you want to restrict the LLM's access (e.g., read-only).
6. Copy the generated MCP Server URL. It will look like https://api.truto.one/mcp/abc123def456...

Method B: Via the API

For automated deployments, you can provision the MCP server via a POST request to Truto's API. This endpoint creates a secure, hashed token in the distributed key-value store and returns the ready-to-use URL.

curl -X POST https://api.truto.one/integrated-account/{integrated_account_id}/mcp \
  -H "Authorization: Bearer YOUR_TRUTO_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Claude Autobound Intent Integration",
    "config": {
      "methods": ["read", "write", "custom"]
    }
  }'

The response will contain the url required by Claude.

Step 2: Connect the MCP Server to Claude

Claude Desktop supports connecting to remote MCP servers over HTTPS. You can add the Truto URL through the Claude UI or by editing the configuration file.

Method A: Via Claude Desktop UI

1. Open Claude Desktop.
2. Navigate to Settings > Connectors > Add custom connector.
3. Paste the Truto MCP Server URL.
4. Click Add. Claude will perform a handshake, fetch the available tools, and make them accessible in your chat interface.

Method B: Manual Configuration File

If you prefer managing configurations as code, edit your Claude Desktop configuration file.

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Add the Autobound connector configuration:

{
  "mcpServers": {
    "autobound-truto": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-everything",
        "--url",
        "https://api.truto.one/mcp/YOUR_TRUTO_TOKEN"
      ]
    }
  }
}

Restart Claude Desktop. The tools will now appear as available integrations.

Autobound Tool Inventory

Truto converts Autobound's endpoints into callable MCP tools using function calling. For the complete schema, visit the Autobound integration page.

Hero Tools

These are the most critical tools for empowering an LLM to act on sales signals.

1. get_single_autobound_credit_by_id

Fetches the current credit balance and usage limit. Because this call costs zero credits, you should instruct Claude to call this before running large bulk enrichment tasks.

• Prompt Example: "Check my Autobound credit balance. Do I have enough remaining credits to run a 50-contact intent export?"

2. get_single_autobound_company_by_id

Enriches a target company with signal intelligence by domain, company name, or LinkedIn URL. Returns detected signals (funding rounds, leadership changes, etc.) and a summary.

• Prompt Example: "Get the signal summary and recent timelines for stripe.com."

3. create_a_autobound_personalized_content

Generates hyper-personalized outreach content (email, call scripts, connection requests) utilizing Autobound's native models combined with real-time insight data.

• Prompt Example: "Generate a personalized email sequence for the contact at john.doe@example.com using the recent signal data you found for his company."

4. create_a_autobound_buyer_intent_export

Submits an asynchronous job to generate a downloadable JSONL file of contacts showing high intent across specific topics.

• Prompt Example: "Start a buyer intent export for the topic ID 'cloud-security'. Let me know the job ID so we can poll it."

5. autobound_companies_bulk_create

Bulk enriches up to 100 companies simultaneously. Useful when processing a list of targets from a CSV or CRM dump.

• Prompt Example: "Here is a list of 20 target domains. Run a bulk enrichment on them and summarize the top three companies with the highest signal scores."

Workflows in Action

When Claude has access to these tools, it acts as an autonomous sales researcher. Here are real-world ways you can interact with the agent.

Scenario 1: Automated Buyer Intent Extraction

"Claude, start a buyer intent export for topics related to 'cybersecurity compliance'. Once it's ready, grab the download URL. Keep an eye on our credit balance first."

Execution Steps:

1. get_single_autobound_credit_by_id: Claude verifies sufficient credits exist to fund the 10-credit export cost.
2. create_a_autobound_buyer_intent_export: Claude initiates the job, passing the required topic IDs.
3. get_single_autobound_buyer_intent_export_by_id: Claude polls this endpoint based on its internal logic until the status returns as ready and outputs the expiring download_url to the user.

Output: Claude confirms the job completion and provides the secure link to the JSONL intent data, noting that the URL expires in 24 hours.

Scenario 2: Hyper-Personalized Outreach Sequence Generation

"Take these three domains: linear.app, vercel.com, and supabase.com. Enrich them to find recent buying signals, then draft a personalized connection request for the primary technical contact at each."

Execution Steps:

1. autobound_companies_bulk_create: Claude passes the array of domains to fetch the company profiles, signal intelligence, and signal summaries in a single batch.
2. create_a_autobound_personalized_content: Iterating over the enriched data, Claude makes three separate calls with contentType: 'connection request', passing the target domains and the signals as context.

Output: The user receives three distinct, highly relevant LinkedIn connection requests based on real-time organizational changes or funding news detected by Autobound.

Security and Access Control

When generating MCP servers via Truto, you are granting powerful capabilities to external LLM clients. Truto provides four key mechanisms to safely give an AI agent access to third-party SaaS data:

• Method Filtering (config.methods): Restrict the MCP server to specific HTTP methods. For example, setting this to ["read"] limits Claude to only using get and list operations, preventing it from accidentally consuming credits via bulk creation endpoints.
• Tag Filtering (config.tags): Integration resources in Truto are tagged (e.g., enrichment, settings). You can restrict an MCP server to only expose tools matching specific tags.
• API Token Authentication (require_api_token_auth): Enable this to require a Truto API token in the Authorization header of the MCP request. This ensures that simply possessing the URL is not enough to execute tools.
• Time-to-Live (expires_at): Set an ISO datetime string to automatically revoke the server URL. Truto handles the cleanup asynchronously via scheduled alarms and distributed key expiration, leaving no stale access vectors behind.