Data visualization is the first step of any application, but data interpretation is where we add real value. Until yesterday, my Cloud Financial Agent could tell me how much I spent on coffee. Today, it can tell me if that habit is ruining my future.
In this phase of the project, I faced an interesting challenge: How do I convert a raw JSON list of transactions into a quality-of-life metric (0-100)?
The Algorithm (Backend Strategy)
I avoided using LLMs for the core calculation because AI can hallucinate with strict math. Instead, I opted for a deterministic Python function injected into my AWS Lambda handler. The logic is strict:
Base: Every user starts with 50 points.
Savings Rate: If (Income - Expenses) / Income exceeds 50%, the algorithm injects +40 points.
Frugality Bonus: If total absolute expenses are under €500, it triggers a "Frugal Month" flag, adding +10 points.
Penalties: If cash flow is negative, the score plummets immediately.
The Transparency Layer (Frontend)
The biggest issue with fintech apps is ambiguity. To solve this, my API doesn't just return score: 100. It returns two key data structures:
short_reasons: A list of emojis for immediate visual impact (🔥 High Savings Rate).
audit_log: A technical breakdown that explains the math behind the number line-by-line.
Visual Integration
On the frontend (React + Tailwind + Recharts), I designed a custom Gauge component using dynamic SVGs that animate stroke-dashoffset based on the score received. Right below it, I implemented a console-style Audit Log to keep the developer aesthetic of the project.
The final result is a system that doesn't just show you your balance; it educates you on your own financial habits in real-time.
Top comments (0)