100K+ Financial Documents Enriched for Smarter Spending Insights
AI-powered transaction mapping for an Egyptian credit card management platform
About the customer
Our client is an Egyptian fintech start-up whose platform helps users repay credit card debt through fixed, lower monthly instalments, and gives them tools to track payments and understand their spending. This matters in Egypt, where credit card debt exceeds $2 billion. The platform could not yet show users where their money was going in detail. We built a data enrichment pipeline that reads bank statements and adds the missing details to each transaction.
What needed solving
Missing spending insight
Users had no way to see their spending patterns in detail, a capability the platform was missing
Bank statements
Individual bank statements had to be processed to extract key transaction attributes
Detailed attributes
City, country, region, merchant name and ID, and payment provider and ID had to be extracted for each transaction
How we built it
Vendor identification
The Serper API uses Google searches to identify vendor names accurately.
Attribute extraction
NLP and LLMs extract merchant names, payment providers and locations from transaction descriptions.
Validation
Custom algorithms cross-check the matched data against existing records.
Data handling
Pandas, PyCSV and Python handle the data, with HuggingFace Embeddings for the language tasks.
Spending categories
Each transaction is enriched so spending can be grouped by Merchant Category Codes (MCC).
Deployment
Packaged with Docker and deployed on the client’s platform, with LangChain and Qdrant managing the vector database.
Users gain new insight into their spending
Technologies
Domains
Have a Query? Let's Have a Conversation
Do you have a project in mind? We’d love to know what it is. Email us or send a message from our contact page.
Have a similar problem?
Tell us what you are trying to solve and our team will get back to you.