What Real AI Leadership Looks Like

New Roles for Managers in the Age of AI

Speaking of these managers, in The Financial Times, Andrew Hill ​points out​ how managerial roles should already be shifting.

To get the most out of AI, he says, we need to assume new roles such as a “possibility catalyzer,” “uncertainty mapper,” and even a “ideas evaluator”:

• Possibility catalyser. We have to educate our workforce en masse about what is possible with AI. Being a manager is now about being a catalyst and championing what could be done differently with the new technology.
• Organisational designer. Managers must determine what parts of any task can be automated, what can be augmented, and what the new processes look like. They need to redesign the work to ensure staff “do a lot more of the ‘why’ and the ‘what’, and let the machine do a lot more of the ‘how’.
• Ideas evaluator. If the output of the AI-augmented worker is less likely to be lines of code or quantity of PowerPoint slides, the manager’s role becomes more one of peer review or quality assurance. This also means re-examining how we’ll compensate someone who did a 7-hour job in 2 hours with AI.

He speaks of working in a “business hospice,” the final destination for corporate operations at risk of being wiped out by artificial intelligence.

It’s a great metaphor, because it forces leaders to acknowledge the emotional and structural toll of not adapting and the investments they need to make in ​AI change management​.

But how many people in your organization are actually adopting these roles? (If you’re a team of one – how about you?)

How many are fixated on tools and platforms, rather than the necessary rethink of the org chart and the roles within them in the first place?

Leadership in the Age of Agentic AI

Especially when we have fully agentic AI (see my ​guide for launching your first agent here​), should they have KPIs? Do we need AI managers? What happens when my agents are negotiating with another company’s agents?

Some, like Lattice’s CEO Sarah Franklin, are ​already testing​ what it means to “hire” AI into org structures. The experiment may have been premature, but the instincts were right (even though some said it was a publicity stunt).

As AI agents become collaborators, managers will need to develop new forms of governance, new definitions of accountability, and new playbooks for collaboration across human-machine lines.

Goldman Sachs CIO Marco Argenti even added that we need to ​teach these AIs our company culture​:“Businesses must figure out how to "inject" their cultural traits and leadership principles into AI agents, just as they do with human employees.”

He warned that “without cultural training, these agents risk missing the nuances that define an organization's identity.”

How to Lead Human + AI Teams

Right now, the short-term cost of inaction may seem low. AI adoption is still early. Most teams are still experimenting.

But in 6–12 months, when fully agentic systems go mainstream and “​SuperWorkers​”—those who’ve already built custom GPTs, refined their prompts, and reorganized their workflows—start moving faster and thinking bigger, the knowledge and application gap will be painful to close.

Those who’ve done the hard work of breaking down their jobs into atomic tasks and exploring where AI fits will be ready. Those who haven’t will be left behind, not just in output—but in capability, confidence, and credibility.

I deeply believe that in 12 months, you’ll wish you had started today.

Not because the tools will be gone—but because the learning curve will be steeper, the playing field more advanced, and the edge will belong to those who’ve already built their muscle.

So as I’ve said before: start small, but start now.

1. Reframe Roles—Starting with Your Own

Stop thinking of AI as a tool, and start thinking of it as a teammate:

• Audit your current role and your team’s roles.
• Identify where AI can augment or automate.
• Redesign job descriptions to reflect AI collaboration (e.g., “Works with AI tools to produce insight reports.”)

2. Assign KPIs to Your Human + AI Team

If AI is part of the team, it should be managed like one. For each AI-augmented role:

• Define clear goals (e.g., faster turnaround, improved quality, reduced error rates).
• Track output and outcomes, not just usage.
• Benchmark against human performance—what would a junior or senior analyst be expected to deliver?

Example KPI: “AI research agent generates 3 strategic insights per week, with 80% rated high-value by the strategy team.”

3. Create Incentives for Smart AI Use

Reward outcomes, not effort. Defeat secret AI usage and encourage smart delegation—to humans and machines:

• Recognize employees who use AI to improve output quality, creativity, or efficiency. Make it a part of weekly reviews.
• Consider internal awards, shoutouts in all-hands, or even bonus structures tied to AI-powered innovation.
• Make AI usage a line item in performance reviews: “How have you integrated AI into your work this quarter?”

Whether you’re ready or not, you’re going to be managing a new kind of team.

I’d start doing this work today to lead the team and company of the future.