Salesforce Unleashes Power of GenAI on Unstructured Data

Salesforce has announced the general availability of its Data Cloud Vector Database, which uses GenAI to quickly sift through a company’s trove of PDFs, emails, transcripts, online reviews, and other unstructured data.

The ability to use that previously untapped data quickly will improve efficiency by “orders of magnitude,” says Rahul Auradkar, executive vice president and general manager of Salesforce Unified Data Services and Einstein Units. He chatted with InformationWeek in a one-on-one briefing to show a live demo of the new capabilities.

Data Cloud will work with the Einstein 1 platform to link unstructured and structured data to allow quick data analysis for sales teams, marketing teams, customer services teams and more. The tool improves the accuracy of Einstein Copilot, the company’s conversational AI assistant for enterprises, the company said.

Auradkar during the briefing quickly pulled up a query based on a customer service request. Within seconds, the program showed multiple related hits that could complete the request. It would have taken an agent or even multiple agents hours to sift through the unstructured data, which accounts for 90% of customer data.

“What this advancement means is that our customers can use the full power of 90% of enterprise data -- unstructured data that is locked up in silos and underutilized or not unitized at all -– to drive use cases, AI, automation, analytics experiences across structured and unstructured data.”

Related:Salesforce CTrO on Transparency and Trust in the Era of GenAI

Using Salesforce’s Einstein 1 platform, Data Cloud will enable users to ingest, store, unify, index and allows semantic queries of unstructured data across all applications, the company said. That data can include disparate unstructured content from websites, social media platforms, and other sources -- this broader inclusion of data will allow for more accurate results and outcomes, the company says.

“I would say it’s an order of magnitude improvement in productivity and customer satisfaction,” Aurdakar says. “For example, a large shipping customer might have thousands of customer cases to sift through. And they probably find different manual ways of categorizing them for they can get the information they need. So, this would be a vast improvement.”

Other Announcements

Saleforce also announced new AI and Data Cloud capabilities, including a pilot for hybrid search that will combine semantic search with traditional keyword search; data cloud in sandbox environments that enables IT teams to configure and test innovations before deployment; and Data Cloud on Hyperforce in the UK available July 31.

Auradkar tells InformationWeek that gains competitive advantage by investing in flexibility and letting customers take ownership of their own data.

“We’ll continue with that journey,” he says. “I think the biggest areas that we are investing in going forward is what is the world going to look like as this product grows and grows at scale. We’re building a significant amount of flexibility for our customers to use our gateway and use any model they want to use any large language model they want to use.”