When the data intelligence company Alation Inc. announced its integration with the Salesforce Data Cloud earlier this month, it provided another example of how businesses were seeking to harness the power of metadata and maximize AI initiatives.
The integration of Alation with Data Cloud will enable Salesforce Inc. customers to govern their data estate using essential information, such as tags, governance policies and data quality indicators, while building accurate and compliant AI models.
“Customers consistently leverage Salesforce as a platform, which means they customize it to a great extent,” said Satyen Sangani (pictured), co-founder and chief executive officer of Alation. “That customization obviously is super useful for customers to achieve their own business outcomes and customize Salesforce to their business, but really, really tough to contend with when you’re trying to analyze the information at scale or leverage it in models. What we do is help them understand what they’ve done in order to be able to help them leverage the data in the best possible way.”
Sangani spoke with theCUBE Research’s Christophe Bertrand at Dreamforce, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed data intelligence and how Alation provides data fidelity for AI solutions.
Data intelligence: Ensuring model trust
As organizations build AI models tailored for their needs, the challenge becomes how to gain meaningful insights and govern information appropriately from structured and unstructured data sources. A customer service interaction chatbot might draw from existing structured knowledge bases while processing real-time customer feedback in unstructured form. Alation’s solution addresses the need for data fidelity to help ensure model trust and accuracy.
“It’s all about fidelity in the actual models and making sure that those models are correct,” Sangani explained. “To tell these large language models how to interact with these structured data systems is tricky, because LLMs are mostly fed off of and developed by leveraging unstructured data. You’ve got these two different pipelines, and then the question is how do they come together and make sense for the customer?”
Alation’s technology is designed to provide the answer through automated ingestion of critical metadata and making trusted information easily discoverable. The company’s AI-assisted curation and intelligent search enable customers to manage data and models to optimize outcomes from repositories such as the Salesforce Data Cloud.
“One could describe Alation as a metadata system or a metadata reference system, and we certainly glean and take metadata from all the data sources that we connect to,” Sangani said. “We’re not a data storage system. We don’t try to replicate all of the information that you have. We’re just trying to learn about the information through the information that describes it. The metadata is actually much more critical to determining whether something is useful or correct or usable in a certain context.”
Here’s the complete video interview, part of SiliconANGLE’s and theCUBE Research’s coverage of Dreamforce 2024: