Expert Interview

Jonathan: Steve, thank you for joining me. We are at The AI Summit London, one of the biggest AI events in Europe. Why is this conference so important for IBM?

Steve: Europe is becoming a major opportunity for AI adoption because of the concentration of regulated industries such as healthcare, telecoms, banking and insurance.

It is extremely challenging to scale AI across those sectors. We see this event as a way to move efficiently from pilots to scale.

Jonathan: I agree; Europe is an incredible environment for experimentation. IBM has been talking a lot about the AI operating model. What does that mean, and how is IBM’s approach different from your competitors?

Steve: Much of the market is focused on building agents, but primarily within their own ecosystems.

IBM focuses on doing this across heterogeneous environments. That is challenging and requires a clear AI operating model that scales. It starts with taking advantage of your underlying infrastructure, whether in the cloud or on premises.

You need flexibility and control at the infrastructure layer, then a deep understanding of where your data resides and how to access it safely and securely, with the right context and accuracy to fuel decision making. On top of that, you need an enterprise agent control plane.

The control plane sits across all of this to observe, evaluate, optimise and govern agents so they can scale across heterogeneous environments.

Jonathan: We are hearing a lot about that operating model at The AI Summit this week. It is day one, and IBM is the primary sponsor. What highlights can visitors expect?

Steve: One is what we are doing with watsonx Orchestrate as our enterprise agent control plane, which allows you to take advantage of your growing agent estate, regardless of where agents are built or run, and to control and observe them across your environment.

Another is IBM Bob, a coding platform that supports developers across the entire software development life cycle.

We are also focused on sovereign capabilities, given the importance of compliance and regulation in Europe, to provide flexibility and control across your data and infrastructure estates.

Jonathan: There is a lot of buzz about AI agents. What are you seeing from enterprise customers? How is adoption progressing?

Steve: Every customer we speak to is focused on building agents. The conversation is shifting from how to build to how to maintain control, optimise placement and execution, and manage cost as agents proliferate across the estate.

We have seen a notable shift over the past six months towards governance, observability and optimisation.

Jonathan: You also mentioned IBM Bob, which is getting a lot of attention. How is it changing the software development life cycle?

Steve: Many coding agents and chatbots focus on specific tasks like generating code snippets. IBM Bob differentiates by understanding intent and the full development life cycle, helping developers plan and understand complex legacy code bases and modernise them.

Internally at IBM, over 80,000 developers use Bob, with consistent productivity gains of up to around 40 per cent over existing tools.

Jonathan: One use of Bob is to build agents. The wider conversation is shifting from creation to governance. Why is that happening now?

Steve: The past few years focused on experimentation, and there is clear value.

Many organisations are stuck moving from pilot to scale. To do that, they need to address compliance and security and shift left.

There was a perception that governance would slow you down. The reality is the opposite: if you do not address compliance, governance and data security upfront, you inhibit your ability to scale. That shift is now widely recognised.

Jonathan: watsonx Orchestrate is IBM’s platform for building, deploying and managing agents. How is it evolving to tackle those challenges?

Steve: When we first released watsonx Orchestrate, it focused on orchestration. It has evolved into an enterprise agent control plane.

It provides the ability to observe, evaluate and optimise, with policy controls so agents run safely and securely across the enterprise, regardless of where they are built or run.

Jonathan: Agents are only as powerful as the data they use. We are hearing about data becoming increasingly real time. How does that impact AI and IBM’s strategy?

Steve: Today, we estimate that less than one per cent of enterprise data is represented in large language models, meaning 99 per cent is not being used by AI. Much of that is critical and sensitive customer data, so you need the right controls, security and compliance to access it.

IBM’s approach is to give you the control to use data across your AI environment, wherever it sits in your toolset, and to leverage real‑time data. AI is no longer about static databases; it is real time. You will see capabilities not only in our watsonx.data portfolio but also through real‑time integrations from Confluent.

Jonathan: Looking ahead, what will the enterprise AI landscape look like over the next few years, and where does IBM fit?

Steve: Enterprises will operate large ecosystems of agents embedded across departments and teams. Those agents must operate safely, securely and cost efficiently.

IBM is focused on taking advantage of the capabilities, platforms and tools organisations already have and making AI work within those environments.

Closing

This conversation reinforces a decisive shift in enterprise AI: from building isolated agents to governing ecosystems that span cloud and on‑premises environments. 

IBM’s emphasis on an AI operating model, anchored by an enterprise agent control plane in watsonx Orchestrate, places observability, policy and safety at the centre of scale. Real‑time data has become essential, yet most enterprise information remains untapped due to security and compliance constraints. 

IBM’s strategy aims to unlock that data responsibly while improving developer productivity with IBM Bob. For regulated European industries, the message is clear: plan for governance and data security upfront, design for heterogeneous environments, and invest in control planes that provide visibility and policy enforcement. That foundation turns pilots into production and keeps agents safe, performant and cost‑efficient.