According to a recent LinkedIn post from MotherDuck, Canva staff data scientist Sam Redfern is scheduled to speak at the Data Outpost event on November 5 about a technique called “semantic leaking.” The post indicates this method uses large language models to extract business language from technical artifacts and integrate it into a semantic layer for analytics.
The company’s LinkedIn post highlights a focus on improving how AI-driven analytics interpret ambiguous business terms such as “active customer.” While the content primarily promotes the event, it also suggests ongoing industry efforts to make AI analytics tools more transparent about data limitations, which may influence expectations for future AI-enabled data products.
As shared in the LinkedIn post, the described technique aims to help AI systems better understand when data cannot reliably answer a user’s question. For investors following the broader data and analytics ecosystem around MotherDuck, this emphasis on semantic and AI tooling may signal a growing market interest in enhanced data governance and interpretability solutions, even though the post does not reference specific commercial offerings or financial impacts for MotherDuck.