There’s a word that’s been dominating AI conversations lately: agentic.
It refers to AI systems that don’t just respond to prompts — they take sequences of actions, use tools, make decisions, and pursue goals over time. They have agency.
This is different from ChatGPT answering a question. This is software that can browse the web, write and run code, send emails, update databases, and do research — all as part of completing a complex task you’ve assigned it.
Why this matters for leaders
Most executives have mentally filed “AI” under “productivity tools.” You use it to summarise documents, generate first drafts, and answer quick questions. Useful, but incremental.
Agentic AI is a different category. The implications are structural, not marginal.
When an AI agent can autonomously handle a process end-to-end — from trigger to outcome — you’re not talking about making individuals 20% more productive. You’re talking about rethinking what work your organisation actually needs humans for.
The delegation question
Good leaders know how to delegate. They identify what needs to happen, choose someone capable, give them context and authority, and check in at the right cadence.
Delegating to an AI agent is similar — but the skills transfer imperfectly. AI agents need more explicit instructions. They don’t ask clarifying questions the way a human would (though they’re getting better). They can fail in strange and unexpected ways.
But when you get it right, the leverage is extraordinary.
What hasn’t changed
The fundamentals of leadership haven’t been disrupted. Someone still needs to:
- Set the direction
- Define what “good” looks like
- Build and maintain the culture
- Make the calls that require judgment and ethics
- Represent the organisation to the world
What’s changing is everything beneath that level. The execution layer is increasingly agentic. The leadership layer is increasingly about designing systems that work.