Governed context is becoming a practical requirement for artificial intelligence agents expected to make decisions across a business. As enterprises move beyond personal productivity tools, those agents need information that reflects changing conditions across the systems they use.

IBM Corp.’s approach connects that data foundation to the broader challenge of putting AI into production. A large language model’s reasoning capabilities depend on the information available to it, making data delivery an operational concern, according to Shaun Clowes (pictured), general manager of Confluent and Data at IBM.

“You wouldn’t give your accounts receivable work to a nuclear physicist that worked on your energy production side of your business,” he said. “So, why would you take a very, very smart model, give it all of the possible information about your business and say, ‘Figure out accounts receivable?’ You need the right context for the job.”

Clowes spoke with theCUBE Research’s Dave Vellante and John Furrier for IBM’s AI Operating Model interview series, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed real-time data, agent specialization and production governance ahead of IBM TechXchange 2026. (* Disclosure below.)

Building governed context into the data flow

Enterprise data architectures have evolved around changing storage, computing and application requirements. Agents introduce a need to combine information across operational systems, such as travel bookings, baggage tracking and rental cars, while that information is still relevant to a decision, according to Clowes.

“Now we need a new type of architecture, a type of architecture that can bring together all the state of our business right now and present that directly to LLMs,” he said. “We call that the streamhouse architecture.”

IBM Confluent brings together streaming, connectivity, processing, governance and querying capabilities to support that approach. Rather than requiring all information to be centralized first, the platform supports the preparation of usable data products can happen across the environments where business information originates, Clowes explained.

“Our argument would be that you can actually move all of that very, very close to where the data is first produced, as close as you possibly can,” he said. “The earlier you govern it, the better. The quicker you produce the high-value data, the better. Then you need to just connect up to all the relevant places.”

Giving specialized agents room to reason

Delivering current, governed data is only part of the challenge. Agents also need context tailored to the work they’re performing. Different teams may use the same business term to mean different things. Preparing context for an agent therefore involves establishing the definitions that apply to its task, according to Clowes.

“You can describe, ‘Hey, in this business process, what I mean is X,’” he said. “Then, you can have these governed metrics and these governed datasets that effectively represent the foundation upon which the model will reason.”

Enterprises can begin with high-value tasks and improve individual agents as they gain experience. Once those agents reliably execute established business processes, the opportunity expands to examining how the work itself is organized, Clowes explained.

“Then you can say to the supervising agent, ‘Do I even need these processes? Should I just do something totally different?’” he said. “The best bit about this whole system is it can learn at every layer.”

Making business context available in production

As agents take on more responsibility, access to current business information becomes increasingly important. Moving from human-assisted decisions to autonomous operations changes how businesses must deliver information. Agents working continuously need access to operational data without waiting for a person to gather it, Clowes explained.

“People used to talk about data as the new oil, but what they really meant was [that] data, in some sense, existed and was used by humans to make good decisions,” he said. “Businesses were run by humans … [and] that’s how you win.”

Information may originate on oil rigs, cruise ships or other remote systems. Enterprises must connect those sources into a usable operational view without assuming one centralized solution will contain everything, according to Clowes.

“You should always be asking yourself, ‘How can I know the state of my business right now?’” he said. “Because that is the world in which you make decisions. That is the world in which humans would make decisions, too.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of IBM’s AI Operating Model interview series:

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

Photo: SiliconANGLE