skip to content
.katana

Spend Lens

AI-powered personal finance platform for Nigerian bank statements — parses transactions across major banks, categorizes spending automatically, and answers financial questions through a conversational assistant.

Spend Lens showcase
Role
Full-stack Development, AI Integration
Date
January 15, 2025
Deliverables
Web application, Multi-bank statement parsing, AI chat assistant, Budgeting & receipt tracking, Categorization rules engine
Tools
Next.js, TypeScript, Python, FastAPI, Supabase, PostgreSQL, OpenAI

Spend Lens turns raw Nigerian bank statements into structured, queryable financial data. Upload a PDF from GTBank, Zenith, Access, Sterling, OPay, or FirstBank, and it parses transactions, categorizes spending automatically, and surfaces insights — plus a conversational AI assistant that can answer questions about your money and Nigerian tax obligations.

A FastAPI backend parses statements, runs an LLM-scored categorization-rules engine, and orchestrates an OpenAI-backed agent with tool access to the user's own transaction data. A Next.js 15 / TypeScript frontend renders dashboards, budgets, receipts, and chat. Supabase provides Postgres, auth, and storage; Mono handles open-banking account linking. Every request is scoped to a user through a dedicated service-manager layer, so domain services never touch the database directly.

view repoview live

Overview

Spend Lens turns raw Nigerian bank statements into structured, queryable financial data. Upload a PDF from GTBank, Zenith, Access, Sterling, OPay, or FirstBank, and it parses transactions, categorizes spending automatically, and surfaces insights — plus a conversational AI assistant that can answer questions about your money and Nigerian tax obligations.

Spend Lens

Approach

A FastAPI backend parses statements, runs an LLM-scored categorization-rules engine, and orchestrates an OpenAI-backed agent with tool access to the user's own transaction data. A Next.js 15 / TypeScript frontend renders dashboards, budgets, receipts, and chat. Supabase provides Postgres, auth, and storage; Mono handles open-banking account linking. Every request is scoped to a user through a dedicated service-manager layer, so domain services never touch the database directly.

Features

  • Multi-bank PDF statement parsing (GTBank, Zenith, Access, Sterling, OPay, FirstBank)
  • Conversational AI assistant with tool access to a user's own transaction data
  • LLM-scored automatic categorization, with a user-editable rules engine
  • Budgets, receipt tracking, and recurring email reports
  • Nigerian tax Q&A backed by an ingested tax-law corpus
  • Open-banking account linking via Mono
.see also
view all projects
say hello