A new class of artificial intelligence has arrived, and it doesn’t fit neatly into any existing box. Agentic AI systems can plan, act, and learn on their own. They share qualities of both tools and people.
According to The Emerging Agentic Enterprise, the ninth annual MIT Sloan Management Review–BCG report based on a survey of 2,102 respondents in 21 industries and 116 countries, 35% of companies are already starting to use agentic AI and an additional 44% plan to soon. Yet nearly half say they lack a strategy for what to do with it.
That is both surprising and understandable. The technology’s dual nature—as both a tool and a colleague—defies the traditional divide between capital and labor.
Managing the Duality
The way companies typically treat a capital investment is very different from how they treat a human investment. Agentic AI has characteristics of both. Like capital, agentic AI systems require substantial investment upfront in development, integration, and infrastructure. Like a human workforce, they also require ongoing variable costs for evaluation, tuning, domain adaptation, usage-driven inference, and governance.
This blend of capex and opex makes agentic AI an economic hybrid, fundamentally reshaping enterprise investment models.
Redesigning for an Agentic Future
Managing agentic AI requires cooperation across boundaries, especially IT, HR, finance, and operations.
Here are five practical steps:
- Rethink how work gets done. Don’t just automate tasks—redesign jobs and processes so people and AI can switch smoothly between doing things efficiently and exploring new ideas.
Rather than tweak existing tasks or reengineer workflows, for example, ADP developed its own “agent-building platform” that does both. Individual payroll tasks can be optimized for efficiency while the platform accelerates the rollout of new capabilities such as implementing changes to tax rules across hundreds of locations. - Redefine roles, not just skills. Expect flatter organizations with fewer layers and more mixed human-AI teams. Some people will act as orchestrators, comfortable working across different parts of organizations and supervising human-AI collaboration.
- Clarify decision making. Set clear rules and safeguards for AI use, but within those, give teams flexibility to decide how much freedom to give their systems depending on the task.
- Keep everyone learning. Train people not just to use AI but to question and guide it responsibly. Keep updating the AI systems themselves so they stay accurate and useful.
Just as HR teams recruit, onboard, train, evaluate, and eventually retire employees, organizations adopting agentic AI need parallel support functions for their agents. Firms such as Capital One and Chevron, for example, constantly monitor updated agentic models to ensure that agents don’t drift off course. - Invest for the long haul. Treat agentic AI like an evolving asset—something to nurture and improve over time rather than a one-time purchase.
The Leadership Challenge
Traditional AI raises the question of whether machines are replacing humans or augmenting them. Agentic AI raises a more profound one. How do leaders manage these artificial colleagues like equipment but supervise and develop them like employees?
Managing AI now requires a cross-functional partnership. Success will depend less on mastering the tools than on rethinking how humans interact with them.
Until next time,
 |
| Christoph Schweizer Chief Executive Officer |
