Inspiration
The spark for VeriCarbon came from a frustrating paradox: carbon markets are supposed to be a primary tool for financing climate action, yet they are plagued by a trust deficit. We read reports of “phantom credits,” double-counting scandals, and verification timelines stretching 6–18 months. A market designed to save the planet was struggling to prove its own integrity. We were inspired by the IEEE ClimateChain track’s call for transparent verification, real-time tracking, and smart contracts to prevent double counting—and by emerging work like Hedera’s EcoGuard and academic multi-agent AI consensus frameworks. We asked: What if every carbon credit had a verifiable, real-time birth certificate and a tamper-proof lifecycle? That question became VeriCarbon.
What it does
VeriCarbon is a full-stack integrity engine for carbon markets with three integrated modules:
Automated dMRV (Digital Monitoring, Reporting & Verification): Ingests IoT sensor and satellite data, then uses a multi-agent AI consensus to validate emissions reductions in near real time. Each credit receives an Integrity Score—a confidence metric. Blockchain Registry & Anti-Double-Counting Engine: Mints each verified credit as a uniquely identifiable token (NFT-style) on a low-energy blockchain. The immutable ledger records origin, timestamp, and integrity score, making double-spending or duplicate claims impossible.
Transparent Trading & Automated Retirement: A smart-contract-enabled marketplace where buyers and sellers match via a double-auction mechanism. Buyers can retire credits directly, triggering an on-chain burn and generating a public, verifiable proof of offset. In short: VeriCarbon verifies carbon and trades trust.
How we built it
We built a working prototype over the hackathon period:
Blockchain layer: Deployed smart contracts on Polygon testnet (chosen for low fees and EVM compatibility). Contracts handle minting, transfer, and retirement (burn) of carbon credit tokens. Each token’s metadata includes a unique project ID, vintage, integrity score, and verification hash.
AI/dMRV layer: Used Python with scikit-learn for anomaly detection on simulated IoT time-series data (e.g., industrial stack emissions) and a rule-based multi-agent consensus: three independent “agents” (sensor-fusion, satellite-proxy, and historical-baseline models) must agree within a threshold before a reading is accepted.
Data layer: Lightweight InfluxDB (or JSON mock store) for time-series telemetry; a simple API connects the AI layer to the smart contracts.
Frontend: React/Next.js dashboard with three views: real-time emissions charts, a marketplace for available credits, and a “Retire Credit” button that calls the smart contract and displays the on-chain proof.
Integration: A Node.js backend orchestrates data flow from simulated sensors → AI consensus → token minting → marketplace listing.
We used LaTeX in our documentation to formalize the integrity score and the double-counting prevention condition.
Challenges we ran into
Simulating real-world dMRV data: Without access to live IoT hardware or satellite APIs, we had to design realistic mock data streams and ensure our consensus logic could detect injected anomalies (e.g., a sudden emission drop that looks like fraud).
Smart contract security and gas optimization: Writing mint/burn/transfer logic that prevents re-entrancy and double-retirement while keeping gas costs low on Polygon. We iterated on contract design and added require statements for state transitions.
Bridging AI output to on-chain data: Ensuring the integrity score and verification hash were tamper-evident. We used a signed hash from the AI layer stored in token metadata, but fully decentralized oracle integration was out of scope for the hackathon.
Time constraints: Building three modules (AI, blockchain, frontend) in a short window forced us to prioritize the mint → trade → retire flow as the core demo, while keeping the dMRV layer as a convincing simulation.
User experience for transparency: Making blockchain proofs understandable to non-technical users. We added simple status badges (“Verified,” “Retired,” “Integrity Score: 0.92”) and a public proof link.
Accomplishments that we're proud of
A working end-to-end prototype that mints a verified carbon credit, lists it on a marketplace, sells it via a smart contract, and retires it on-chain—with a public proof page.
Demonstrated anti-double-counting: We showed that attempting to retire the same token twice fails at the contract level, directly addressing a core track requirement.
Built a multi-agent AI consensus (even in simplified form) that flags anomalous sensor readings before minting, showcasing how AI can strengthen dMRV.
Designed a clear Integrity Score metric that gives buyers a confidence signal, moving beyond binary “verified/unverified” labels.
Aligned tightly with the ClimateChain track—covering transparent verification, real-time tracking, smart-contract retirement, and double-counting prevention in one coherent demo.
What we learned
Trust is a technical problem as much as a social one. Blockchain alone doesn’t create trust; it must be paired with credible data ingestion and verification logic. The multi-agent AI layer was essential to make on-chain records meaningful.
Carbon markets are complex. We gained a deeper appreciation for vintage years, project types, retirement registries, and the nuances of double counting (both double issuance and double claiming).
Smart contract design matters. Simple state machines and explicit require statements are powerful for preventing fraud, but they must be carefully audited.
Teamwork and scoping are everything in a hackathon. We learned to build the smallest demonstrable slice of each module rather than perfecting one layer.
Transparency needs UX. A hash on-chain is not transparency to a buyer; a clear proof page with plain-language status is.
What's next for VeriCarbon
Pilot with real IoT and satellite data: Partner with a local emissions-monitoring project or use open satellite APIs (e.g., Sentinel) to replace simulations.
Expand project types: Move beyond industrial emissions to forestry, agriculture, and blue carbon, each with tailored verification agents.
Decentralized oracle integration: Use Chainlink or similar to bring AI verification results on-chain in a trust-minimized way.
Registry interoperability: Connect with existing carbon registries (Verra, Gold Standard) for credit retirement synchronization.
Security audit and mainnet deployment: Conduct a smart contract audit and deploy on Polygon mainnet or Hedera for production use.
Governance and standardization: Propose the Integrity Score as an open metric and explore DAO-based governance for verification agent updates.
VeriCarbon started as a hackathon response to a trust crisis—and we believe it can grow into a real infrastructure layer for credible climate action.
Log in or sign up for Devpost to join the conversation.