An agentic oracle architecture for vehicular carbon microcredit tokenization
Agentic Oracles, Permissioned Blockchain, Carbon Credits, Large Language Models (LLMs), Digital Sustainability.
The transportation sector poses a central challenge for climate change mitigation, requiring automated, reliable Measurement, Reporting, and Verification (MRV) mechanisms. Currently, the tokenization of vehicle carbon credits faces limitations due to the use of traditional passive oracles, which lack autonomous governance and the ability to validate context. In this context, this work proposes an agentic oracle architecture for carbon microcredit tokenization. The system was designed as a decarbonization incentive strategy that rewards emission reductions through Artificial Intelligence and Large Language Models (LLMs). The methodology consisted of developing a multi-agent structure specifically designed to support the stages of an MRV cycle, coordinated by a moderator-orchestrator. The mechanism performs vehicle telemetry processing for CO₂ calculation, anomaly detection, and governance of deliberations. To validate the architecture's viability, the workflow was implemented using smart contracts on the Hyperledger Besu permissioned blockchain, enabling asset issuance at zero cost. Tested with real and synthetic driving session data, the results demonstrated that LLM-based agents enable autonomous, explainable governance with on-chain, auditable records, enabling traceability and efficiency in digital environmental markets.