Civiclens

Inspiration## Inspiration

All of us have walked past a pothole, a broken streetlamp, or overflowing trash and thought, "Someone should fix that." But reporting these issues is often tedious, opaque, and thankless. We wanted to change that narrative. CivicLens was born from the idea that civic engagement should be rewarding—literally. By combining the transparency of blockchain with the verification power of AI, we created a system that pays citizens instantly for helping their community.

What it does

CivicLens is a decentralized civic reporting platform that incentivizes infrastructure maintenance through instant crypto bounties.

Snap & Report: A user takes a photo of a civic issue (e.g., a pothole).
AI Verification: Our backend uses AWS Rekognition to detect objects and AWS Bedrock (Claude) to verify the severity and context of the report.
Instant Payout: Once verified, the Cronos EVM smart contract automatically sends a USDC bounty to the reporter's wallet via the x402 payment rails.
On-Chain Record: Every report is logged on-chain for transparency and distinct severity scoring (1-10) determines the payout tier.

How we built it

Frontend: Built with React and Vite for a snappy user experience, styled with TailwindCSS. We used ethers.js to connect with the Crypto.com DeFi Wallet and MetaMask.
Backend: A Node.js/Express server that handles the "heavy lifting" of AI processing.
AI Layer: We integrated AWS Rekognition to detect labels (e.g., "Road", "Asphalt", "Flood") and AWS Bedrock to act as a logic engine, cross-referencing the image labels with the user's description to prevent fraud.
Blockchain: We wrote Solidity smart contracts deployed on the Cronos EVM Testnet. The contract (CivicLensBounty.sol) holds the bounty pool and features a restricted verifyAndPayBounty function that our trusted AI backend calls to trigger payments.

Challenges we ran into

AI "Hallucinations": Getting the AI to reliably distinguish between a severe pothole and a simple crack was tricky. We implemented a hybrid approach using Rekognition for hard object detection and Bedrock/Claude for contextual reasoning to create a robust "Severity Score."
Oracle Problem: Connecting the off-chain AI verification to the on-chain smart contract required a secure trusted bridge. We secured our backend to act as the sole authority allowed to call the payment function on the contract.

Accomplishments that we're proud of

End-to-End Automation: We successfully automated the flow from a physical photo to a digital crypto payment without human intervention.
x402 Integration: Leveraging the Cronos x402 concept to make payments "embedded" into the user flow rather than a manual administrative process.
Accuracy: tuning our AI prompts to correctly identify valid reports vs. random images.

What's next for CivicLens

Community Validation: Adding a staking mechanism where users stake tokens to vote on reports for extra rewards (DAO governance).
GPS Verification: Integrating strict geolocation checks to ensure reports are coming from the claimed location.
City Partnerships: Offering an enterprise dashboard for city municipalities to view verified, prioritized incident maps.