Hype, investment, innovation and market results — every breakthrough technology has followed a version of this path to dominate the mainstream.

With artificial intelligence and, more specifically, generative AI having reached the innovation stage, what signals are emerging as it primes to deliver on the lofty cross-sector promise?

“I think where we are right now is there’s a lot of potential to move these pilots and [proofs of concept] to production, but the organizations who are going to be successful with that are those who have spent the time to identify the right use cases from the get-go,” said Lilly McNealus (pictured, left), director of outbound product management, cloud artificial intelligence, at Google LLC. “Organizations who have just deployed agents or models without really thinking through where do they have the opportunity to generate value for the business, they are never going to see that ROI.”

McNealus and Rajesh Abhyankar (right), senior vice president at Persistent Systems Ltd., spoke with theCUBE Research’s John Furrier at the “Accelerating Innovation With Persistent” event, during an exclusive interview on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed the shifting tides from hype to real progress, underlined by the recent keynotes from Google Cloud Next, serving as a barometer for industry movement. (* Disclosure below.)

Organizations must navigate gen AI complexity to reap rewards

From adapting business-specific use cases to technical and skill-based demands, companies must brace themselves to tackle several hurdles on their gen AI journeys. Also important is a platform-centric mindset, with robust tooling and services tailored for machine learning operations, according to McNealus.

“If you don’t have the right tooling in place, you’re not able to customize those models, you can’t deploy them at scale. And if you don’t have the right services around MLOps to monitor those models and deployment, you’re never going to realize the benefit,” McNealus said. “I think there’s a lot of potential, but there’s a lot of work that needs to happen for organizations to realize the real benefit of generative AI.”

Large AI technology platform providers, such as Google, provide the full range of tooling that ecosystem partners can harness and adapt to their existing operations. However, it’s up to these companies to leverage the tooling in an effective, purposive manner, according to Abhyankar.

“Google has Model Garden and all the right tooling and Vertex AI,” he said. “The platform and everything that Google can do as a platform provider is there. But the key question that enterprises are asking us is, ‘How do I make that work in my ecosystem?’ There are strict governance and security requirements, and there’s so much going on in an enterprise … and that’s where we come in.”

Unlike their predictive counterparts, gen AI models need specialized services spanning prompt management, evaluation and model observability. The advent of automated side-by-side comparison and rapid evaluation heralds a new era, augmenting the arsenal of tools that will empower enterprises in their respective journeys, according to McNealus.

“What I think is going to be interesting is those organizations who can take that next step and launch consumer-facing applications powered by generative AI,” McNealus said. “We’re starting to see that today, and I think we heard some examples in the keynote. I think that’s where we’re going to realize a lot of the magic of this technology is when we move from just business-to-business and interior facing to business-to-business-to-consumer.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE Research’s coverage of the “Accelerating Innovation With Persistent” event:

(* Disclosure: TheCUBE is a paid media partner for the “Accelerating Innovation With Persistent” event. Neither Persistent Systems Ltd. nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE