Skip to content
BA.
Live2026 · Full-stack — design, frontend, APIBuilt with Claude

FinanceFlow AI

Personal finance intelligence for households.

  • Next.js
  • TypeScript
  • Tailwind CSS
  • Node.js
  • Express.js
  • MongoDB

Overview

FinanceFlow AI is a personal finance manager built for households and joint families — a single place to track expenses, manage monthly budgets, and automate recurring dues and payments.

It grew out of a real need: managing shared family finances across multiple members without spreadsheet sprawl. I designed and built it end to end, using an AI-assisted workflow to move from idea to working product faster than a traditional solo build.

Problem

Shared household finances are messy: expenses live across people and apps, budgets drift silently, and recurring dues are easy to miss. Most trackers assume a single user and don't model a family with shared and personal categories.

  • No shared view across family members with per-member and pooled budgets.
  • Recurring payments tracked manually, leading to missed dues.
  • Data entry with no guidance on where the money actually goes.

Solution

A multi-member finance manager with shared and personal budgets, automated recurring-payment scheduling, and a clean dashboard that surfaces spending at a glance.

  • Member-aware accounts with shared + personal budget envelopes.
  • Recurring dues that schedule and remind automatically.
  • Dashboard with category breakdowns and monthly trends.

Architecture

A Next.js frontend backed by a Node/Express API and a document data model, with auth-gated household workspaces isolating each family's data.

  • Next.js App Router for fast, SEO-friendly rendering.
  • Express.js API with typed request/response contracts.
  • Data model covering households, members, budgets, and recurring rules.

Tech Stack

  • Frontend: Next.js, TypeScript, Tailwind CSS
  • Backend: Node.js, Express.js
  • Database: MongoDB

Challenges

  • Modelling shared vs. personal budgets without confusing the UX.
  • Designing a recurring-rules engine that's predictable and editable.
  • Keeping the data model flexible for a future multi-tenant SaaS.

Lessons Learned

  • An AI-assisted workflow compresses prototyping — but architecture still needs a human owner.
  • Modelling the domain up front saved far more time than it cost.
  • Shipping a v1 to real users (my own family) surfaced the right priorities fast.

Future Improvements

  • Bank/UPI import for automatic transaction ingestion.
  • Cash-flow forecasting and spending insights.
  • Multi-tenant SaaS packaging with role-based access.