Deedlock Holmes

Every deed verified. Every transfer trusted.

Comment 6

Inspiration

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Deedlock Holmes

🕵️‍♂️ DeedLock Holmes — Securing Property Rights with AI + Blockchain

🌍 What Inspired Us

Imagine sitting down for breakfast when someone knocks on your door and claims your home isn’t yours — holding a forged deed as proof.

In many developing countries, this is not hypothetical. Property deed fraud costs billions every year. Forged documents, duplicate transfers, corrupt registries, and no reliable way to verify ownership create instability. In places like Myanmar and Bangladesh, property disputes can escalate into community conflict and political unrest.

That reality inspired us to build DeedLock Holmes — a system that combines AI-powered document screening with immutable blockchain records to protect property ownership.

Our goal was simple:

Make property fraud impossible to fake and easy to verify.

🧠 What We Learned

Building this project taught us far more than just how to use Solana.

1️⃣ Blockchain Is Not a Database — It’s a Trust Layer

We learned that Solana programs don’t store data like traditional systems. Instead:

  • Programs define rules
  • Accounts store property records
  • Program Derived Addresses (PDAs) ensure uniqueness and prevent duplication

This forced us to carefully design how ownership and data would live on-chain.

2️⃣ AI Is Powerful — But Needs Guardrails

Gemini AI extracts deed information and assigns a fraud risk score.

But we learned:

  • AI output must be validated
  • JSON responses must be sanitized and parsed carefully
  • Fraud thresholds need explicit logic

AI became our screening layer — not our final authority.

3️⃣ Full-Stack Web3 Architecture Is Different

We had to coordinate:

  • React frontend
  • Node.js backend
  • MongoDB authentication
  • Gemini API
  • Anchor (Rust) smart contract
  • Solana Devnet

Making these systems communicate reliably was one of our biggest engineering challenges.

🏗 How We Built It

Our architecture follows a clean trust flow:

🔐 Authentication Layer

We implemented:

  • Google login
  • MongoDB user storage
  • JWT authentication
  • Wallet mapping to Solana addresses

Each user is linked to a Solana wallet that associates them with property tokens.

📄 Property Registration Flow

When a user uploads a deed:

  1. Gemini AI analyzes the PDF

    • Extracts address, owner name, and sale price
    • Assigns fraud risk and confidence score
    • Provides a reasoning summary

  2. Fraud Check

    • If flagged → blocked from the blockchain
    • If valid → proceed to on-chain registration

  3. Solana Registration

    • A deterministic PDA prevents duplicate property entries
    • A token is minted or associated with the property
    • A transaction is recorded on Solana Devnet
    • An Explorer link is returned for verification

This guarantees:

  • One property → one owner
  • No duplicate registrations
  • Immutable transaction history

🔄 Ownership Transfer

Ownership is fully transferable:

  • A new owner is entered
  • A new transaction is written on-chain
  • Previous ownership is permanently recorded
  • History is transparent and publicly verifiable

If someone attempts to register the same property twice:

They are blocked instantly with a “Property Already Registered” alert — enforced by deterministic PDA logic.

🌐 Public Verification Dashboard

Anyone can:

  • Browse all registered properties
  • View fraud risk and AI confidence scores
  • See recorded prices
  • Click directly to Solana Explorer

Everything is transparent.
Nothing can be secretly altered.

⚠️ Challenges We Faced

1️⃣ Solana Devnet Faucet Limits

We encountered rate limits while testing and had to strategically switch between local validators and Devnet.

2️⃣ PDA Derivation & Anchor Setup

Understanding how program-derived addresses prevent duplication required deeper knowledge of:

  • Seed generation
  • Account ownership
  • Deterministic address derivation

3️⃣ AI Output Formatting

Gemini sometimes returned markdown-wrapped JSON.
We had to clean and parse responses carefully to ensure reliability.

4️⃣ Coordinating Multiple Systems

Getting:

  • MongoDB authentication
  • Gemini AI validation
  • Anchor program logic
  • Frontend state updates

to work seamlessly required careful debugging and iteration.

🚀 What Makes DeedLock Holmes Powerful

  • ✅ AI fraud detection before blockchain write
  • ✅ Immutable on-chain storage via Solana
  • ✅ Deterministic duplicate protection
  • ✅ Transparent public verification
  • ✅ Transferable ownership with full history

We combined AI’s intelligence with blockchain’s permanence.

🌎 Our Vision

Through DeedLock Holmes, our hope is to bring certainty, security, and reliability to property ownership around the world — especially in regions where paper records fail the people who depend on them.

One deed.
One Solana connection.
One step toward trust.

Thank you.

+ 3 more
Share this project:

Updates