ShopChain — AI-Powered Safe Deals with Solana Escrow

Inspiration

Buying expensive items from strangers online — bikes, laptops, cameras, tickets, sneakers — always feels risky.
You see a listing, you like the price, but something in your mind whispers:

“What if this is a scam?”

Most platforms still rely on trust, screenshots, and hope.
We wanted to replace trust with verifiable safety.

Our core idea was simple:

$$ Trust = Transparency+ Verification $$

If we could combine AI (to evaluate risk) and blockchain escrow (to protect the payment),
we could create a system where deals feel safe, predictable, and transparent — even between total strangers.

That is what inspired ShopChain.

What It Does

ShopChain makes online peer-to-peer buying safe, smart, and visual.

Listing Analysis

Paste any listing from Facebook Marketplace, eBay, Craigslist, Discord, WhatsApp — anything.
ShopChain uses an NLU model to extract:

Item name
Price
Delivery terms
Seller behavior
Missing or suspicious details

It then generates a risk score and a short explanation.

Solana Escrow

If you decide to proceed, ShopChain creates a Solana escrow contract:

Buyer funds the escrow vault
Seller only gets paid after shipment
If anything goes wrong, funds are refunded

Interactive Deal Card UI

Both buyer and seller get a beautiful animated Deal Card showing:

Buyer side vs seller side
A central lock that “closes” when funded
Coins moving during escrow operations
Real-time status: Funded → Shipped → Released

It feels like a real digital contract.

How We Built It

Blockchain — Solana Web3

Escrow vault contract
Buyer → Escrow → Seller flow
Refund path
Deployed on devnet for demo stability

Frontend — React + Solana Wallet Adapter

Input for listing text/link
Real-time risk display
Dynamic risk bar
Animated Deal Card
Phantom wallet integration

Data collection, condensation, and Gemini AI interpretation

Using URL given in input (as stated above), we parse data from the product site
Using variety of data optimization methods we simplify/shorten data to most relevant information for AI
AI makes determination on risk given the necessary data

Deployment

Frontend on Vultr
Backend and Hosting on Vultr
PostgreSQL Database entirely on Vultr
Solana programs on devnet

Challenges We Ran Into

1. Messy Listings

Listings vary wildly — emojis, slang, broken grammar.
We built preprocessing + heuristics to stabilize the NLU pipeline.

2. Solana Transaction Timing

Devnet confirmation delays caused UI desync.
We added fallback states like:
“Waiting for transaction settlement…”

3. Complex Animations

Synchronizing animations with real blockchain events required
a clean state machine for each escrow stage.

4. Time Pressure

Building AI + blockchain + UI in a hackathon forced us to prioritize aggressively.

Accomplishments We're Proud Of

A working Solana escrow system
A clean, helpful AI risk analyzer
A fully animated Deal Card UI
End-to-end flow: listing → analysis → escrow → release
A project that feels like a real product, not just a prototype

What We Learned

We learned:

How to build real trust infrastructure
How to integrate LLMs for structured information extraction
How escrow systems operate behind the scenes
How to manage complex frontend state tied to blockchain events
Why UX and trust-signals matter for financial tools

We also learned that a small insight can grow into
a powerful product with the right architecture.

What’s Next for ShopChain

Near Future

Chrome extension: “Scan this listing”
Fraud-detection ML trained on real scam datasets
Shipment verification via QR scans + photo proof
Stronger NFT proof system

Long-Term Vision

We aim to transform ShopChain into a full Safety-as-a-Service ecosystem, where marketplaces, apps, and communities can instantly plug into our escrow, verification, and fraud-analysis APIs. From small sellers to large platforms, anyone can add a single line of code to unlock trusted peer-to-peer commerce.