The Agentic Enterprise – The evolution of enterprise architecture in the AI era

Over the past few months, it has been practically impossible to avoid the discussion on the Agentic Enterprise. So, it was no surprise that the Agentic Enterprise was also one of the main themes at Dreamforce 2025. For a while now we have been told that AI agents will become an integral part of how work gets done – embedded in workflows, collaborating with people, and driving action. But what does the Agentic Enterprise mean in all of this?

Quite simply the Agentic Enterprise is a new way of running an organization, in which intelligent agents (AI systems with some autonomy) and humans work together. Essentially, AI agents take on more proactive, goal-oriented, multi-step responsibilities rather than passive automation.

The essence of the Agentic Enterprise can be defined by a few characteristics:

These agents can reason, plan, and act within defined boundaries and guardrails to achieve business goals.
They are embedded into workflows, not just as mere chatbots or assistants, but as “digital workforce” components that work alongside human workers.
The enterprise architecture needs to evolve from traditional layers to new architectural layers supporting agents across data, integration, tooling, and governance.
This is much more than automation. It is about dynamically adaptive systems, cross-functional orchestration, and humans shifting to higher value work while agents handle scale and 24/7 tasks.

Visionary or merely ”enterprise-washing”?

The Agentic Enterprise is an important evolution of enterprise architecture in the AI era: one, in which agents (AI systems with autonomy, reasoning, and action) become integrated alongside humans in workflows and operations. If we use the term Agentic Enterprise when, in fact, we simply mean automation + chatbot there is a danger that we are “enterprise-washing” our business. Much like whitewashing it can result in a lot of hype over not much content or proof.

In order to avoid ”enterprise-washing”, we have to design and implement agentic AI with depth and discipline. Simply renaming automated processes as “agentic” results in a hefty amount of hype without real business value.

To turn the Agentic Enterprise vision into reality, you will need the right architecture, data and integration maturity, organisational commitment, governance framework, and phased execution to avoid shallow implementations.

How to succeed with the Agentic Enterprise?

At Dreamforce 2025 the Agentic Enterprise was positioned as a real architectural shift with products, architecture guidance, and transformation and governance talk. This shift will pivot from copilots to agents and the need for governance, observability, and security at scale.

This change is far from being simple and requires a deep understanding of the following key success factors:

1. Start with business goals and use cases.

Instead of starting from narrow tasks, select use cases where autonomy and orchestration bring clear value (e.g., end-to-end workflows and cross-functional orchestration). Avoid shallow, never-ending pilots without scaling, i.e., define how to scale your pilots across the organisation and make them operational.

2. Pay careful attention to foundational architecture and data readiness.

Legacy architecture often lacks the layers needed for agents. Agents rely on deep, reliable access to systems of record, real-time data, and the ability to execute actions. If systems are siloed, the agents will fail. This highlights the importance of data, integration, and APIs.

3. Plan governance, robust controls, and ethical and operational oversight.

Governance, security, and auditability are needed, as agents act autonomously. You must guarantee that agents’ decisions are understandable, traceable, aligned with your business policies, and compliant with regulations. As part of risk mitigation, you must define what level of autonomy is acceptable, where you need to keep your human workers in the loop, and how to handle failures.

4. Plan for organisational change management.

Your employees must understand how to work with agents, what the boundaries are, and how decisions get made. Your culture must support and enable transparency and drive trust in AI, as well as willingness to delegate to agents within designed boundaries, of course. It is inevitable that human roles shift from “task-doers” to “agent-bosses”, oversight, exception handling, and strategy.

5. Treat the Agentic Enterprise vision as a program (data + integration + governance + operating model), not a feature or one-time project.

Look for concrete enablers, such as agent orchestration, policy guardrails, observability, and clear value use cases.

6. Always aim for scalable implementation.

Monitor metrics and outcomes beyond “tickets closed”. Measure productivity, value creation, and customer and employee experience. Maintain flexibility and optimisation, as models, tools, and pretty much everything evolves very quickly. Your architecture must allow swapping components and updating strategies.

From passive systems to autonomous workflows

The Agentic Enterprise marks a pivotal shift in enterprise architecture from passive systems to autonomous, outcome-driven agents embedded in every workflow. It is truly both a visionary concept and a test of leadership maturity. Success depends not on technology alone, but on the clarity of purpose, governance, and trust between humans and machines.