Monie
A personal-finance analysis tool that treats budgeting as a pattern-recognition problem — built as a Social Action Project.
What it is
Most people don't overspend because they lack discipline — they overspend because nobody shows them the pattern. Monie reads transaction history and surfaces what a spreadsheet won't volunteer: where the money actually goes each month, which transactions don't fit the usual shape, and where a small adjustment would matter.
How it's built
- Transaction categorisation — the deceptively hard part; real-world merchant strings are messy and inconsistent.
- Monthly pattern analysis across categories, with anomaly flags for transactions that break the pattern.
- Budget suggestions phrased as concrete, small adjustments rather than lectures.
- Designed to run locally — your transaction history shouldn't need to visit a server to be understood.
Status
In active development. The categoriser is the current battle: teaching a model that "AMZN MKTP" and "Amazon Marketplace" are the same store is exactly the kind of unglamorous problem ML is actually made of. Code goes public once it's something worth reading.
Links
code coming soon ← All projects