With software consuming the computing world, artificial intelligence is the flavor of the moment.

Several tech advances, such as computer vision and, more particularly, generative AI, are unraveling a swathe of new capabilities. Many forward-thinking companies are already investing to drive increased end-user value and stay competitive.

“We have been working on generative AI for some time,” said Bratin Saha (pictured, center), vice president and general manager of AI and machine learning at Amazon Web Services Inc. “Last year we released CodeWhisperer, which is about using generative AI for software development, and a number of customers are using it and getting real value out of it. So generative AI is now mainstream and enterprise users can use.”

Saha; Tom Mason (right), chief technology officer of Stability AI Ltd.; and Aidan Gomez (left), co-founder and chief executive officer of Cohere Inc., spoke with theCUBE industry analyst John Furrier at the AWS Startup Showcase: “Top Startups Building Generative AI on AWS” event, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed the increasingly potent use cases for generative AI and how all three companies are addressing them. (* Disclosure below.)

Partnerships galore toward shared goals

Key to AWS’ efforts to explore generative AI is industry partnerships, and both Stability AI and Cohere form a core part of that ecosystem. It’s approaching the goal from three areas, the first of which is creating efficient, purpose-built infrastructure. Following that comes the AI models and, lastly, a bustling application ecosystem, according to Saha.

Stability AI cut its teeth in creating capable open-source foundational models for AI operations, and AWS has, over the years, strengthened ties with the company to train some of those models within its garden, according to Mason.

“Stable Diffusion was our first big model which we trained on AWS,” he explained. “And we’re excited to take it further this year as we develop a commercial strategy of the business and build out the ability for enterprise customers to come and get all the value from these models that we think they can get.”

Some of the planned updates coming soon from the Stability/AWS camp include new modalities, video models and more consistent handling of image recognition.

Cohere, for its part, builds large language models for the enterprise, at a similar scale as ChatGPT’s underpinnings. And one limiting factor standing in the way of LLMs reaching the enterprise mainstream, as the company has noticed, is the difficulties in creating polished products out of them, according to Gomez.

“About six months ago, we released our command models, which are some of the best that exist for large language models,” he explained. “And, in December, we released our multilingual text understanding models. And that’s been trained on over a hundred different languages with authentic data from native speakers.”

Where LLMs are headed

Despite the already brilliant capabilities that LLMs are showing with products like Bard and ChatGPT, their range of use is still quite limited. Advancing their potential would imply things like an external knowledge base, where these models stay up to date in real time. Equipping these AI tools to access and use APIs, for instance, can greatly improve how they approach end-user queries.

“What happens when you give these models the ability to use tools, to use APIs?” Gomez said. “What can they do when they can actually effect change out in the real world, beyond just streaming text back at the user? I think that’s the really exciting piece.”

But while laying projections for the future, the present must always be evaluated. And in the meantime, early adopters within the AWS product ecosystem are already leveraging the visual and textual models from Stability and Cohere.

“One of the things we have been seeing, both with the text models and the visual models that stability.ai does is customers are really using it to change the way you interact with information,” Saha said. “One example is of a customer that we have who’s using that to query customer conversations and ask questions around customer issues and how to solve them, in addition to trying to get those kinds of insights that were previously much harder to get.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of the AWS Startup Showcase: “Top Startups Building Generative AI on AWS” event:

(* Disclosure: This is an unsponsored editorial segment. However, theCUBE is a paid media partner for the “Top Startups Building Generative AI on AWS” event. Amazon Web Services and other sponsors of theCUBE’s event coverage have no editorial control over content on theCUBE or SiliconANGLE.)