Klariefy

Bank statements are dense and hard to read, and most tools are built for foreign formats. People rarely understand their own spending patterns, and there was no product tuned to Nigerian statement structures.

I studied the layout of statements from major Nigerian banks and catalogued the inconsistencies in how transactions are described, dated, and categorised.

I planned a pipeline that ingests a statement, normalises it, categorises transactions, and produces behavioural summaries. Authentication and data privacy were designed in from the first sprint.

The product presents finance without intimidation. Insights are written in plain language, supported by restrained charts, so a first time user understands their money at a glance.

Built on Next.js 15 with Auth.js v5 for authentication and Supabase with PostgreSQL for data, deployed on Vercel. The parsing engine is the heart of the system, converting messy inputs into consistent records.

Statement formats vary widely between banks and even between versions, which made reliable parsing the central technical risk.

I built a normalisation layer that handles format variation gracefully and falls back cleanly when a pattern is unrecognised, so the product stays trustworthy rather than guessing.

Klariefy gives users a readable, honest picture of their financial behaviour and lays a foundation for richer financial intelligence features.

When the input is unpredictable, investing early in a strong normalisation layer pays off across every feature that depends on it.