Connect Autobound to ChatGPT: Write Personalized Sales Content

Connecting Autobound's sales intelligence engine to ChatGPT transforms it from a generic writing assistant into a hyper-personalized outbound machine. By exposing Autobound's endpoints via the Model Context Protocol (MCP), ChatGPT gains real-time access to buyer intent data, company signals, and credit-aware content generation. If your team uses Claude, check out our guide on connecting Autobound to Claude, or if you are architecting headless workflows, read about connecting Autobound to AI Agents.

This guide breaks down how to generate an Autobound MCP server using Truto, connect it to ChatGPT, and safely execute workflows without burning through your API credits.

The Engineering Reality: Autobound API Quirks

Autobound's API is a specialized system designed for signal detection and content generation. Before giving an LLM unfettered access, you need to account for a few vendor-specific realities:

1. Credit-Based Billing Model: Autobound API operations are metered tightly. Generating a sequence email consumes 2 credits, while a buyer intent export costs 10 credits per topic. You must structure your AI agent prompts to explicitly check the get_single_autobound_credit_by_id endpoint (which is free) before executing high-volume bulk or export commands.
2. Asynchronous Polling for Intent Data: Buyer intent exports are not synchronous. Your agent must first POST to create the export, receive a job ID, and then actively poll a secondary endpoint until the status returns as ready. The resulting download URLs expire in 24 hours.
3. Polymorphic Identifier Requirements: Enrichment endpoints like get_single_autobound_company_by_id do not rely solely on a primary key. They require at least one valid identifier from a subset - either a domain, a company name, or a LinkedIn URL. Vague inputs will result in validation errors.

Factual note on rate limits: Truto does not retry, throttle, or apply backoff on rate limit errors. When an upstream API returns HTTP 429, Truto passes that error directly to the caller. Truto normalizes upstream rate limit info into standardized headers (ratelimit-limit, ratelimit-remaining, ratelimit-reset) per the IETF spec. The caller - in this case, your ChatGPT agent loop or middleware - is strictly responsible for handling retries and backoff.

Creating the Autobound MCP Server

Truto dynamically generates an MCP server from the Autobound integration's underlying schema and documentation. You can spin this up either via the UI or programmatically.

Method 1: Via the Truto UI

1. Log in to Truto and navigate to your connected Autobound integrated account.
2. Select the MCP Servers tab.
3. Click Create MCP Server.
4. Define a name (e.g., "Autobound GPT Server") and apply any desired method or tag filters.
5. Click Create and securely copy the generated https://api.truto.one/mcp/<token> URL.

Method 2: Via the Truto API

For DevOps workflows, you can generate the server directly via REST. Make a POST request to the /integrated-account/:id/mcp endpoint:

curl -X POST https://api.truto.one/integrated-account/<your_autobound_account_id>/mcp \
  -H "Authorization: Bearer <your_truto_api_token>" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "ChatGPT Content Generator",
    "config": {
      "methods": ["read", "write", "custom"]
    }
  }'

The response payload contains the url required by ChatGPT. Truto hashes the token instantly - the raw token is never stored.

Connecting the MCP Server to ChatGPT

ChatGPT supports custom MCP connectors. Depending on your workspace configuration, you can add this connection via the interface or through an enterprise config file.

Via the ChatGPT UI

1. Open ChatGPT and go to Settings > Apps > Advanced settings.
2. Toggle on Developer mode to enable MCP capabilities.
3. Under MCP servers / Custom connectors, add a new server.
4. Name: "Autobound Content Engine".
5. Server URL: Paste the Truto MCP URL generated in the previous step.
6. Save the configuration. ChatGPT will automatically handshake with Truto and pull the Autobound tool schemas.

Via the Manual Config File Approach

For programmatic or enterprise ChatGPT environments that rely on JSON configuration files for workspace agents, you inject the MCP server configuration similarly to standard tool manifests:

{
  "mcpServers": {
    "autobound_sales": {
      "command": "https",
      "url": "https://api.truto.one/mcp/<token>",
      "requireAuth": false
    }
  }
}

Note: If you configured your Truto MCP server with require_api_token_auth: true, you must pass your Truto API token in the authorization headers within your client configuration.

Autobound Tool Inventory

Truto translates Autobound's endpoints into granular JSON-RPC tools. Below are the most critical tools for content generation and intelligence workflows. For the complete schema, visit the Autobound integration page.

Hero Tools

1. create_a_autobound_personalized_content

Generates hyper-personalized outreach content utilizing Autobound's AI models. It accepts formats like email, sequence, call script, or SMS.

• Contextual usage: Requires contentType alongside either email or LinkedIn URL for both the contact and the user. Watch out for credit consumption: when contentType is 'sequence', each email costs 2 credits.
• Example prompt: "Draft a personalized sequence for john@example.com from my email (me@example.com). Use Autobound to generate the content."

2. get_single_autobound_company_by_id

Enriches a single company profile with deep signal intelligence, returning timelines, scores, and summaries.

• Contextual usage: Requires at least one of domain, company_name, or linkedin_url. Excellent for contextualizing a prospect before writing a pitch.
• Example prompt: "Look up the Autobound intelligence for acmecorp.com and summarize their recent buying signals."

3. create_a_autobound_buyer_intent_export

Submits an asynchronous job to generate a downloadable JSONL file of contacts exhibiting buyer intent.

• Contextual usage: Costs 10 credits per topic. Requires topic_ids. Since it is asynchronous, you must instruct the LLM to take the returned export_id and poll the status endpoint.
• Example prompt: "Start a buyer intent export for topic ID 45. Give me the export ID when it's queued."

4. get_single_autobound_credit_by_id

Retrieves your current credit balance and cycle details.

• Contextual usage: Crucial for agentic guardrails. This call is free and should be executed prior to any bulk creation or sequence generation tool to ensure you have sufficient remaining credits.
• Example prompt: "Check my Autobound credit balance. Do I have enough to run an intent export?"

5. autobound_companies_bulk_create

Bulk enriches up to 100 companies with signal intelligence in a single request.

• Contextual usage: Takes an array of domains. Highly efficient for list scrubbing before executing mass outreach.
• Example prompt: "Bulk enrich these five domains: stripe.com, twilio.com, truto.one, vercel.com, and render.com using Autobound."

Workflows in Action

Scenario 1: Automated Intent Prospecting & Content Generation

Sales development reps need highly targeted emails based on verified company changes. Here is how ChatGPT orchestrates the research and drafting phases sequentially.

User prompt: "Check my credit balance first. If I have more than 5 credits, research the domain 'example.com'. If they have any recent funding or hiring signals, draft a personalized email to jane@example.com from me (sales@mycompany.com)."

Step-by-step execution:

1. ChatGPT calls get_single_autobound_credit_by_id to verify the remaining balance exceeds 5.
2. It calls get_single_autobound_company_by_id using the domain example.com to extract the signal_summary.
3. After finding relevant signals in the payload, ChatGPT calls create_a_autobound_personalized_content with contentType: "email", passing the required contact and user emails.

Result: The user receives a finalized, high-quality email draft directly inside the chat interface, backed by real-time intelligence, without ever leaving the ChatGPT window or risking a credit overdraft.

Scenario 2: Async Buyer Intent Polling

Marketing operations needs a fresh list of leads for a specific intent topic but doesn't want to deal with manual API polling.

User prompt: "Kick off a buyer intent export for topic ID 102. Poll the status until the download URL is ready, and give me the link."

Step-by-step execution:

1. ChatGPT calls create_a_autobound_buyer_intent_export with topic_ids: [102] and receives an export_id with a pending status.
2. ChatGPT pauses, then repeatedly calls get_single_autobound_buyer_intent_export_by_id using the retrieved ID.
3. Once the status changes to completed, ChatGPT extracts the download_url and total_contacts.

Result: The user receives the temporary, secure download link directly in the chat, abstracting away the asynchronous complexity of Autobound's backend.

Security and Access Control

Exposing an endpoint that actively consumes paid credits to an LLM requires strict boundary setting. Truto's MCP architecture provides native security features at the token level:

• Method Filtering: You can restrict the MCP server config to methods: ["read"], allowing ChatGPT to query signals and company data while blocking expensive operations like create_a_autobound_personalized_content.
• Tag Filtering: Limit the server to specific tool tags (e.g., exposing only the webhooks endpoints for a DevOps agent).
• Extra Authentication (require_api_token_auth): For enterprise workspaces, mandate that the client passes a valid Truto API token in the headers alongside the server URL. This prevents an leaked URL from being exploited.
• Time-Bound Access (expires_at): Generate ephemeral servers with a strict TTL. Once the timestamp passes, the server self-destructs and the Durable Object alarm wipes the keys from the KV store instantly.