SubPilot AI

Inspiration

Subscription costs are easy to ignore because they are small, recurring, and spread across many merchants. Over time, this creates hidden monthly burn and avoidable annual spend.
We built SubPilot AI to make subscription spending visible, actionable, and explainable from one place.

What It Does

SubPilot AI helps users detect, understand, and act on recurring payments:

Detects subscription-like transactions from imported Monzo-style history (12-month view)
Groups recurring merchants and estimates cadence/cost
Shows unified subscription, upcoming, pay, reminders, actions, and risk views
Provides AI-powered recommendations (overpaying, alternatives, savings opportunities)
Surfaces budget risk with Monte Carlo simulation across multiple horizons
Supports wallet top-up and payment workflows via Stripe (checkout + webhook path)
Sends reminder confirmation + scheduled reminder emails with calendar support (ICS + Google Calendar link)

How We Built It

Backend

Django + DRF API server
SQLite persistence
Core modules for:
- recurring payment detection/grouping
- AI advisor orchestration (Gemini)
- risk simulation (Monte Carlo)
- reminders + Brevo transactional email integration
- Stripe payment/session/webhook handling

Frontend

React + TypeScript (Lovable-started, then custom implementation)
Multi-page product UI:
- Dashboard
- Subscriptions
- Upcoming
- Pay
- AI Advisor
- Reminders
- Actions
- Risk
- Settings

Tech Stack

Backend: Django, Django REST Framework
Database: SQLite
Frontend: React + TypeScript
AI: Gemini API
Payments: Stripe
Email/Calendar: Brevo + ICS + Google Calendar deep link
Analytics: Monte Carlo risk engine (probabilistic balance-outflow simulation)

Challenges We Ran Into

Branch coordination and merge conflicts while iterating quickly on both UI and API
Keeping frontend and backend models aligned during rapid feature expansion
Handling inconsistent real-world transaction labels/intervals for recurring detection
Balancing AI response quality, latency, and deterministic grounding from app data

Accomplishments We’re Proud Of

Built a true full-stack fintech prototype, not just a static demo
Integrated multiple real systems: AI, Stripe, reminders/email scheduling, calendar events
Delivered decision-support features beyond listing subscriptions:
- risk probability
- savings drivers
- actionable suggestions
Created a product flow that turns noisy transaction data into clear user decisions

What We Learned

Tight API contracts and migration discipline are critical in fast-moving teams
Product value comes from explainability and actionability, not only detection
Combining deterministic finance logic with AI explanations gives better trust and UX