Validate Payment Receipts with Artificial Intelligence

The problem: manual payment verification is slow and error-prone

If your business receives payments via SPEI (Mexico's interbank transfer system), you know the routine well. A customer sends you a screenshot or PDF of their payment receipt, and you need to confirm that the transfer actually went through. Maybe it happens once a day, maybe ten times, maybe a hundred.

The manual process is always the same. You open the Banxico verification site, copy the tracking number from the receipt, select the date, choose the sending and receiving banks, enter the destination CLABE account number, type in the exact amount, and submit the query. Then you repeat this for every single receipt.

Across dozens or hundreds of daily payments, this adds up to hours of repetitive work. And the risk of errors is high: one mistyped digit, a wrong date, or an incorrect bank selection and the verification fails, forcing you to start over.

The solution: an AI assistant that reads and verifies for you

Modern AI can process images and extract text from them with high accuracy. By combining this capability with Tlaloc's verification tools, you can automate the entire receipt validation process.

The concept is straightforward: you send your AI assistant (such as ChatGPT or Claude) an image of the payment receipt. The assistant reads it, extracts all the relevant data, and uses it to query Banco de Mexico directly. Within seconds, you get a clear answer about whether the payment is real and settled.

This works through MCP (Model Context Protocol), an open standard that lets AI assistants use external tools during a conversation. Tlaloc's MCP server gives the assistant direct access to Banxico's CEP (Comprobante Electronico de Pago) verification system.

How it works, step by step

The entire flow happens in four steps, all within a single conversation with your AI assistant:

1. You send the receipt image or PDF. You can attach a screenshot from your online banking, a photo of a printed receipt, or a PDF file. The assistant accepts any of these formats.
2. The assistant reads the image and extracts the data. Using its vision capabilities, it identifies and extracts the required fields: transaction date, tracking number, sending bank, receiving bank, destination CLABE, and transfer amount.
3. The assistant queries Banxico. With the extracted data, it uses Tlaloc's tools (via MCP) to verify the transfer against Banco de Mexico's official records.
4. You get a clear response. The assistant tells you whether the payment is legitimate, the exact amount, who sent it, who received it, and its current status.

All of this happens in the same chat window. No need to open another website, copy data, or fill out forms.

Sample prompt for your AI assistant

You can copy and use this prompt directly. Just attach the receipt image along with your message:

You can use this prompt with any AI assistant that supports MCP and vision (image processing). The most common ones today are ChatGPT and Claude. If your assistant has Tlaloc's MCP server configured, it will run the verification automatically.

What to expect from the response

After processing the image and querying Banxico, the assistant will respond with something like this:

If the transfer is not found or the data doesn't match, the assistant will let you know as well, pointing out which fields have discrepancies. This helps you detect altered receipts or payments that haven't settled yet.

Prerequisites

Before using this feature, you'll need the following:

• A Tlaloc account. Sign up for free at console.tlaloc.sh.
• Account balance. Each CEP verification costs $0.10 MXN. You can add funds via SPEI transfer from the console.
• An API key configured in your AI assistant. The API key connects your assistant to Tlaloc's tools.

If you haven't set up the MCP server yet, follow our step-by-step MCP configuration guide. You'll have everything ready in under five minutes.

Beyond a single receipt

Individual verification is just the starting point. Once you have the flow working, you can take it further:

• Validate multiple receipts in a single conversation. Send several images and ask the assistant to verify them all: "Here are 5 payment receipts. Verify each one and give me a summary of which are valid."
• Integrate into your collections workflow. When a customer reports a payment, send the receipt directly to the assistant to confirm before crediting their account.
• Set up validation rules. Ask the assistant to check specific conditions beyond basic verification: "Verify this receipt and confirm that the amount is at least $1,000 MXN and the recipient is Empresa Ejemplo SA de CV."
• Programmatic automation. For high volumes, you can use Tlaloc's API directly from your own system to verify payments without manual intervention. See the technical details of CEP verification for API parameters.

Next steps

If you haven't yet set up your AI assistant with Tlaloc, these resources will help you get started:

• Set up your AI assistant with Tlaloc's MCP server — step-by-step guide for ChatGPT, Claude, and other assistants.
• Learn the technical details of CEP verification — how Banxico's payment verification works and what data you need.