Case Studies | FinTech

Reading the merchant behind the transaction

Merchant Identification AI

About

A FinTech business working with Cloud Combinator on AWS. The client is anonymised at their request.

Challenge

The challenge had three focus areas.

Unidentified and misattributed merchants

The existing system could not always identify a merchant ID or name from a transaction, and points were not always allocated correctly as a result. The client needed a more reliable way to work out which retailer a transaction belonged to.

Messy, inconsistent descriptions

Transaction descriptions are short, abbreviated and vary wildly between banks and merchants. Rules alone struggle with this kind of text, which is exactly the sort of ambiguity a language model can reason through.

Working at production scale, in their own account

The solution had to cope with hundreds of thousands of transactions a month and run inside the client's own AWS environment, so the team could operate and extend it after handover rather than depend on us.

Solution

We ran the engagement through our Cloud Accelerator programme, a staged path from discovery to a deployed proof of concept and a Well-Architected review. The work centred on getting a language model to reason reliably about transaction text and wrapping it in a simple, serverless API.

350k

Monthly transactions the pipeline is designed to handle

3

Modes: validate, match, or propose a new merchant name

~$1,237

Indicative monthly running cost, projected

By the numbers:

  • 350k - Monthly transactions the pipeline is designed to handle
  • 3 - Modes: validate, match, or propose a new merchant name
  • ~$1,237 - Indicative monthly running cost, projected
Changes

The engagement met its scope: a working AI proof of concept that identifies the vendor behind a transaction from its description, deployed into the client's own AWS account and reviewed against the Well-Architected Framework.

  • Better merchant identification — A language model reasons about the transaction description to validate, match or name the merchant, addressing the gaps in the previous rules-based approach.
  • Built for the messiness — In-context learning lets the model cope with abbreviated and inconsistent descriptions that defeat fixed matching rules.
  • Serverless and scalable — An API and AWS Lambda front a model on Amazon Bedrock, a pattern sized for hundreds of thousands of transactions a month.
  • Handover — We delivered documentation and a live demonstration, and completed a Well-Architected review so the team could operate and extend the solution.

With a proven identification pipeline in place and a Well-Architected foundation to build on, the client has a base it can extend, tuning the prompts, growing the merchant database and taking the approach further across its AI roadmap.

AWS Stack

Amazon Bedrock

For large language model access, using Anthropic Claude, to validate, match and propose merchant names.

Amazon SageMaker

For developing and testing the prompt logic before productionising it.

AWS Lambda

For serverless execution of the identification logic on each transaction.

Amazon API Gateway

For receiving transaction JSON and returning results for validation.

Amazon DynamoDB

For low-latency storage of merchant records.

YOU MIGHT LIKE

Related success stories

View all case studies

Case Studies | Insights

Utilising Language Recognition, Speed, and Enhanced Security to Make Social Media a Force for Good

  • Here, we take a detailed look at how the Cloud Combinator team collaborated with another cutting-edge AI service provider that provides intelligent systems to “make social media more social” for brands and users alike.
  • Arwen AI is a UK-based startup specialising in AI solutions to manage and enhance brands’ social media interactions. Founded in 2020 by Matt McGrory, Dr. David Cole, and Joel Bailey, Arwen. AI focuses on using AI to automatically detect and remove spam, toxic comments, and other unwanted content from social media platforms.
  • The team at Arwen have three core products. ‘Moderate’ is focused on identifying and removing toxic content from social media channels. ‘Engage’ helps brands identify and engage with meaningful conversations on social media, and ‘Customize’ allows brands to apply bespoke algorithms to their channels - creating an even more effective moderation and engagement.
Read more
CONTACT US

Ready to turn AI into impact?

We'll help you spot the highest-value opportunities, reduce risk around your first AI initiative, and define a clear path to results from day one.

Why talk to us:

Outcome-driven recommendations

AWS-recognised delivery expertise

Risk-aware AI adoption

Clear next step, not a sales pitch

Start with a focused 20-minute conversation about your goals — no pressure, no commitment.

We use cookies

This website uses cookies to enhance user experience and to analyze performance and traffic on our website.

See our Privacy Policy for details.