Enterprise AI governance critical as AI scales

With research estimating that two-thirds of AI pilots stall before reaching production scale, most organizations are clearly not only behind on AI — they’re stuck on the starting line. Still, at the same time as different corporations take different approaches to AI, interoperability and governance have emerged as clear cornerstones, based on Murali Swaminathan (pictured), chief technology officer of Freshworks Inc.

“There’s quite a lot of change that has happened within the AI world. There’s a lot proliferation of AI-native tools [and] third-party products. Everybody is doing AI differently,” Swaminathan said. “There’s no one-size-fits-all — every little thing needs to attach together. We’d like to interoperate with other AI systems, and that’s the training lesson here.”

Swaminathan spoke with theCUBE’s Bob Laliberte on the Freshworks Refresh event, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed enterprise AI governance in addition to the shift from human-assisted workflows toward more autonomous AI-driven operations. (* Disclosure below.)

The importance of enterprise AI governance

Together with interoperability, trust has emerged as a foundational requirement for any successful AI adoption, especially as organizations begin delegating more work to automated systems. To construct confidence in AI-driven outcomes, enterprises need visibility and assurance that the technology is working as intended, based on Swaminathan.

“[You need] trust in the way you arrange the AI, the way you arrange the controls, the way it performs, how you’ll be able to trace what’s happening after which how we will report on what it did and what it didn’t do,” Swaminathan said. “Trust is there in lots of layers, and that’s what we imagine in constructing the product or providing the tools so that you can set it up the proper way.”

Trust in AI requires layered governance, and Freshworks approaches that through controls built at every level of the stack — from LLM guardrails that outline what’s permitted, to agent-level boundaries that keep automated systems focused on their intended tasks, Swaminathan explained. Data anonymization and sovereignty policies add further guardrails, ensuring personal information is protected and that data stays inside designated regions. Organizations are increasingly focused on these sorts of safeguards to make sure AI systems operate securely in a way that feels invisible to the user and indispensable to the administrator.

“We deal with data sovereignty — we make certain that the information stays within the zone and doesn’t go outside,“ he said. “So far as the user is anxious, it doesn’t matter to them, but behind the scenes, the people who find themselves governing it have to know that it’s working the proper way.”

Here’s the entire video interview, a part of SiliconANGLE’s and theCUBE’s coverage of the Freshworks Refresh event:

(* Disclosure: TheCUBE is a paid media partner for the Freshworks Refresh event. Neither Freshworks, the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE