Connect Infor M3 to ChatGPT: Manage Finance, Orders, and Inventory

If you need to connect Infor M3 to ChatGPT to automate supply chain workflows, manage general ledger vouchers, or orchestrate warehouse inventory, you need a Model Context Protocol (MCP) server. This server acts as the translation layer between ChatGPT's tool calling capabilities and Infor M3's highly specific MI transaction APIs. You can either build and host this translation layer 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 Infor M3 to Claude or explore our broader architectural overview on connecting Infor M3 to AI Agents.

Giving a Large Language Model (LLM) read and write access to an enterprise ERP like Infor M3 (similar to how we connect Infor SyteLine to ChatGPT) is a massive engineering challenge. You must handle complex, multi-step transaction programs, manage strict division-level data contexts, and map rigid legacy data types to natural language arguments. Every time your M3 administrators expose a new MI program, your custom server code must be updated, redeployed, and tested.

This guide breaks down exactly how to use Truto to generate a secure, managed MCP server for Infor M3, connect it natively to ChatGPT, and execute complex ERP workflows using natural language.

The Engineering Reality of the Infor M3 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 Infor M3's highly rigid architecture is exceptionally painful.

If you decide to build a custom MCP server for Infor M3, you own the entire API lifecycle. Here are the specific integration challenges that break standard REST CRUD assumptions when working with M3:

MI Transaction Programs Instead of REST Resources

Unlike modern SaaS platforms with intuitive endpoints like POST /orders, Infor M3 exposes functionality via MI (Movex API) transaction programs. Operations are grouped into monolithic programs like OIS100MI (Customer Orders) or MMS200MI (Item Toolbox). To create an order, your LLM must know to execute OIS100MI/AddBatchHead. If you attempt to feed raw MI documentation to an LLM, it will quickly hallucinate parameters or use the wrong transaction program entirely. Your MCP server must meticulously map these transaction endpoints into clear, descriptive JSON-RPC tools with strictly defined schemas.

Strict Division and Facility Context

Nearly all data in Infor M3 is siloed by company (CONO), division (DIVI), and facility (FACI). When querying an item's warehouse balance, you cannot simply ask for "Item 12345" - you must provide the exact warehouse (WHLO). If your MCP server does not enforce or correctly prompt the LLM for these organizational parameters, M3 will reject the transaction or, worse, execute it in the wrong financial division. This requires building complex guardrails around how the AI agent accesses Master Data.

Multi-Step Data Creation Lifecycles

Creating a customer in Infor M3 is not a single API call. It requires executing CRS610MI/Add for basic data, followed by CRS610MI/AddAddress for location details, and potentially CRS610MI/ChgFinancial to configure billing parameters. If you expose these as raw, disconnected tools, ChatGPT will likely execute them out of order or miss required steps. Building an MCP server means you have to orchestrate these multi-step processes or provide incredibly detailed prompt instructions within the tool descriptions.

Handling API Rate Limits

When exposing Infor M3 to an autonomous AI agent, rate limiting becomes a critical architectural concern. Agents can rapidly execute loops (e.g., retrieving thousands of individual inventory records) that overwhelm the ERP.

It is important to understand how rate limits are handled at the integration layer. Truto does not retry, throttle, or apply backoff on rate limit errors. When the upstream Infor M3 API returns an HTTP 429 error, Truto passes that error directly to the caller. Truto normalizes the upstream rate limit information into standardized headers (ratelimit-limit, ratelimit-remaining, ratelimit-reset) per the IETF specification. The caller - in this case, your AI agent framework or ChatGPT configuration - is strictly responsible for interpreting these headers and executing its own retry and exponential backoff logic.

Generating the Infor M3 MCP Server

Rather than building the JSON-RPC infrastructure from scratch, you can use Truto to dynamically generate an MCP server scoped specifically to an authenticated Infor M3 environment.

Truto derives MCP tools directly from the underlying API documentation and resources. If a transaction program is documented, it becomes an AI-ready tool.

There are two ways to generate your MCP server URL: via the Truto UI, or programmatically via the API.

Method 1: Via the Truto UI

This is the fastest method for internal development or ad-hoc agent testing.

1. Log into your Truto dashboard and navigate to the Integrated Accounts page.
2. Select your connected Infor M3 account.
3. Click on the MCP Servers tab.
4. Click Create MCP Server.
5. Select your desired configuration. You can filter the server to only allow read methods, or restrict it to specific tags like inventory or finance.
6. Click Save and copy the generated MCP server URL (e.g., https://api.truto.one/mcp/a1b2c3d4...).

Method 2: Via the API

For production use cases where you are provisioning agents programmatically, you can generate the MCP server via a REST call. This provisions a secure token and returns the 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": "Infor M3 Agent Server",
    "config": {
      "methods": ["read", "write", "custom"],
      "tags": ["finance", "orders", "inventory"]
    }
  }'

The API will return a response containing the self-contained MCP url. Treat this URL as a sensitive credential - it contains the cryptographic token required to route and authenticate the JSON-RPC requests.

Connecting the MCP Server to ChatGPT

Once you have your Truto MCP URL, you need to register it with ChatGPT. You can do this through the ChatGPT user interface or via a local configuration file for desktop development.

Method A: Via the ChatGPT UI

If you are using ChatGPT Pro, Plus, Business, Enterprise, or Education, you can connect remote MCP servers directly in the web app.

1. Open ChatGPT and navigate to Settings -> Apps -> Advanced settings.
2. Ensure Developer mode is enabled.
3. Under the MCP servers or Custom connectors section, click to add a new server.
4. Enter a recognizable name (e.g., "Infor M3 ERP").
5. Paste the Truto MCP URL into the Server URL field.
6. Click Add or Save.

ChatGPT will immediately perform a handshake with the Truto MCP router, fetch the tool definitions, and make them available in your chat sessions.

Method B: Via Manual Config File

If you are running local experiments or using the ChatGPT macOS app in a developer environment, you can use the standard Server-Sent Events (SSE) proxy configuration.

Create or update your MCP configuration JSON file:

{
  "mcpServers": {
    "infor_m3": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-sse",
        "--url",
        "https://api.truto.one/mcp/<YOUR_TOKEN_HERE>"
      ]
    }
  }
}

Infor M3 Hero Tools for ChatGPT

When the MCP server is initialized, it exposes Infor M3 MI transactions as distinct tools. Here are the highest-leverage tools available for orchestrating ERP workflows.

Add Standard Voucher (Finance)

Tool: infor_m_3_gls_015_mi_standard_voucher_add

This tool executes the GLS015MI/Add transaction to create a standard financial voucher in the general ledger. It requires the accounting string and division context.

"Create a new standard voucher for division 100, year 2024. The voucher amount is $5000 and it should be booked against account code 4010."

Create New Customer Record (CRM)

Tool: infor_m_3_crs_610_mi_customer_add

This tool executes CRS610MI/Add to create the master data for a new customer. Note that in M3, this establishes the basic identity; addresses and financial data are often handled in subsequent MI calls.

"Create a new customer record for 'Acme Corp'. Set their customer type to 'Retail' and assign them to sales group 'NA-EAST'."

Create Batch Order Head (Sales)

Tool: infor_m_3_ois_100_mi_customer_order_add_batch_head

This tool executes OIS100MI/AddBatchHead. It is the first step in the customer order lifecycle, creating the order header before lines are attached.

"Start a new customer order for customer ID 'CUST-8890'. Set the facility to 'FAC-01' and the requested delivery date to next Friday."

Get Item Warehouse Balance (Inventory)

Tool: infor_m_3_mms_200_mi_item_toolbox_get_itm_whs_bal

This tool executes MMS200MI/GetItmWhsBal to retrieve the real-time stock balance, allocated quantity, and available-to-promise (ATP) metrics for a specific item in a specific warehouse.

"Check the current warehouse balance for item 'WIDGET-X' in warehouse 'WH-MAIN'. I need to know the available quantity minus current allocations."

Confirm Purchase Order Line (Procurement)

Tool: infor_m_3_pps_001_mi_purchase_order_confirm_line

Executes PPS001MI/ConfirmLine to acknowledge and confirm a specific line on a supplier purchase order, moving it to the next stage in the procurement lifecycle.

"Confirm line 10 on purchase order 'PO-55432'. The confirmed quantity is 100 units."

Approve Supplier Invoice (AP)

Tool: infor_m_3_aps_110_mi_recode_and_approve_supplier_approve_invoice

Executes APS110MI/ApproveInvoice to authorize a supplier invoice for payment after goods receipt and reconciliation have occurred.

"Approve supplier invoice 'INV-9988' from supplier 'SUP-001' for payment processing in division 100."

Confirm Picking List (Warehouse)

Tool: infor_m_3_mws_420_mi_picking_list_confirm

Executes MWS420MI/Confirm to record that items have been physically picked from the warehouse location, finalizing the outbound logistics stage for those goods.

"Confirm the picking list for delivery number 'DEL-88776'. All items were picked as requested."

To see the full schema details, required parameters, and the complete inventory of supported Infor M3 MI transactions, visit the Infor M3 integration page.

Workflows in Action

With Truto handling the complex MI program mapping and authentication, ChatGPT can act as an autonomous ERP operator. Here is how multi-step workflows execute in practice.

sequenceDiagram
    participant User as User
    participant ChatGPT as ChatGPT
    participant TrutoMCP as Truto MCP Server
    participant M3 as Infor M3 API

    User->>ChatGPT: "Check stock for WIDGET-A and create an order if we have 50+ units"
    ChatGPT->>TrutoMCP: call get_itm_whs_bal
    TrutoMCP->>M3: MMS200MI/GetItmWhsBal
    M3-->>TrutoMCP: Returns 120 units
    TrutoMCP-->>ChatGPT: Balance result
    ChatGPT->>TrutoMCP: call add_batch_head
    TrutoMCP->>M3: OIS100MI/AddBatchHead
    M3-->>TrutoMCP: Order number ORD-123
    TrutoMCP-->>ChatGPT: Order created
    ChatGPT->>TrutoMCP: call add_batch_line
    TrutoMCP->>M3: OIS100MI/AddBatchLine
    M3-->>TrutoMCP: Line added
    TrutoMCP-->>ChatGPT: Success
    ChatGPT-->>User: "Order ORD-123 created for 50 units."

Scenario 1: Quote-to-Order Automation

Customer service representatives often waste time manually checking stock before keying in orders. ChatGPT can orchestrate this entire check-and-create process.

"Look up the warehouse balance for item 'WIDGET-A' in warehouse 'MAIN'. If we have more than 50 units available, create a new customer order for customer 'CUST-100' and add an order line for 20 units."

1. ChatGPT calls infor_m_3_mms_200_mi_item_toolbox_get_itm_whs_bal to retrieve the current inventory levels.
2. Recognizing sufficient stock, it calls infor_m_3_ois_100_mi_customer_order_add_batch_head to generate the new order header.
3. It extracts the new order number from the response and calls infor_m_3_ois_100_mi_customer_order_add_batch_line to attach the 20 units of 'WIDGET-A' to the order.
4. The user receives confirmation of the created order number.

Scenario 2: AP Invoice Approval and Journaling

Finance teams can use ChatGPT to review pending payables and immediately book the associated general ledger entries.

"List the supplier invoice headers for supplier 'SUP-404'. Approve invoice 'INV-999' for payment, then create a standard voucher in the GL for $2500 against account 6020."

1. ChatGPT calls aps_260_mi_supplier_invoice_header_display_lst to verify the invoice exists and is pending.
2. It calls infor_m_3_aps_110_mi_recode_and_approve_supplier_approve_invoice to release the invoice in the AP subledger.
3. Finally, it calls infor_m_3_gls_015_mi_standard_voucher_add to record the corresponding journal entry in the general ledger.
4. The user is presented with the voucher number and approval confirmation.

Security and Access Control

Granting an LLM access to an ERP requires strict governance. Truto's MCP servers provide granular configuration controls to limit the blast radius of AI agents:

• Method Filtering (config.methods): Restrict the server to specific operation types. Setting this to ["read"] ensures the agent can execute Lst and Get MI programs, but strictly blocks Add, Upd, or Del programs.
• Tag Filtering (config.tags): Scope access by functional area. You can restrict an agent to only see tools tagged with inventory, preventing it from accessing finance or hr transactions.
• Secondary Authentication (require_api_token_auth): When set to true, possessing the MCP URL is not enough. The client must also pass a valid Truto API token in the Authorization header, preventing unauthorized access if the URL is leaked.
• Ephemeral Servers (expires_at): You can set an exact ISO datetime for the server to expire. Once expired, the underlying infrastructure automatically cleans up the tokens, terminating access. This is ideal for granting a contractor temporary reporting capabilities.

Automate the ERP Back-Office with Truto

Infor M3's MI transaction architecture was not built for natural language interfaces. Building a custom MCP server means taking on the burden of translating LLM JSON into rigid, mainframe-era payloads, maintaining complex authentication lifecycles, and managing enterprise rate limits.

By using Truto, you offload the entire integration layer. You get dynamic, documentation-driven MCP tools that automatically adapt as new MI programs are enabled. Your engineers can focus on building intelligent agent workflows, while Truto handles the heavy lifting of enterprise ERP connectivity.