Summary
This initiative leverages cutting-edge AI models and institutional-grade oracle infrastructure to create unified golden records of corporate action data that work across multiple blockchains. Participants included Euroclear, Swift, UBS, Franklin Templeton, Wellington Management, CACEIS, Vontobel, and Sygnum Bank.
Key Takeaways
Corporate actions present one of the most complex unstructured data problems in the financial world. Initially presented in human-readable formats like PDFs and press releases, this critical information—covering mergers, dividends, and stock splits—undergoes a complex journey through custodians, brokers, fund managers, exchanges, and ultimately investors.
By using Chainlink oracles paired with multiple large language models (LLMs), we were able to source unstructured, offchain data and convert it autonomously into structured, onchain data that is available in near real-time and into predefined standards of specific corporate action types, modeled on the ISO 20022 framework and aligned with the Securities Market Practice Group (SMPG) guidelines.
This approach not only reduces uncertainty around using novel LLM models but also distributes data nearly instantaneously across public and private blockchains to provide a unified golden record for asset managers, CSDs, and custodian service providers that is consumable via Swift’s messaging protocol.