Inspiration
Managing shared money in a group is more complicated than simply splitting a bill.
Whether people organize college events, travel together, share accommodation, manage family expenses or plan celebrations, financial information often gets scattered across messages, receipts, payment screenshots and documents.
People struggle to keep track of who paid, how much was spent, who owes money and what remains within the budget.
Existing expense-splitting solutions primarily focus on dividing expenses. We wanted to go further.
Our inspiration was to create a platform that could understand and organize a group's entire financial activity, rather than just calculate who owes whom.
That idea became SplitSmart AI — The Financial Memory of a Group.
What it does
SplitSmart AI is an AI-powered group financial management platform designed to support any type of shared group activity.
Users can create groups according to their own requirements, customize activities, manage members and roles, set budgets, organize expenses and track settlements.
What makes SplitSmart AI different is its use of AI throughout the financial management process.
With Google Gemini, users can describe their requirements in natural language and generate customized group structures and activities. They can also use AI to understand expense information and interact with their group's financial history.
The platform is designed to support multiple information formats, including:
Natural-language text.
Receipt and bill images.
Payment screenshots.
PDF documents.
Supported audio and video files.
Voluntarily exported chat conversations.
AI helps extract and organize relevant information, identify potential duplicate records and prepare expense entries for user review.
The application also focuses on:
Member-wise contributions and expense sharing.
Customizable budgets and activities.
Category-wise spending analysis.
Outstanding balances and settlement suggestions.
Evidence management.
An AI assistant for group-related financial questions.
PayPal checkout integration and transaction verification.
Our goal is to bring a group's financial information into one organized and understandable workspace.
How we built it
We are building SplitSmart AI using a Python-based backend and a web-based frontend.
Technology stack:
Python: Core application logic and financial calculations.
FastAPI: Backend APIs and application services.
SQLAlchemy and SQLite: Database management.
HTML, CSS and JavaScript: User interface and frontend interactions.
Google Gemini API: Natural-language understanding, information extraction and AI-powered interactions.
PayPal Developer APIs: Payment checkout and transaction verification.
A key architectural decision is separating AI understanding from financial calculations.
Gemini helps understand user requests and extract information from supported evidence. The Python backend validates that information and handles official expense calculations, contributions, balances and settlements.
This approach helps maintain accurate financial records rather than relying on AI-generated arithmetic.
The application also uses a structured database to manage groups, members, activities, expenses, evidence and settlements.
Challenges we ran into
One of our main challenges was designing an application that could support different types of group activities without forcing users into fixed templates.
Another challenge was making AI useful beyond ordinary conversations. We wanted Gemini to understand user requirements and interact with actual application functionality, including creating groups, organizing activities and preparing financial records.
Handling different evidence formats also introduced challenges involving file uploads, information extraction, validation and associating related documents with the correct expenses.
Financial accuracy was another important consideration. Equal and custom splits, multiple payers, partial settlements and rounding require reliable backend calculations.
We also had to consider secure API credentials, group permissions, evidence access and payment verification.
These challenges shaped our approach: AI for understanding and assistance, supported by validated backend operations for actual financial management.
Accomplishments that we're proud of
We are proud of developing the concept beyond traditional expense splitting into a more comprehensive group financial management platform.
Some of the key accomplishments behind SplitSmart AI include:
Designing a flexible group structure that supports different real-world activities.
Integrating AI into the planned group creation and customization workflow.
Bringing multiple forms of financial information into one concept.
Designing an architecture that separates AI understanding from financial calculations.
Developing group expense, contribution and settlement management capabilities.
Incorporating evidence management and financial history into the same application.
Planning a secure PayPal checkout integration.
Most importantly, we are proud of the idea of making financial management more natural: users should be able to explain what they need instead of being restricted to predefined application workflows.
What we learned
Working on SplitSmart AI taught us that building an AI-powered application involves much more than connecting an AI API to a chatbot.
AI must understand the user's requirements, work with structured information and interact with the actual application through controlled backend operations.
We also learned the importance of financial accuracy, database design, secure file handling, user permissions and payment verification.
One of our biggest takeaways was that AI and traditional application logic should complement each other. AI makes interaction more intelligent and natural, while reliable backend services ensure that financial records and calculations remain consistent.
What's next for SplitSmart AI
Our vision is to make SplitSmart AI a flexible financial management platform for groups of all kinds.
Future improvements include:
More advanced document and receipt understanding.
Better connections between related financial evidence.
More flexible activity and role customization.
Improved financial analytics and explainable spending estimates.
More conversational AI-powered group management.
Enhanced settlement workflows.
Further payment integration and verification capabilities.
Improved deployment, scalability and data security.
We want users to spend less time organizing scattered financial information and more time understanding and managing their group's finances.
One Group. Any Activity. Every Expense Understood. Every Payment Accounted For.