Agentic AI’s rise is making the enterprise architect role more fluid
The CIO’s key lieutenant has to become more business focused to better chart the business processes. It’s a key requirement for successful agentic AI adoption, which, in turn, will drive greater agility.
In a previous feature in CIO magazine about enterprise architects, generative AI had emerged but its practical impact on enterprise technology had not yet been widely felt. Today, it has sparked a wave of agentic AI solutions from major SaaS providers, and that is reshaping both enterprise architecture and the enterprise architect role. CIOs and architects are moving from designing around relatively stable applications and integrations to planning operating models where agents can execute parts of business processes.
The momentum is being driven from the top, with CEOs increasingly vocal about using AI to boost productivity and restore growth, a view echoed by analysts. Gartner, for example, forecasts that 75 percent of IT work will be completed by human employees using AI over the next five years, which it says will require a proactive search for new value creating IT work such as entering new markets, launching new products and services, or adding higher margin features. If that productivity leap happens, organisations will need to redesign business processes and the technology that runs them, because recent history suggests that without new operating models, the expected returns from technology investments do not fully materialise.
Agentic AI raises two immediate architecture challenges: process change and complexity. CIO magazine notes warnings that the idea of giving everyone a bot is simplistic, and that organisations must analyse roles and processes to avoid deploying agents that are unnecessary, too limited for complex work, or that inflate cloud costs. Governance and control become critical as the number of agents grows, particularly around data segregation and the ability to manage agents once published. A further architectural focus is the difference between deterministic and non deterministic systems, with non deterministic outcomes needing guardrails, often supported by deterministic components, and with experimentation seen as essential to identify where AI can strengthen enterprise architecture and deliver clear business outcomes.
