04/12/2025

We are entering a strange phase in the AI market – a phase I would describe as AI bluster. It’s the overconfident, overpolished, and often misleading promise that a shiny new “AI platform” can replace strategic and critical thinking, automate strategic decision-making, or run an organisation with little to no human leadership.

This is not where AI is today, but some sellers desperately want you to believe it is.

The problem begins when agencies position their tools as “CEO in a box,” “fully autonomous managers”, “AI that runs your business”, or “AI that will replace the whole department”. These claims aren’t just exaggerated; they distort understanding, inflate expectations, and put organisations at risk of making high-stakes decisions based on marketing hype rather than capability.

These unicorns are easy to spot:
– Overpromised “Agentic AI” that claims to think, prioritise and act strategically.
– Tools sold as replacements for specialists and managers, not as support systems.
– Platforms that advertise autonomy but still require constant human correction behind the scenes.

Case studies built on prototypes, not real deployments.
Why is this happening?
Because the market rewards confidence, and some AI agencies are chasing short-term wins, hype sells. “Autonomous leaders” sound exciting. And many buyers don’t yet have the technical understanding to separate what’s possible from what’s fantasy.

So what’s the real status of AI Agents today?

Current AI agents lack contextual awareness, organisational judgement, ethical reasoning, cultural intelligence, and the accountability that leadership requires.

A practical solution is to treat AI as a decision-support partner, not a managerial replacement, and design workflows that keep humans firmly in the loop. Organisations should demand evidence, not hype, requesting real case studies, transparent limitations, and live demonstrations before adopting any “agentic” system.

And finally, leaders need AI literacy training to understand what today’s systems can actually do – so they can challenge overclaims and deploy AI responsibly.