HyperJet

Dashboard and settings
Insights

Inspiration

The Problem: We realized that for most people, the journey to homeownership is disconnected. They see a potential home in a listing, but they can't visualize its potential, and the financing process feels like a separate, dry world of spreadsheets. We wanted to move beyond just "rates and mortgages" to help people emotionally connect with their future home.

The Technical Goal: Simultaneously, we wanted to prove that a complex, data-heavy financial application doesn't need three separate codebases. We set out to build a technically robust Cross-Platform app that offers a native experience on Mobile and Web from a single source of truth.

Our Solution: HypoJet bridges the gap between dreaming and planning. We use Generative AI to visually "renovate" property listings in real-time, instantly showing users what a house could look like. Then, we immediately calculate the financial feasibility using a Mortgage Budget API. It turns abstract numbers into a concrete, visual dream.

What it does

HypoJet is a unified Kotlin Multiplatform (KMP) application that serves as an intelligent companion for home buyers in the DACH region.

Smart Affordability Check (The "Plan" Factor): * Once the user falls in love with the visual, the app checks the math. * We integrated a dynamic maximum credit calculation based on income and existing wealth

Cross-Platform Accessibility: * The exact same business logic runs on Android, iOS, and Web (Wasm). * Users can start their renovation on their phone at a viewing and finish the financial application on their desktop at home.

Localization: Support for German and English languages.

How we built it

We focused on a "Write Once, Run Everywhere" architecture to maximize efficiency during the hackathon.

Core Logic (Kotlin Multiplatform):
- We built a shared domain layer containing our Data Models (Property, LoanOffer) and Business Logic.
- Ktor was used for all networking, handling both the financial REST APIs and the AI endpoints in a unified client.
- Koin handles Dependency Injection across all platforms.
UI (Compose Multiplatform):
- We share 95% of our UI code. The screens for "Renovation," "Budget Input," and "Results" are identical on Android and Web.
- We implemented a responsive design that adapts to mobile touchscreens and desktop mouse inputs automatically.
AI Integration: We built a custom Repository in the shared module to communicate with the Gemini 3 Pro model. It handles the complex requests required to send images + text prompts to the inference engine.
Localization: We utilized the new composeResources API to separate string files into values (English) and values-de (German).

Challenges we ran into

Wasm Image Handling: Sending an image to an AI API requires Base64 encoding. While this is straightforward on Android, it is complex in a shared KMP environment. We had to write a custom expect/actual implementation to use the browser's FileReader API on Web/Wasm and native streams on Android.
API Response Times: High-quality AI image generation takes several seconds. To prevent the user from feeling "stuck," we implemented a progressive loading state with fun, localized tips (e.g., "Calculating your dream home..." / "Traumhaus wird berechnet...") to keep engagement high.

Accomplishments that we're proud of

True "Single Codebase" Deployment: We are deploying to Android (.apk) and Web (via Kotlin/Wasm) from the exact same project structure.
Emotional UI: We didn't just build a form filler. The transition between the "Old" house and the "AI Renovated" house provides a genuine "Wow" moment that pure financial apps lack.
Business Value Fit: We successfully integrated the Budget & Property APIs, giving the "fun" AI features a serious, actionable business value.

What we learned

Prompt Engineering for Real Estate: We learned to generate cartoon AI animations.
KMP is Production Ready: The ease of sharing the Repository layer meant we could fix an API bug once, and it was instantly fixed on both Android and Web.

What's next for HyperJet

Real-Time Valuation: Using a Property Search API to find comparable renovated homes nearby to estimate the post-renovation value of the user's project.
Sustainability Score: Automatically analyzing the image to suggest green upgrades (e.g., "This roof looks perfect for solar panels") and linking to specific green mortgage products.