Fiduciary grade AI sets the bar as Thomson Reuters and Snowflake bring governed intelligence to the professions

Professionals such as lawyers and auditors require AI systems that are reliable and accurate, as they bear personal liability for their decisions. Thomson Reuters and Snowflake are addressing this need by developing "fiduciary-grade AI," which is built on governed, authoritative data rather than being a constraint on adoption. This approach allows for the deployment of agentic systems that can handle complex tasks while maintaining trust. Thomson Reuters' data estate, spanning over 37,000 governed tables and 350 databases, serves as the foundation for these AI tools. The company's legal assistant, CoCounsel, has evolved into a fully agentic system that leverages authoritative content like Westlaw and Practical Law. Crucially, Thomson Reuters' responsible AI team vets every capability for hallucinations and bias before release, ensuring that the human lawyer remains in control of the process. This same standard of governed intelligence is applied internally, where a semantic capability on Snowflake unifies fragmented customer master data sources, providing finance teams with a single, trusted definition of business terms. The partnership points toward a future of enterprise-scale semantic intelligence that serves the public interest while being fit for purpose.

(Source：Siliconangle)