Inspiration
Most expense-splitting apps are bloated: they ask for logins, require everyone to download an app, and lock data behind accounts. SplitFool was built as a minimalist, open-source alternative. Simple, accountless group management using short codes, with a clean web UI and deploy-anywhere backend. The vision: group splitting that works like a shared whiteboard, not a social network.
What it does
SplitFool is a full-stack web app that lets users track and simplify group expenses without accounts, emails, or integrations. Key features:
- Groups are created and joined using 6-character codes, no login required
- Anyone with the code can add/remove expenses and members
- Automatically calculates net balances and optimal payments
How we built it
SplitFool is built with a modern, minimal stack:
- Frontend: Svelte, TypeScript, Vite, Tailwind CSS hosted on Cloudflare
- Backend/API: Hono framework with ElectroDB and DynamoDB (single-table design), deployed via AWS Lambda
- Dev containers: Docker for local development and hot reload
- Infra tooling: Terraform
Challenges we ran into
- Designing a truly stateless user experience, without logins or sessions, that’s still secure and collaborative
- Implementing full local development parity with AWS services (like simulating DynamoDB locally with Docker)
- Balancing minimalist UI with enough clarity for users to track and settle expenses in multi-user flows
Accomplishments that we’re proud of
- Fully functional MVP without requiring any user data
- Clean separation of concerns between frontend, backend, and infra for both dev and prod environments
- Designed from day one for scalability via serverless AWS
What we learned
- Serverless can be lean if you build with the right primitives (e.g., ElectroDB + Lambda)
- Svelte remains unmatched for fast, expressive UIs with minimal overhead
- Even in 2025, most “simple” apps aren’t actually simple — minimalism takes discipline
What’s next for SplitFool
Build a Gemini-powered chat interface that allows users to control expenses using natural language, including:
- Adding expenses via receipts, screenshots, or chat messages
- Parsing group chat screenshots to auto-import members
- Running rebalancing and summaries conversationally
- Integrating Gemini 2.5 Flash for fast multimodal parsing
Built With
- amazon-dynamodb
- amazon-web-services
- cloudflare
- lambda
- svelte
- typescript
- vite

Log in or sign up for Devpost to join the conversation.