Connect Wealthbox to ChatGPT: Manage Contacts, Tasks, and Workflows

If you need to connect Wealthbox to ChatGPT to manage client contacts, track opportunities, or automate financial advisory workflows, you need a Model Context Protocol (MCP) server. This server acts as the translation layer between ChatGPT's tool calls and Wealthbox's REST APIs. If your team uses Claude, check out our guide on connecting Wealthbox to Claude or explore our broader architectural overview on connecting Wealthbox to AI Agents.

Giving a Large Language Model (LLM) read and write access to a specialized CRM like Wealthbox is an engineering challenge. You have to handle OAuth 2.0 token lifecycles, map complex relationship graphs to MCP tool definitions, and deal with strict pagination limits. Every time the API updates or a custom field is changed, 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 Wealthbox, connect it natively to ChatGPT, and execute complex workflows using natural language.

The Engineering Reality of the Wealthbox 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, implementing it against domain-specific APIs is painful. If you decide to build a custom MCP server for Wealthbox, you own the entire API lifecycle. Here are the specific integration challenges that break standard CRUD assumptions when working with Wealthbox:

Complex Household and Relationship Graphs Wealthbox is built for financial advisors, which means its data model is deeply relational. You are rarely just dealing with isolated contacts. A contact is part of a household, has specific contact roles, and is tied to teams. Modifying a household requires hitting specific household_member endpoints, not just updating a contact record. If an LLM attempts to add a spouse to a client profile by updating the primary contact's custom fields, the request will fail. Your MCP server must explicitly expose the household graph logic to the model.

Polymorphic Activity Streams When an advisor asks ChatGPT, "What happened with the Smith account this week?" the LLM needs to query the activity stream. The Wealthbox Activity Stream API returns a highly polymorphic array of events - notes, tasks, events, opportunities, and emails are all mixed together, and each object has different nested fields. Exposing this to an LLM requires carefully tuned JSON Schemas that instruct the model on how to parse the varying payloads without hallucinating fields that belong to a task onto an email record.

Rate Limits and 429 Errors Wealthbox enforces rate limits to protect infrastructure. If your AI agent gets stuck in a recursive loop or tries to paginate through a massive contact directory too quickly, the upstream API will reject the request with an HTTP 429 error. Truto standardizes upstream rate limit information into IETF-compliant headers (ratelimit-limit, ratelimit-remaining, ratelimit-reset). Truto passes the 429 error directly back to the caller. It does not retry, throttle, or apply backoff automatically. Your MCP server or the LLM orchestrator must be capable of reading these headers and executing exponential backoff, otherwise the LLM will assume the action succeeded when it actually failed.

Deploying the Wealthbox MCP Server

Instead of building a proxy layer from scratch, you can use Truto to dynamically generate an MCP server. Truto reads the API documentation, generates JSON Schemas, handles token refreshes, and outputs a secure JSON-RPC 2.0 endpoint.

Step 1: Create the MCP Server

You can generate the MCP server for a connected Wealthbox account using either the Truto Dashboard or the API.

Method A: Via the Truto UI

1. Navigate to the integrated account page for your Wealthbox connection in the Truto dashboard.
2. Click the MCP Servers tab.
3. Click Create MCP Server.
4. Select your desired configuration (e.g., restrict to read-only tools, or apply specific tags).
5. Copy the generated MCP server URL (e.g., https://api.truto.one/mcp/abc123def...).

Method B: Via the API You can programmatically generate a server URL by making a POST request to the Truto API. This is ideal if you are deploying AI agents for multiple tenants.

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": "Wealthbox Advisor Agent",
    "config": {
      "methods": ["read", "write"],
      "tags": ["crm", "tasks"]
    }
  }'

The response returns the secure URL that ChatGPT will use to discover and invoke tools:

{
  "id": "mcp-789",
  "name": "Wealthbox Advisor Agent",
  "url": "https://api.truto.one/mcp/a1b2c3d4e5f6..."
}

Step 2: Connect ChatGPT to the MCP Server

Once you have the URL, you need to register it with ChatGPT. You can do this through the UI for simple use cases, or via a configuration file for headless agents.

Method A: Via the ChatGPT UI

1. Open ChatGPT and navigate to Settings -> Apps -> Advanced settings.
2. Enable Developer mode to expose MCP functionality.
3. Under MCP servers / Custom connectors, click Add new server.
4. Enter a name like "Wealthbox CRM" and paste the Truto MCP URL.
5. Save the configuration. ChatGPT will immediately perform an initialize handshake and load the Wealthbox tools.

Method B: Via Manual Config File If you are using Claude Desktop or managing a custom server environment, you can map the MCP server using an SSE transport configuration file. You will use the official MCP SSE wrapper:

{
  "mcpServers": {
    "wealthbox_crm": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-sse",
        "--url",
        "https://api.truto.one/mcp/a1b2c3d4e5f6..."
      ]
    }
  }
}

The Execution Architecture

When ChatGPT calls a tool, the data flow is strictly managed without caching the underlying CRM data.

sequenceDiagram
    participant LLM as ChatGPT
    participant MCP as Truto MCP Server
    participant Core as Truto Proxy API
    participant API as Wealthbox API

    LLM->>MCP: JSON-RPC call (create_a_wealthbox_task)
    MCP->>MCP: Validate Token & Parse Flat Schema
    MCP->>Core: Inject OAuth Token
    Core->>API: POST /v1/tasks (Mapped Body)
    API-->>Core: 201 Created (or 429 Too Many Requests)
    Core-->>MCP: Raw JSON Response
    MCP-->>LLM: JSON-RPC Result Content

Hero Tools for Wealthbox Automation

Truto automatically maps Wealthbox endpoints into discrete, descriptive tools. Here are the highest-leverage tools available for your AI agents.

list_all_wealthbox_contacts

This tool retrieves the contact directory. It supports query parameters for filtering by name, email, or tag, which is critical because LLMs cannot ingest an entire database at once. The generated schema explicitly instructs the LLM on how to pass pagination cursors back unchanged.

"Find the contact record for Sarah Jenkins. Search using her last name and retrieve her active profile ID."

get_single_wealthbox_contact_by_id

Retrieves the full, deeply nested schema for a specific contact. This is the primary tool used to fetch background context before drafting an email or preparing for a client review. It returns all custom fields associated with the client profile.

"Fetch the full contact record for ID 892314. Summarize her employment history and current household status."

list_all_wealthbox_activity_stream

This tool pulls the chronological history of a contact, team, or workspace. Because the objects are polymorphic, the tool description guides the LLM to inspect the event_type field before parsing the contents of the payload.

"Pull the activity stream for the Smith Household over the last 30 days. Did anyone from our team log a call note regarding their estate plan?"

create_a_wealthbox_note

Notes are the lifeblood of a CRM. This tool allows the LLM to write back to Wealthbox. It takes a body schema that includes the raw text of the note and the associated contact_id or opportunity_id.

"I just finished a 30-minute review call with Mark. Write a summary note detailing our discussion about tax-loss harvesting and save it to his Wealthbox record."

create_a_wealthbox_task

Automates workflow handoffs. The LLM can use this tool to assign follow-up actions to specific users or teams, setting due dates and priorities based on natural language context.

"Create a task assigned to the compliance team to review the updated KYC documents for the Johnson account. Set the due date for next Tuesday."

list_all_wealthbox_opportunities

Exposes the sales and pipeline mechanics of Wealthbox. The LLM can retrieve opportunities to generate pipeline summaries or check the status of pending asset transfers.

"List all active opportunities in the 'Asset Transfer' pipeline. Which ones have been stuck in the pending stage for more than 14 days?"

create_a_wealthbox_household_member

Allows the LLM to safely manipulate relationship graphs. Instead of hacking a contact record, it uses the dedicated resource to link a new or existing contact to a household, requiring specific title mappings.

"Add Jane Doe to the Doe Household record in Wealthbox. Her role title is 'Spouse' and her existing contact ID is 99812."

For the complete inventory of available Wealthbox tools, endpoints, and schema details, visit the Wealthbox integration page.

Workflows in Action

Connecting Wealthbox to ChatGPT unlocks agentic workflows that save hours of administrative overhead. Here is how specific personas utilize these tools in production.

Scenario 1: The Client Review Briefing

An advisor is preparing for an annual review with a high-net-worth client and needs a quick, comprehensive summary of all recent interactions.

"I have a meeting with David Chen in 10 minutes. Get his profile, summarize the last 5 things in his activity stream, and check if we have any open opportunities pending for his account."

Execution Flow:

1. The agent calls list_all_wealthbox_contacts with name=David Chen to retrieve his specific contact ID.
2. The agent calls list_all_wealthbox_activity_stream filtering by David's ID to fetch the latest logged emails, notes, and tasks.
3. The agent calls list_all_wealthbox_opportunities to see if there are pending pipeline items tied to his record.
4. ChatGPT synthesizes the JSON responses into a clean, bulleted briefing document.

Scenario 2: Post-Meeting Task Orchestration

Following a client call, an advisor dumps raw voice-to-text notes into ChatGPT and expects the CRM to be updated seamlessly.

"Here is the transcript from my call with Sarah. Extract the key takeaways, save them as a note on her profile, and create a task for the operations team to send her the new beneficiary forms by Friday."

Execution Flow:

1. The agent parses the provided text and calls list_all_wealthbox_contacts to find Sarah's CRM ID.
2. The agent calls create_a_wealthbox_note using the ID and the extracted summary.
3. The agent calls create_a_wealthbox_task, formatting the due date programmatically for Friday and assigning it to the operations team ID.
4. ChatGPT confirms that the note is logged and the compliance task is queued.

Scenario 3: Pipeline Management via Email Summaries

A team lead wants to know the status of a specific opportunity without leaving their chat interface.

"What is the latest status on the 'Q3 401k Rollover' opportunity? Check the activity stream for that specific deal and tell me who touched it last."

Execution Flow:

1. The agent calls list_all_wealthbox_opportunities and filters the response array for the specific rollover deal name to extract the opportunity ID.
2. The agent calls list_all_wealthbox_activity_stream scoped to that opportunity ID.
3. ChatGPT reads the most recent event, identifies the user ID attached to the event, and summarizes the latest status update for the team lead.

Security and Access Control

Giving an LLM access to sensitive financial CRM data requires strict guardrails. Truto's MCP servers provide granular access control at the infrastructure level.

• Method Filtering: You can restrict a server entirely. By passing config: { methods: ["read"] } during creation, the LLM will only receive get and list tools. It becomes physically impossible for the model to hallucinate a create_a_wealthbox_contact call.
• Tag Filtering: Limit the server scope to specific functional domains. If you apply tags: ["crm"], the server will exclude tools related to internal user management or global account settings.
• Require API Token Auth: By default, possessing the MCP URL grants access. By setting require_api_token_auth: true, the client must also pass a valid Truto API token in the Authorization header. This guarantees that even if the URL leaks, the endpoint remains secure.
• Expiring Servers: Generate ephemeral access. By passing an expires_at ISO datetime string, the KV storage and database entries will automatically self-destruct via a scheduled durable object alarm. This is perfect for granting temporary AI agent access during a specific batch processing job.

Moving Past Manual CRM Entry

Connecting Wealthbox to ChatGPT shifts your workflow from data entry to data orchestration. Instead of clicking through five different screens to log a note, update a pipeline stage, and assign a task, your AI agent can execute the entire sequence through natural language via standard JSON-RPC tool calls.

By leveraging Truto's dynamic, documentation-driven MCP generation, you avoid the technical debt of maintaining schemas, handling OAuth refreshes, and building backoff logic from scratch. You get a production-ready integration layer that scales with your workflows.