These records were distributed simultaneously to blockchain networks and legacy systems like the interbank messaging system SWIFT, significantly reducing manual work and the risk of error.
The process used a blend of large language models, including OpenAI’s GPT, Google’s Gemini, and Anthropic’s Claude, to extract structured data from unstructured corporate action announcements. These were then published as unified gold records on-chain to create a “single source of truth that all participants can easily access, verify, and build upon.”
Chainlink’s Runtime Environment (CRE) validated model outputs, while its interoperability protocol (CCIP) relayed data to blockchains, including Avalanche and DTCC’s private network.
Data attesters cryptographically attested the outputs and contributed to potentially missing data fields. According to Chainlink, the system achieved a near 100% data consensus across all test events.
The current system for processing corporate actions is costly. Citi’s 2025 Asset Servicing report shows that the average corporate action touches 110,000 interactions and costs $34 million to process. The global financial industry is now spending an estimated $58 billion annually in processing corporate actions.