Agentic AI: Moving beyond pilots to enterprise impact

Using a combination of autonomy, planning, memory, and integration, agentic AI has the potential to automate complex business processes. Agentic AI can be a company’s goal-driven virtual collaborator rather than a reactive tool. It can supercharge operational agility, create new revenue opportunities, and provide a competitive edge.

However, although 80 percent of companies are using generative AI in some form in at least one business function, its impact on the bottom line is not yet material at the enterprise-wide level. During a McKinsey Live event, partner Dave Kerr and senior partner Klemens Hjartar discussed the state of agentic AI at forward-looking companies, the importance of being in the race to realize agentic AI’s full potential and be an early adopter, and the requirements to scale.

It’s still early days for agentic AI, but realizing its full value requires transformation, not just new technology. Leading companies are resetting their AI transformation by reimagining their structures, processes, and workflows from the ground up and using agents to automate end-to-end workflows. These companies are shifting from siloed vertical deployments of individual initiatives to an integrated and orchestrated ecosystem.

Capturing the value of agentic AI requires wielding the power of an agentic AI mesh—a composable, distributed, and governed architecture that enables multiple agents to reason, collaborate, and act autonomously across systems in a secure and scalable way. The agentic AI mesh becomes the connective tissue of the enterprise; it coordinates the work of multiple agents, integrates with core systems, and evolves along with the technology itself.

Resetting a company’s approach to AI transformation is no simple task. Among the requirements:

• An enterprise-wide AI/agentic strategy. Senior leaders are aligned with the strategy, and enterprise-level programs tie directly to the company’s biggest priorities. Agents are used to reimagine business models and create new sources of value as well as propel efficiency.
• These critical capabilities:
  • People. Employees are equipped for a world of human–agent collaboration. Some companies have introduced new roles such as agent orchestrator and human-in-the-loop designer.
  • Governance. A strong governance framework defines agents’ autonomy levels, decision boundaries, and overnight mechanisms.
  • Tech architecture. Over time, enterprise architecture will be reshaped by agents as they become the connective tissue of day-to-day operations.
  • Data. Agents can only be as effective as the data they rely on.
• Adoption, scaling, and implementation. To attain industrialized, scalable delivery, systems for integration and monitoring are thoroughly thought through in advance and designed for efficiency from day one. Systems are technically and economically sustainable and allow the company to react to changes with agility.

* * *

For more on this topic, see the research Seizing the AI advantage, The state of AI: How organizations are rewiring to capture value, Superagency in the workplace: Empowering people to unlock AI’s full potential, and the McKinsey Quarterly article “Why agents are the next frontier of generative AI,” all on McKinsey.com.