---
title: "How Ramp Fixes Merchant Matches with AI"
description: "Ramp describes the difficulty of mapping payment-processor card acceptor data to useful merchant records when names, MCCs, and locations are vague, misleading, or affected by rebranding. Users can submit a replacement merchant name, website, and category, but deciding whether the request is legitimate and whether to update an existing record or create one requires additional context from related receipts, extracted line items, and transaction memos. Ramp addresses this with an AI agent built around an LLM, embeddings, rapid online analytical processing queries, multimodal retrieval-augmented generation, and guardrails that constrain permitted changes. The agent handles requests in under 10 seconds, and monitoring indicates it improves nearly 99% of transaction classifications, while fewer than 10% receive a second correction request. The same flow has also been extended to batch remapping and credit-card statement matching."
---

# How Ramp Fixes Merchant Matches with AI

[Ramp](https://yomu.fyi/company/ramp) · Chami Lamelas, Anton Biryukov · Jun 16, 2025

**Type:** Problem & solution

## Summary

Ramp describes the difficulty of mapping payment-processor card acceptor data to useful merchant records when names, MCCs, and locations are vague, misleading, or affected by rebranding. Users can submit a replacement merchant name, website, and category, but deciding whether the request is legitimate and whether to update an existing record or create one requires additional context from related receipts, extracted line items, and transaction memos. Ramp addresses this with an AI agent built around an LLM, embeddings, rapid online analytical processing queries, multimodal retrieval-augmented generation, and guardrails that constrain permitted changes. The agent handles requests in under 10 seconds, and monitoring indicates it improves nearly 99% of transaction classifications, while fewer than 10% receive a second correction request. The same flow has also been extended to batch remapping and credit-card statement matching.

## Context

Payment-processor card acceptor data can be insufficient for accurate merchant mapping: names may be vague, MCCs may be misleading, locations may not reflect the transaction, and rebranding can create mismatches. Manual correction involved customer support, finance, and engineering teams, took hours, and did not scale with transaction volume.

## Approach / What changed

Ramp uses an AI agent combining an LLM, embeddings, rapid online analytical processing queries, multimodal retrieval-augmented generation, and guardrails. The agent evaluates user requests using card acceptor data, MCCs, related receipt information, extracted merchant details and line items, and user memos, then selects a constrained resolution action.

## Takeaways

- Card acceptor names, MCCs, and location data can all be unreliable signals; examples include vague names such as SERVICE FEE and hotel transactions whose MCC reflects another service.
- The agent uses receipts, extracted merchant names and line items, user memos, and transaction metadata to judge whether a requested merchant correction is legitimate.
- The agent handles requests in under 10 seconds and close to 100% of them, compared with 3% handled in 2023 and 1.5% in 2024 by support and engineering teams.

**Tags:** [AI Agents](https://yomu.fyi/topic/ai-agents), [LLMs](https://yomu.fyi/topic/llm)

- Source: [Ramp](https://builders.ramp.com/post/fixing-merchant-classifications-with-ai)
- Source URL: https://builders.ramp.com/post/fixing-merchant-classifications-with-ai
- Ingested by Yomu: 2026-09-01T01:34:34.382Z

[Read original post](https://builders.ramp.com/post/fixing-merchant-classifications-with-ai)