Connect SendGrid to ChatGPT: Run Campaigns and Track Metrics

If you need to connect SendGrid to ChatGPT to automate email marketing campaigns, sync contact lists, or monitor delivery metrics, you need a Model Context Protocol (MCP) server. This server acts as the translation layer between ChatGPT's tool calls and SendGrid's REST APIs. You can either build and maintain this infrastructure yourself, or use a managed integration platform like Truto to dynamically generate a secure, authenticated MCP server URL.

If your team uses Claude, check out our guide on connecting SendGrid to Claude or explore our broader architectural overview on connecting SendGrid to AI Agents.

Giving a Large Language Model (LLM) read and write access to your primary email infrastructure is an engineering challenge. You have to handle API key scopes, map massive JSON schemas to MCP tool definitions, and deal with SendGrid's specific rate limits. Every time SendGrid updates an endpoint, you have to update your server code, redeploy, and test the integration. This guide breaks down exactly how to use Truto to generate a secure, managed MCP server for SendGrid, connect it natively to ChatGPT, and execute complex workflows using natural language.

The Engineering Reality of the SendGrid API

A custom MCP server is a self-hosted integration layer. While the open MCP standard provides a predictable way for models to discover tools, the reality of implementing it against vendor APIs is painful. You aren't just integrating "email" - you are integrating the Marketing Campaigns API, the Contacts API, the Event Webhooks API, and the legacy Mail Send API, all of which have different design patterns, error formats, and behavioral quirks.

If you decide to build a custom MCP server for SendGrid, you own the entire API lifecycle. Here are the specific integration challenges that break standard CRUD assumptions when working with SendGrid:

The Asynchronous Job Pattern for Contacts

Updating a single contact in SendGrid is not a synchronous REST call. The PUT /v3/marketing/contacts endpoint accepts up to 30,000 contacts at once but processes them asynchronously. When an LLM sends a request to add a subscriber, the API immediately returns a 202 Accepted with a job_id. The contact does not exist in the database yet. If your agent immediately tries to search for that contact to verify the update, the search will return a 404. Your custom server - or your agent's system prompt - must implement polling logic against the import status endpoint using that job_id before proceeding to the next step.

The Draft-Then-Schedule Campaign Workflow

SendGrid's Single Sends (marketing campaigns) require a rigid, multi-step state machine. You cannot create and send a campaign in one API call. First, you must call the endpoint to create the Single Send, which generates a draft and returns an ID. Then, you must call an entirely separate endpoint to schedule it, passing either an ISO 8601 timestamp or the string "now". If your MCP server abstracts this poorly, the LLM will think it has sent an email when it has only generated a draft.

Strict Rate Limits and Standardized Headers

SendGrid enforces strict rate limits that vary heavily by endpoint. For example, marketing campaign endpoints allow far fewer requests per second than raw mail send endpoints. 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.

However, Truto normalizes the upstream rate limit information into standardized headers (ratelimit-limit, ratelimit-remaining, ratelimit-reset) per the IETF specification. The caller - your AI agent or framework - is completely responsible for reading these headers and executing the retry or exponential backoff. Do not assume the infrastructure will magically absorb these errors; your agent must be engineered to handle the 429 response natively and wait until the reset window clears.

Custom Field Validation Silences

When upserting contacts, you can map data to custom fields. However, SendGrid requires these custom fields to be pre-created in the account via the UI or the Custom Fields API. If an LLM hallucinates a field name or tries to pass a field that hasn't been defined, SendGrid will often drop the unmapped data silently rather than throwing a hard 400 error. The agent assumes the data was saved, leading to massive gaps in your marketing personalization.

The Managed MCP Approach

Instead of forcing your engineering team to build, host, and maintain a custom translation layer, Truto provides a managed MCP server. Truto generates MCP tools dynamically based on SendGrid's API documentation and your specific environment configuration.

Because the MCP server is hosted by Truto, it inherently supports standardized authentication, schema generation, and tool filtering without writing any custom Rust or TypeScript boilerplate.

Generating a SendGrid MCP Server

Truto scopes each MCP server to a single integrated account. This means the server URL contains a cryptographic token that maps directly to a specific user's connected SendGrid instance. You can generate this server via the Truto UI or programmatically via the API.

Method 1: Via the Truto UI

If you are setting up an internal tool for your own team, the UI is the fastest path.

1. Log into your Truto dashboard and navigate to the integrated account page for your SendGrid connection.
2. Click the MCP Servers tab.
3. Click Create MCP Server.
4. Select your desired configuration (e.g., restrict to read methods only, or filter by specific tags like marketing).
5. Copy the generated MCP server URL. It will look like https://api.truto.one/mcp/abc123def456...

Method 2: Via the Truto API

If you are building an AI product and need to provision MCP servers dynamically for your customers, you use the REST API. Make a POST request to /integrated-account/:id/mcp.

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

The API returns a fully provisioned, securely hashed token URL ready for immediate use:

{
  "id": "mcp_8f7d6c5b",
  "name": "SendGrid Marketing Agent",
  "config": {
    "methods": ["read", "write", "custom"]
  },
  "expires_at": null,
  "url": "https://api.truto.one/mcp/a1b2c3d4e5f67890"
}

Connecting the MCP Server to ChatGPT

Once you have your Truto MCP URL, you can connect it to your LLM environment. There are two primary ways to connect this to ChatGPT - via the web UI for standard ChatGPT Enterprise/Pro users, or via a manual configuration file if you are using developer frameworks or local agents.

A. Via the ChatGPT UI

If you have a ChatGPT Plus, Pro, Team, or Enterprise account, you can add custom connectors directly in the web interface.

1. In ChatGPT, navigate to Settings -> Apps -> Advanced settings.
2. Toggle Developer mode to ON.
3. Under MCP servers / Custom connectors, click Add a new server.
4. Set the Name to something descriptive (e.g., "SendGrid via Truto").
5. Paste your Truto MCP URL into the Server URL field.
6. Click Save. ChatGPT will immediately ping the server, execute the handshake, and ingest the tool schemas.

B. Via Manual Config File

If you are running local agents, using Claude Desktop, or building a custom AI application that interfaces with OpenAI's models via an MCP framework, you define the server in a JSON configuration file. Truto provides an SSE (Server-Sent Events) transport layer.

Use the official @modelcontextprotocol/server-sse package to connect to the Truto URL.

{
  "mcpServers": {
    "sendgrid-truto": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-sse",
        "https://api.truto.one/mcp/a1b2c3d4e5f67890"
      ]
    }
  }
}

When your agent boots, it will execute the initialize handshake, request tools/list, and dynamically load all the SendGrid endpoints you authorized.

Hero Tools for SendGrid

Truto exposes the entire SendGrid API as standardized tools. Here are the core "hero tools" your AI agent will use most frequently when running campaigns and tracking metrics.

create_a_send_grid_single_send

This tool creates a draft marketing campaign. It requires a name and accepts complex configurations including target list IDs, segment IDs, HTML content, plain text content, and sender identity IDs. Remember that this only creates a draft - it does not dispatch the email.

"Draft a new SendGrid Single Send named 'Q4 Feature Announcement'. Target list ID 98765. Set the subject line to 'Major Platform Updates for Q4' and use the standard sender identity."

create_a_send_grid_schedule_single_send

Once a draft is created, this tool schedules it for delivery. You must pass the Single Send ID and a send_at parameter. To send immediately, pass the string "now". Otherwise, provide a valid ISO 8601 timestamp.

"Take the Single Send draft I just created (ID 12345) and schedule it to dispatch 'now'."

list_all_send_grid_global_statistics

This tool retrieves aggregated email statistics for a specific date range. It returns hard metrics like requests, delivered, opens, clicks, bounces, and spam reports. This is critical for agents acting as marketing analysts.

"Fetch the global statistics for SendGrid from October 1st to October 31st. Summarize the total opens, unique clicks, and highlight the overall bounce rate."

create_a_send_grid_contact

This tool upserts up to 30,000 contacts at once. It requires at least one identifier (email, phone number, or external ID). The API returns a job_id which must be tracked. If your agent needs to add custom fields, those fields must already exist in SendGrid.

"Add a new contact to SendGrid with the email 'cto@example.com', first name 'Sarah', and last name 'Connor'. Add her to list ID 45678."

list_all_send_grid_messages

This tool filters raw email activity logs. It is essential for debugging deliverability issues. You must pass a specific SGQL (SendGrid Query Language) string in the query parameter to filter by email address, status, or date.

"Search the SendGrid message activity logs for any emails sent to 'cto@example.com' in the last 3 days that resulted in a 'bounce' or 'block' status."

send_grid_contacts_search_emails

This tool allows an agent to look up specific contacts by passing an array of email addresses. It returns the full contact profile including list memberships, custom field values, and creation dates.

"Search SendGrid for the contact profile associated with 'billing@acme.com'. Tell me what lists they are currently subscribed to."

For the complete tool inventory and detailed schema definitions, visit the SendGrid integration page.

Workflows in Action

Providing an LLM with raw tools is only half the battle. The real value comes from orchestrating multi-step workflows. Here are two real-world sequences an AI agent can execute using the SendGrid MCP server.

Workflow 1: The Campaign Deployment Engine

A marketing manager wants to quickly spin up a targeted email blast without logging into the SendGrid UI and clicking through the campaign builder.

"Create a new SendGrid Single Send called 'Urgent Maintenance Notice'. Target list ID 112233. The subject is 'Scheduled Downtime - Nov 12'. The body should warn users of a 2-hour window. Once the draft is created, schedule it to send immediately."

Step-by-step execution:

1. create_a_send_grid_single_send: The agent parses the prompt, formats the plain text and HTML body, and calls the API to create the draft. SendGrid returns an ID (e.g., ss_998877).
2. create_a_send_grid_schedule_single_send: The agent immediately takes the returned ID, maps it to the id parameter, sets the send_at parameter to "now", and executes the scheduling call.

Result: The agent replies confirming the campaign has been queued for immediate delivery, entirely bypassing the manual UI process.

sequenceDiagram
    participant User
    participant ChatGPT
    participant MCP Server
    participant SendGrid API

    User->>ChatGPT: "Create & schedule Maintenance Notice"
    ChatGPT->>MCP Server: create_a_send_grid_single_send (List 112233)
    MCP Server->>SendGrid API: POST /v3/marketing/singlesends
    SendGrid API-->>MCP Server: 201 Created (ID: ss_998877)
    MCP Server-->>ChatGPT: Result: Draft ss_998877 created
    ChatGPT->>MCP Server: create_a_send_grid_schedule_single_send (ID: ss_998877, "now")
    MCP Server->>SendGrid API: PUT /v3/marketing/singlesends/ss_998877/schedule
    SendGrid API-->>MCP Server: 200 OK
    MCP Server-->>ChatGPT: Result: Scheduled
    ChatGPT-->>User: "The campaign is scheduled and sending now."

Workflow 2: The Deliverability Auditor

A customer support representative receives a complaint that a user is not getting password reset emails. The agent needs to audit the contact's status and message history.

"Check if 'user@startup.com' is on any of our global suppression lists. Then, check the message activity logs to see if we've attempted to send them anything in the last 48 hours and what the exact status was."

Step-by-step execution:

1. get_single_send_grid_global_suppression_by_id: The agent checks the global suppression list. If the user accidentally marked a previous email as spam, they will appear here.
2. list_all_send_grid_messages: The agent constructs an SGQL query (to_email="user@startup.com") and fetches the activity logs for the past 48 hours.

Result: The agent reports back: "The user is not on the global suppression list. However, I found 2 message attempts in the last 48 hours that resulted in a 'deferred' status due to a temporary mailbox full error at their domain."

Security and Access Control

Exposing your email infrastructure to an LLM introduces significant risk. If an agent hallucinates, it could delete entire contact lists or send unauthorized campaigns. Truto mitigates this by enforcing strict configuration rules at the MCP server level.

• Method Filtering: Restrict servers to specific operational categories. Setting methods: ["read"] ensures the agent can query metrics and logs but cannot physically dispatch an email or delete a contact.
• Tag Filtering: Group tools by functional area. You can restrict an MCP server to only expose tools tagged as contacts and metrics, preventing the agent from touching settings or sender_verification endpoints.
• Require API Token Auth: By default, anyone with the MCP URL can interact with the server. Setting require_api_token_auth: true forces the client to pass a valid Truto session or API token in the headers, adding a secondary identity check.
• Time-to-Live (TTL): Use the expires_at property to create temporary servers. If you are generating an agent for a one-off audit task, setting the server to expire in 24 hours ensures stale access paths are automatically destroyed by the infrastructure.

Build Faster Agent Workflows

Connecting SendGrid to ChatGPT shouldn't require your engineering team to study legacy V3 API docs, manage async job polling, or write custom backoff logic for 429 rate limit headers. By abstracting the integration layer into a managed MCP server, Truto allows you to treat SendGrid as a native component of your LLM framework.

Instead of maintaining infrastructure, you can focus on writing better system prompts and orchestrating more intelligent marketing workflows.