Connect Arketa to ChatGPT: Sync Class Schedules and Client Bookings

If you want to connect Arketa to ChatGPT to automate studio operations, sync class schedules, and handle client bookings, you need a Model Context Protocol (MCP) server. A managed MCP server translates ChatGPT's unstructured natural language requests into structured JSON-RPC calls that Arketa's REST API can process safely.

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

Giving a Large Language Model (LLM) read and write access to a fitness studio management platform is an engineering hazard. You have to map the LLM's flat arguments into complex polymorphic reservation schemas, handle proprietary cursor pagination, and enforce strict security boundaries so an AI agent does not accidentally wipe out a week of class schedules.

This guide breaks down exactly how to use Truto to generate a secure, authenticated MCP server for Arketa, connect it natively to ChatGPT, and execute complex booking workflows using natural language.

The Engineering Reality of the Arketa API

Building a custom integration layer for LLMs means you own the entire API lifecycle. While the open MCP standard provides a predictable way for models to discover tools, implementing it against Arketa's specific domain logic requires handling several architectural quirks.

Here are the specific integration challenges you face when exposing the Arketa API to an AI agent:

Polymorphic Reservation Schemas

Booking a client into a class seems like a straightforward POST request, but Arketa's reservation endpoint uses polymorphic payload requirements. If the client already exists in the system, the API requires a clientId and purchaseId. If the client is brand new, the API requires firstName, lastName, and email instead. An LLM natively struggles with conditional parameter requirements. Your MCP schema must explicitly instruct the LLM on which payload branches to use based on the context of the user's prompt, otherwise the API will reject the request with validation errors.

Destructive Operations Disguised as Updates

In many REST APIs, a PUT or PATCH to an existing resource simply updates standard fields. In Arketa, the update_a_arketa_reservation_by_id endpoint is actually a cancellation orchestrator. It does not allow general metadata updates. Instead, it expects boolean flags like isLateCancel to penalize a no-show, or issueRefund to return credits to the client. Exposing this tool to an LLM without strict method filtering or explicit descriptions risks accidental mass cancellations if the agent hallucinates an "update" intent.

Unforgiving Cursor Pagination

Arketa relies on a strict start_after cursor model for paginating through classes and clients. The API does not accept traditional page numbers or offsets. If you map this directly without context, ChatGPT will frequently attempt to guess page numbers (e.g., passing start_after: 2). Truto solves this at the schema generation level by dynamically injecting a next_cursor property into the JSON schema with strict, hardcoded instructions telling the LLM to pass the raw string value back completely unmodified.

Rate Limits and Edge Cases

A factual note on API rate limits: Truto does not retry, throttle, or apply automatic backoff on rate limit errors. When Arketa returns an HTTP 429 Too Many Requests error, Truto passes that error directly back to the caller. However, Truto does normalize upstream rate limit information into standardized headers (ratelimit-limit, ratelimit-remaining, ratelimit-reset) per the IETF spec. Your MCP client implementation is responsible for catching these errors and handling backoff logic before retrying the tool call.

Arketa to ChatGPT Quickstart Guide

To bridge the gap between ChatGPT and Arketa, we will use Truto to spin up an ephemeral, authenticated MCP server. This server dynamically generates tool definitions based on Arketa's documentation records and handles the JSON-RPC translation layer.

Step 1: Connect Arketa to Truto

Before you can expose Arketa tools, you must establish an authenticated connection to the underlying account.

1. Log into the Truto dashboard.
2. Navigate to Integrated Accounts -> New Integrated Account.
3. Select Arketa from the list of integrations and complete the authentication flow.
4. Make a note of your integrated_account_id. This ID maps the connection securely within Truto's multi-tenant architecture.

Step 2: Generate the Arketa MCP Server

Truto scopes MCP servers to a single integrated account. This ensures isolation - a URL generated for Studio A cannot access data for Studio B. You can generate this server via the Truto UI or programmatically via the API.

Option A: Via the Truto UI

1. Navigate to the Integrated Accounts page for your Arketa connection.
2. Click the MCP Servers tab.
3. Click Create MCP Server.
4. Configure the server name, select any necessary method filters (e.g., restricting to "read" only), and set an optional expiration date.
5. Copy the generated MCP server URL (it will look like https://api.truto.one/mcp/<token>).

Option B: Via the API

For developers orchestrating this programmatically, you can issue a POST request. The API validates the configuration, hashes the secure token, stores it in edge storage for ultra-fast reverse lookups, and returns the ready-to-use endpoint.

curl -X POST https://api.truto.one/integrated-account/$INTEGRATED_ACCOUNT_ID/mcp \
  -H "Authorization: Bearer $TRUTO_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Arketa Booking Agent",
    "config": {
      "methods": ["read", "write"],
      "tags": ["classes", "clients", "reservations"]
    }
  }'

The response contains the exact URL you need to hand to ChatGPT:

{
  "id": "mcp_abc123",
  "name": "Arketa Booking Agent",
  "config": {
    "methods": ["read", "write"],
    "tags": ["classes", "clients", "reservations"]
  },
  "expires_at": null,
  "url": "https://api.truto.one/mcp/a1b2c3d4e5f6..."
}

Step 3: Connect the MCP Server to ChatGPT

You have the server URL. Now you must point your client at it. There are two primary ways to do this depending on how you are using OpenAI's models.

Option A: Via the ChatGPT UI (Custom Connectors)

If you have a ChatGPT Pro, Plus, Enterprise, or Education account with Developer mode enabled:

1. In ChatGPT, navigate to Settings -> Apps -> Advanced settings.
2. Ensure Developer mode is enabled.
3. Under MCP servers / Custom connectors, click to add a new server.
4. Set the Name to "Arketa Studio Manager".
5. Paste the Truto MCP URL into the Server URL field.
6. Click Add. ChatGPT will immediately perform a protocol handshake, calling the initialize and tools/list methods to discover your Arketa capabilities.

Option B: Via Manual Configuration File

If you are using a custom desktop client, agent orchestration framework, or the official MCP inspector, you can connect via Server-Sent Events (SSE) using a standard JSON configuration.

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

When the client starts, it uses the @modelcontextprotocol/server-sse transport adapter to convert standard stdio MCP communication into the HTTP POST requests that Truto expects.

Security and Access Control

When handing an LLM the keys to a live database, boundaries are mandatory. Truto applies constraints at the token level, meaning the LLM physically cannot bypass them.

• Method Filtering (config.methods): Restrict the MCP server to specific categories of operations. Passing ["read"] ensures the LLM can list classes and fetch client data, but completely drops write tools like reservations from the generated schema.
• Tag Filtering (config.tags): Arketa operations are grouped logically. Passing ["classes"] strips out unrelated endpoints (like purchases or referrals), keeping the LLM focused and saving context window overhead.
• Time-to-Live (expires_at): You can assign an ISO datetime for temporary access. Truto automatically schedules a durable edge alarm to purge the token metadata and KV records exactly when the TTL hits.
• Strict Authentication (require_api_token_auth): By default, the cryptographically secure URL token is sufficient for access. Setting this flag to true injects a secondary middleware layer requiring the caller to also pass a valid Truto API token in the Authorization header.

Arketa Hero Tools for ChatGPT

Truto automatically generates tool definitions by analyzing Arketa's resource documentation. Here are the highest-leverage tools exposed to your agent.

list_all_arketa_classes

Retrieves a paginated list of studio classes. This is the foundational tool an agent uses to check schedule availability before attempting to book a client. The tool automatically handles Arketa's start_after cursor implementation.

"Show me all the upcoming yoga classes scheduled for tomorrow afternoon. Are there any spots left?"

create_a_arketa_reservation

Creates a reservation for a class. The agent must pass the class_id. The payload schema explicitly instructs the agent to supply either existing client credentials (clientId, purchaseId) or new client onboarding details (firstName, lastName, email).

"Book Jane Smith into the 5:00 PM Pilates class. She is a new client, her email is jane@example.com."

create_a_arketa_reservation_check_in

Marks an existing reservation as checked in. This is critical for front-desk automation, allowing an agent to quickly reconcile physical attendance against the booked roster.

"Mark John Doe as checked in for the Advanced HIIT class that started at 8:00 AM."

update_a_arketa_reservation_by_id

Despite the "update" nomenclature, this tool specifically handles reservation cancellations. It requires the id of the reservation and the class_id. The agent can specify isLateCancel and issueRefund to dictate business logic.

"Cancel Sarah's booking for tomorrow's spin class. It is outside the cancellation window, so mark it as a late cancel and do not issue a refund."

list_all_arketa_clients

Fetches a paginated roster of clients. Agents use this to look up internal clientId values required for booking existing users into new classes.

"Can you find the client ID for Marcus Johnson? I need to verify his active status."

list_all_arketa_purchases

Retrieves purchase history for clients. This allows the agent to verify if a user has an active membership, a drop-in pass, or a purchaseId valid for a specific class tier.

"Check the purchase history for client ID 1093. Do they have an active 10-class pack we can draw from?"

For the complete inventory of available Arketa tools and their underlying query and body schemas, visit the Arketa integration page.

Workflows in Action

Once connected, ChatGPT can orchestrate multi-step business logic by chaining Arketa tools together. Truto's JSON-RPC router extracts the flat arguments provided by the LLM and reliably routes them to the correct query and body parameters on Arketa's API.

Workflow 1: The Front Desk Waitlist Manager

When a popular class is full and a cancellation occurs, studio managers must quickly backfill the spot. An agent can automate this entirely.

"Cancel Michael's spot in the 6:00 PM Strength Training class and issue a refund. Then, find the first person on the waitlist and book them into the newly opened spot."

1. list_all_arketa_classes: The agent queries the schedule to retrieve the specific class_id for the 6:00 PM Strength Training class.
2. list_all_arketa_reservations: The agent pulls the roster for that class to locate Michael's reservation ID.
3. update_a_arketa_reservation_by_id: The agent executes the cancellation, setting isLateCancel to false and issueRefund to true.
4. list_all_arketa_reservations: The agent queries the class again, specifically checking for records carrying a waitlist status.
5. create_a_arketa_reservation: Using the clientId of the waitlisted user, the agent pushes a new booking request to fill the vacancy.

sequenceDiagram
  participant User as User
  participant ChatGPT as ChatGPT
  participant TrutoMCP as Truto MCP Server
  participant Arketa as Arketa API
  User->>ChatGPT: "Cancel Michael's spot..."
  ChatGPT->>TrutoMCP: Call update_a_arketa_reservation_by_id
  TrutoMCP->>Arketa: PUT /reservations/{id}
  Arketa-->>TrutoMCP: 200 OK (Cancelled)
  TrutoMCP-->>ChatGPT: JSON result
  ChatGPT->>TrutoMCP: Call list_all_arketa_reservations
  TrutoMCP->>Arketa: GET /reservations?class_id=XYZ
  Arketa-->>TrutoMCP: 200 OK (Waitlist data)
  TrutoMCP-->>ChatGPT: JSON result
  ChatGPT->>TrutoMCP: Call create_a_arketa_reservation
  TrutoMCP->>Arketa: POST /reservations
  Arketa-->>TrutoMCP: 201 Created
  TrutoMCP-->>ChatGPT: Success confirmation
  ChatGPT-->>User: "Michael is cancelled. Next user is booked."

Workflow 2: New Client Lead Conversion

Marketing often drives leads who want to try a class immediately. An agent can handle the end-to-end data entry and booking.

"A new lead named Alex Chen (alex@example.com) wants to drop into tomorrow morning's beginner yoga class. Create a profile for him and secure his spot."

1. list_all_arketa_classes: The agent fetches the schedule for the next day, filtering for the "beginner yoga" string to extract the correct class_id.
2. create_a_arketa_reservation: The agent detects that this is a new user workflow. It constructs a polymorphic payload containing the class_id, along with firstName, lastName, and email, bypassing the standard clientId requirement.

The user receives an immediate confirmation that the schedule has been queried, the class found, and the new profile initialized directly within the booking transaction.

Stop Writing Custom LLM Integration Code

Wiring an AI agent to a dynamic, production API like Arketa is notoriously difficult. If you try to build a custom MCP server in-house, your engineering team assumes the burden of maintaining auth state, tracking cursor pagination edge cases, and manually mapping API specs to JSON-RPC tools. Every time Arketa adds a feature, your server code rots.

Truto eliminates this technical debt. By deriving tool schemas dynamically from authoritative documentation and managing all token validation, routing, and pagination logic at the edge, Truto allows you to treat integrations as configuration rather than code. You generate the URL, hand it to your LLM framework, and start shipping features.