Sound&Clear

Inspiration

I've been scammed more than once. I did everything right — checked everything I knew, searched everything online — and still lost money. That taught me something most fraud pitches miss: the problem isn't a lack of information, it's a structural gap. Privacy law rightly protects everyone's personal data, which means a sender can never verify the stranger they're sending to. Scammers live inside that gap. They don't hack accounts; they talk people into sending willingly, sailing past every login screen and transaction filter. Every other approach I saw tried to build a smarter wall. I decided to light up the blind spot instead.

What it does

Sound & Clear shows a live trust signal for any recipient — phone, email, or payment tag — in the seconds before money moves. The score is computed from real data: actual member transfer history and live community vouches and reports — never mock data, never guesswork. Unknown recipients honestly display "no established record" rather than a fabricated score. The end state is a simple read-only API inside existing bank apps: from link or QR code to one small card at send time. It never blocks a legitimate transfer, so false positives are structurally impossible — it helps members make an educated decision instead of policing them.

How we built it

Solo, full-stack: React + Vite client, Node/Express API, Supabase (Postgres + Auth) for member records and community signals, Plaid for verified account activity over open banking. Every identifier is a one-way hash — no names, balances, or transaction details stored anywhere. The trust score draws on genuine account telemetry of past transaction or future treat and real community evidence, and the system is designed to stay honest about uncertainty: where the data is thin, it says so instead of inventing confidence. Because Sound Credit Union works with open banking APIs, the whole thing runs on real data flows rather than mockups.

Challenges we ran into

Two walls. The first was legal: privacy protections create the very blind spot scammers exploit, so the architecture had to prevent fraud without seeing personal data — that constraint shaped everything. The second was creative: AI assistants accelerated coding and basic legal checks, but every idea they suggested converged on the same generic concepts. That disagreement over originality is why I left my team and went solo. All core ideas here are my own, and scoping for one builder forced hard cuts — every surviving feature had to earn its place. I needed to show how my personal experience made me creative and technically unique, even within this narrow topic, compared to other teams.

Accomplishments that we're proud of

A working system on real data, not slides: live lookup, transfer-backed vouching, reason-labeled reports, verified account activity, and a live community feed. But the accomplishment I'm proudest of is a decision, not a feature — restraint. The system refuses to fake confidence, refuses to block innocent users, and refuses to hide behind a black box. Most student projects demo what their tech can do; this one is defined by what it won't do.

What we learned

Lived experience beats literature review: being scammed taught me the process, the procedures, and exactly where they fail. AI multiplies execution speed but not originality — the distinctive thesis still has to come from a human who felt the problem. And narrow scope turned out to be an advantage: with one builder and one topic, there was nowhere to hide behind feature count, so depth and coherence had to carry the project.

What's next for Sound & Clear

How I stand apart, and where that leads: other teams will show smarter detectors and bigger models — all competing on accuracy inside the bank's walls, where every false positive punishes a real member. I compete at the one moment nobody owns yet: the seconds before Send, on the member's side, with community evidence no single institution can replicate and a network effect that strengthens with every user. Next is a three-month pilot with one credit union — overlay on the existing send flow, judged on warn-and-stop rate with zero false blocks — then partnerships with member-facing fintechs to distribute the signal where conversations already happen.

Looking further ahead, I want to solve the hardest part of the puzzle: post-fraud recovery. Even with prevention, some scams slip through. Today, victims face a confusing, cold, and fragmented reporting process across banks, law enforcement, and credit bureaus. I am building an automated post-incident assistant that immediately freezes compromised channels, compiles verified transaction evidence for law enforcement, and guides members through dispute filings. By closing the loop from pre-send protection to post-fraud recovery, Sound & Clear evolves from a transaction safeguard into a complete, end-to-end financial safety net.

Built With

• express.js
• lucide
• node.js
• plaid
• postgresql
• react
• supabase
• tailwind
• vite