At the AI4 conference in Las Vegas, Worth and IBM co-hosted a candid conversation about the future of enterprise AI. On stage, I sat down with Sunil Murthy, IBM’s Field CTO for Data and AI, to unpack how businesses can move from AI experimentation to operational impact.
“I’m in the field talking to three customers a day,” Murthy told the audience. “The most common challenge I hear is: Where do we start?” His answer starts not with a model or tech stack, but with customer requirements—especially governance and compliance. “In highly regulated industries like healthcare, governance isn’t an afterthought—it’s the driver,” he explained. One recent healthcare engagement shaped the AI solution from the ground up around regulatory demands, and those lessons now influence IBM’s governance tools across sectors.
That governance-first mindset is core to IBM’s approach. Through watsonx Orchestrate, IBM helps enterprises build and deploy pre-trained AI agents that integrate seamlessly with existing systems. “You need AI that hits the ground running—AI that knows your business almost as well as you do,” Murthy said. The platform orchestrates workflows across departments, reduces manual work, and accelerates decisions without sacrificing compliance or interoperability.
Integration, Murthy stressed, is often where enterprises struggle. Many already have AI inside platforms like Salesforce or SAP, but they don’t always work together. “Our focus is on augmentation,” he said. “We look at the AI routines that come with those platforms and see where we can add orchestration and intelligence.” IBM’s partnerships with SAP, Salesforce, ServiceNow, and Adobe are designed to extend—not replace—the value clients already have in their IT stack.
Real-world deployments make the case. Murthy cited a major Airline kiosk at one of the busiest Airports as a prime example of AI orchestration in a high-volume environment. By automating workflows and integrating across systems, IBM helped streamline passenger check-ins and service requests at scale—exactly the type of outcome enterprise leaders want to see before committing further investment.
Still, scaling AI isn’t just technical—it’s organizational. Developers are juggling as many as 15 different tools to build AI applications, according to IBM research. “Fragmentation slows you down,” Murthy said. “Our goal is to unify the experience without locking clients into a closed system.” That balance—between reducing complexity and preserving choice—is why watsonx supports open models alongside IBM’s Granite models and third-party options from Meta, Mistral, and others.
Choosing the right model, he added, is a business decision as much as a technical one. “It impacts your brand, your costs, and your outcomes.” IBM’s curated model library is designed to help enterprises select the right tool for the right job—and do it responsibly.
Looking ahead, Murthy predicted that AI agents will become more collaborative, autonomous, and deeply embedded in business operations over the next three years. But he warned that orchestration remains underestimated. “An AI agent isn’t just a chatbot with a fancy brain,” he said. “It’s a set of capabilities that need to work together, securely, across multiple systems and partners.”
For the AI4 audience, the takeaway was clear: the messy middle—where strategy meets integration—is where pilots either scale or stall. As Murthy put it, “Our role is to meet clients there, understand their requirements, and get them to value faster. If we do that, AI stops being an experiment and starts being part of the business.”
