Mortgage Navigator

Mortgage Navigator will guide borrowers from pre-approval to closing with AI risk alerts, a smart rate calculator, and milestone tracking. Mortgage doesn't have to be scary, it just needs a map.

Comment

Mortgage Navigator

What Inspired Us

One of us comes from computer engineering. The other from biomedical engineering. Different disciplines, same instinct: when a process is opaque and error-prone, you instrument it.

The mortgage process is a system with no feedback loop for the person it matters most to. Underwriting follows rules, but borrowers never see them. Applications stall with no explanation. Documents get rejected without context. We saw a tractable engineering problem: take a workflow that already has defined stages, inputs, and decision criteria, and make it legible.

We also wanted to build something with a physical layer. A web dashboard tells you where you stand. A house on your desk that glows green when you close is something you remember.

What We Built

Mortgage Navigator is a borrower intelligence platform that walks you from pre-approval to closing with:

  • An AI assistant that flags risks in plain English
  • Location-aware calculators that benchmark price-per-sq-ft against your state average
  • A 3D-printed house that physically lights up as you hit milestones

The hardware was a risk for a software hackathon. But we wanted something you could touch, something that made progress feel real.

How We Built It

Backend: Base44 We used Base44 for data storage and authentication, creating entities for loan applications, documents, and users. It handled all the CRUD and auth automatically, letting us focus on the borrower experience.

Frontend: Appifex The borrower dashboard was built in Appifex. Users create applications, upload documents, and move through a 5-stage workflow. Each stage triggers AI-powered actions: credit checks, risk flags, and underwriting summaries written in plain English.

Hardware: Arduino Uno R4 WiFi + RGB LED When the app advances a loan stage, it posts to an ntfy.sh topic, a free pub/sub message relay. An Arduino Uno R4 WiFi polls that topic every 5 seconds and updates an RGB LED, with each color mapped to a loan stage.

The full signal path: $$\text{App Stage Change} \rightarrow \text{ntfy.sh topic} \rightarrow \text{Arduino HTTP poll} \rightarrow \text{RGB LED color}$$

Challenges

Arduino over WiFi is unforgiving. We hit ntfy.sh rate limits, wrestled with HTTP client timeouts, and watched the Arduino crash from heap fragmentation caused by dynamic String objects. The fix: seed the topic on boot and replace every String with a char[]. Stable after that.

Common anode vs. common cathode. We spent an embarrassing amount of time debugging an LED that wouldn't behave. We had a common anode LED wired as common cathode, meaning the long leg goes to ground, not power. A small thing that cost us an hour.

Backend paywalls mid-build. Base44's backend functions required a paid plan. Rather than pay or pivot off the platform entirely, we routed the hardware integration through ntfy.sh as a free relay. Constraint turned into a creative solution.

What We Learned

Arduino over WiFi is harder than it looks. HTTP clients timeout. Strings fragment memory. Rate limits appear exactly when you're demoing.

We also learned that hardware changes how software feels. When the house lit up for the first time, we both smiled. That's the moment we kept coming back to, and the feeling we want borrowers to have when they close on a home.

What's Next

Mortgage Navigator has a real future as a lender onboarding tool. The physical house, a keepsake that stays lit after closing, gives borrowers something tangible to mark the journey.

On the software side, we'd add live rate APIs, deeper document analysis, and expanded underwriting rule coverage. For two weeks of work, we're proud of what we built.

Built With

Share this project:

Updates