Connect AccountsIQ to ChatGPT: Automate Financials & GL Journals

If you want to connect AccountsIQ to ChatGPT so your AI agents can automate general ledger entries, pull trial balances, investigate AP/AR discrepancies, and manage stock items, you need a Model Context Protocol (MCP) server.

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

Giving a Large Language Model (LLM) read and write access to an enterprise financial system like AccountsIQ is an engineering hurdle. The platform relies heavily on legacy SOAP architecture. You either spend weeks building, hosting, and maintaining a custom MCP server to translate LLM JSON arguments into strict XML payloads, or you use a managed infrastructure layer.

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

The Engineering Reality of the AccountsIQ API

A custom MCP server is a self-hosted translation layer. While the open MCP standard provides a predictable way for models to discover tools using JSON-RPC, implementing AI-to-ERP integrations against AccountsIQ's SOAP-heavy API surface is exceptionally painful for modern developers.

If you decide to build a custom MCP server for AccountsIQ, you own the entire integration lifecycle. Here are the specific engineering challenges that break standard REST assumptions when working with this financial platform:

The SOAP Envelope and XML Serialization

AccountsIQ relies on SOAP 1.1 operations. When an LLM wants to pull a trial balance, it naturally wants to pass a flat JSON object like {"pointInTime": "2024-10-31"}. However, the upstream API expects a strictly typed XML payload wrapped in a SOAP envelope, with a specific xmlns namespace (http://www.visorsoftware.com/visor/accountsiq/dashboard/Integration/).

Your custom MCP server must serialize the LLM's JSON into XML, inject the required AiqSoapHeader, attach the exact SOAPAction header to the HTTP request, and then parse the verbose XML response back into a clean JSON structure that the LLM can understand without blowing up its context window.

Token Management Inside the Payload

Unlike modern APIs that accept Bearer tokens in the Authorization header, many AccountsIQ SOAP operations require passing a session <token> directly inside the XML body. Your server must intercept the tool call, securely retrieve the integration token, and inject it into the correct node of the XML tree before forwarding the request.

Nested Result Wrappers

When you ask AccountsIQ for a list of active suppliers via the GetActiveSupplierList operation, you do not simply get an array of suppliers back. You receive a heavily nested XML response containing a Status, ErrorCode, ErrorMessage, HasExpired flag, and finally a Result element that holds the actual data. If you pass this raw structure back to ChatGPT, the model wastes reasoning cycles navigating boilerplate status codes. A production-grade MCP server must flatten this response.

Truto abstracts all of this away. The platform dynamically generates MCP tools based on the API's documentation, automatically translating the LLM's flat JSON input into valid SOAP envelopes, injecting the token, and returning normalized JSON responses.

Creating the AccountsIQ MCP Server

To connect ChatGPT to AccountsIQ, you first need to generate a self-contained MCP server URL. This URL holds a cryptographic token that handles authentication and routing. You can create this server in two ways.

Method 1: Via the Truto Dashboard

For quick prototyping or manual setup, use the dashboard:

1. Log into Truto and navigate to Integrated Accounts.
2. Connect a new AccountsIQ instance (or select an existing one). Truto handles the credential exchange.
3. Click the MCP Servers tab on the account detail page.
4. Click Create MCP Server.
5. Select your desired configuration (e.g., restrict to read methods only, or filter by specific tags like gl_accounts and journals).
6. Copy the generated MCP server URL (e.g., https://api.truto.one/mcp/a1b2c3d4...).

Method 2: Via the Truto API

For automated provisioning - like generating a unique MCP server for every tenant in your AI application - use the API. This endpoint scopes the server to a specific integrated account and applies your security filters.

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": "AccountsIQ GL Automation",
    "config": {
      "methods": ["read", "write"],
      "tags": ["gl_accounts", "journals", "invoices", "suppliers"]
    }
  }'

The API responds with the secure URL:

{
  "id": "mcp-789-xyz",
  "name": "AccountsIQ GL Automation",
  "config": { 
    "methods": ["read", "write"],
    "tags": ["gl_accounts", "journals", "invoices", "suppliers"]
  },
  "expires_at": null,
  "url": "https://api.truto.one/mcp/a1b2c3d4e5f67890"
}

Treat this URL like a secret. It contains everything a client needs to discover tools and execute SOAP operations against the connected AccountsIQ instance.

Connecting the MCP Server to ChatGPT

Once you have your Truto MCP URL, you can plug it into ChatGPT.

Method A: Via the ChatGPT UI

If you are using ChatGPT Pro, Plus, Business, Enterprise, or Education, you can add custom connectors directly in the web interface:

1. Open ChatGPT and navigate to Settings -> Apps -> Advanced settings.
2. Toggle on Developer mode.
3. Click Add custom connector under the MCP servers section.
4. Name the connection (e.g., "AccountsIQ Financials").
5. Paste the Truto MCP URL into the Server URL field.
6. Click Add.

ChatGPT will immediately ping the server's /initialize and /tools/list endpoints, discovering all available AccountsIQ operations based on your configured filters.

Method B: Via Manual Config File

If you are deploying a custom headless AI agent, or using an MCP-compatible client that requires manual configuration (like Claude Desktop or standard local runners), you can connect via a JSON config file. Because Truto provides a remote HTTPS endpoint, you use a Server-Sent Events (SSE) adapter to bridge the standard MCP stdio protocol to the remote URL.

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

When your agent boots up, the SSE client proxy connects to the Truto URL and handles the JSON-RPC lifecycle automatically.

Security and Access Control

Giving an LLM access to corporate accounting data requires strict guardrails. Truto's MCP architecture enforces security at the token level, ensuring the model cannot accidentally drop tables or query unauthorized ledgers.

• Method Filtering: The config.methods array restricts operations by type. Passing ["read"] limits the server to get and list operations, ensuring the agent cannot execute create or update SOAP calls.
• Tag Filtering: The config.tags array scopes the available API surface. By restricting tags to ["gl_accounts"], you prevent the LLM from hallucinating calls to the HR or procurement modules.
• Secondary Authentication (require_api_token_auth): If this flag is set to true, possession of the MCP URL is not enough. The client must also pass a valid Truto API token in the Authorization header, forcing the AI agent to run in a secure, authenticated backend rather than a public client.
• Time-to-Live (expires_at): You can set an exact ISO datetime for the server to self-destruct. Truto uses distributed key-value storage with TTLs and durable alarms to ensure the endpoint vanishes at the precise expiration time, leaving no stale access vectors.

Handling Rate Limits and Pagination Context

AI agents are notoriously aggressive when polling APIs or scraping paginated lists.

Factual note on rate limits: Truto does not retry, throttle, or apply backoff on rate limit errors. When the upstream AccountsIQ API returns an 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.

Your AI agent code is strictly responsible for inspecting these headers, handling retry logic, and implementing exponential backoff.

Additionally, Truto handles the complex SOAP pagination cursors by injecting limit and next_cursor properties into the query schemas during tool generation. The tool descriptions explicitly instruct the LLM to pass cursor values back unchanged, preventing the model from hallucinating invalid offset numbers.

AccountsIQ Hero Tools

Truto automatically generates descriptive snake_case tool names from the underlying AccountsIQ integration documentation. Because the LLM receives a flat input namespace, query parameters and body arguments are combined into a single JSON object, which Truto maps back into the correct SOAP payload.

Here are the highest-leverage financial tools for your AI agents. View the full tool inventory on the AccountsIQ integration page.

List General Ledger Accounts

Tool Name: accounts_iq_gl_accounts_get_list Runs the GetGLAccountList SOAP operation. This is foundational for any agent doing financial analysis, as it maps account codes to human-readable names and categories.

"Fetch the full list of GL accounts from AccountsIQ. Give me a markdown table of the account codes, names, and their sub-categories."

Create a General Journal

Tool Name: accounts_iq_general_journals_create_general_journal Executes the CreateGeneralJournal operation. Useful for automated month-end adjustments or reclassifying expenses based on AI analysis.

"Create a new general journal entry for $500. Debit the 'Software Subscriptions' account and credit 'Prepaid Expenses', using today's date and the description 'Monthly SaaS Amortization'."

Get Trial Balance for Period

Tool Name: accounts_iq_trial_balances_get_for_period Calls the GetTrialBalanceForPeriod endpoint. This tool requires a specific date and returns the foundational accounting data needed for automated health checks.

"Pull the trial balance for the period ending October 31st. Identify any accounts with unexpected negative balances and summarize the total debits versus credits."

Query Invoices by Criteria

Tool Name: accounts_iq_invoices_get_by Executes the GetInvoicesBy SOAP operation to query the ledger. You can filter by date ranges, status codes, or specific account IDs.

"Find all sales invoices generated in the last 7 days that are currently marked as unpaid. Calculate the total outstanding revenue."

List Active Suppliers

Tool Name: accounts_iq_suppliers_get_active_list Triggers GetActiveSupplierList. Essential for AP automation workflows where the agent needs to verify vendor codes before routing invoices.

"Get the list of all active suppliers in the system. Check if we have an active vendor profile for 'Acme Corp' and return their supplier code."

Post (Finalize) an Invoice

Tool Name: accounts_iq_invoices_post Commits an invoice to the ledger using the PostInvoice operation. This transitions a draft invoice into a finalized financial record.

"Take invoice ID 45902 and post it to the ledger so it becomes finalized. Confirm if the operation succeeded or if there were validation errors."

For a complete list of tools, including stock items, budgets, tax codes, and currency exchange rates, visit the AccountsIQ Integration detail page.

Workflows in Action

Connecting an LLM to an ERP isn't just about reading data - it's about chaining operations to replace manual accounting tasks. Here are two real-world workflows you can execute via ChatGPT.

Scenario 1: Month-End Trial Balance & GL Review

Finance teams spend hours manually reviewing trial balances for misclassifications. An AI agent can pull the data, analyze it, and propose adjusting journal entries in seconds.

"Pull the trial balance for the month of November. Check the GL accounts for 'Office Supplies' and 'Software Subscriptions'. If any software expenses were accidentally booked to 'Office Supplies', draft a general journal entry to reclassify them and ask for my approval before creating it."

Execution Steps:

1. The agent calls accounts_iq_trial_balances_get_for_period for November.
2. The agent calls accounts_iq_gl_accounts_get_list to map the specific account codes for supplies and software.
3. It identifies a $1,200 AWS charge sitting in Office Supplies.
4. It asks the user for approval.
5. Upon approval, it calls accounts_iq_general_journals_create_general_journal to post the reclassification entry.

sequenceDiagram
  participant ChatGPT as ChatGPT
  participant Truto as Truto MCP Server
  participant AIQ as AccountsIQ SOAP API
  
  ChatGPT->>Truto: Call accounts_iq_trial_balances_get_for_period
  Truto->>AIQ: Inject XML Envelope & SOAPAction
  AIQ-->>Truto: Return nested XML response
  Truto-->>ChatGPT: Return flat JSON Result
  
  ChatGPT->>Truto: Call accounts_iq_gl_accounts_get_list
  Truto->>AIQ: Inject XML Envelope & SOAPAction
  AIQ-->>Truto: Return XML
  Truto-->>ChatGPT: Return JSON
  
  Note over ChatGPT: Agent analyzes data & detects error<br>User approves reclassification
  
  ChatGPT->>Truto: Call accounts_iq_general_journals_create_general_journal
  Truto->>AIQ: Inject XML Envelope & Token
  AIQ-->>Truto: Return Success XML
  Truto-->>ChatGPT: Return JSON confirmation

Scenario 2: AP Invoice Validation and Posting

When raw invoice data arrives, it often sits in a draft state until an accountant verifies the supplier and posts it. An AI agent can handle this automatically.

"Check for any draft purchases invoices submitted today. Verify that the supplier exists in our active vendor list. If the vendor is active and the total amount is under $5,000, post the invoice to the ledger."

Execution Steps:

1. The agent calls accounts_iq_invoices_get_by to find today's draft purchases invoices.
2. For each invoice, it extracts the supplier code.
3. It calls accounts_iq_suppliers_get_active_list to verify the code matches a valid, active vendor.
4. It evaluates the amount logic (under $5,000).
5. It iterates through the valid invoices and calls accounts_iq_invoices_post to finalize them on the ledger.

Strategic Wrap-up

Forcing LLMs to parse legacy SOAP architectures is a recipe for hallucinations, excessive token usage, and integration failure. By deploying Truto's managed MCP server, you offload the complex XML serialization, authentication injection, and nested payload flattening to a dedicated infrastructure layer.

This architecture allows your AI agents to interact with AccountsIQ using clean, predictable JSON-RPC tool calls, enabling you to build automated GL workflows, smart AP/AR reconciliation, and financial reporting bots without writing a single line of SOAP parsing logic.