You can buy AI. You cannot buy a team that makes it work.

I recently opened a masterclass I delivered on building great teams in the age of AI with that sentence, and it actually summarizes the story quite well.

McKinsey said something similar this year: oversight capacity limits agentic scale. Translated: how far AI can take your organisation is not determined by the model. It is determined by how well your team adapts to it.

Beneath that statement lie four shifts. They all arrive at the same team, within the same timeframe.

The four shifts
The first is the operating model itself, the way in which a company creates value, now with AI as a new source of that value. Most companies still operate on an industrial-age operating model with AI bolted onto it, and that delivers speed, not value. Only a small minority have redesigned themselves around agile, cross-functional, product-oriented ways of working. Rewiring comes before automation, not after.

The second is system architecture. Traditional software is deterministic; AI is not. This means that validation and security can no longer sit at the end of a process as a final check. They must be designed in from the start: prepare, propose, evaluate, execute, with a human where necessary.

The third is how work gets done. People are shifting from building to specifying, supervising and owning the outcome. Employees become owners, defining what success looks like and where the boundaries lie, and verifiers, catching what AI gets wrong before it spreads further.

The fourth is how teams are organised. Two to five people can already oversee fifty to one hundred specialised agents executing an end-to-end process. Teams become smaller, cycles become shorter, and the skills within a team shift. Accountability, however, remains firmly human. When an airline once argued in court that its chatbot was effectively a separate party, the court still held the company responsible. Whatever an agent does in your name, it is still your name attached to it.

Add these four together and you arrive at one conclusion. The limitation on how far AI can go within your organisation does not lie in the model. Your team is your AI ceiling.

Where the Cherry Model helps
I like to think of a team as a cherry. The pit represents the hard skills, the technical expertise for which a team is hired. The flesh around it represents the soft skills, how people actually work together. A team is effective when both are in balance, never on the basis of the pit alone, and we explored that perspective in greater depth several years ago together with Tim Meeuwissen in our book The Cherry Model.

What is new is that AI is moving into the pit. It is steadily taking over more of the hard skills, and that will continue. That is precisely why the flesh, the soft skills, becomes the differentiating factor. Not because it is a nice addition, but because it now determines the ceiling.

That same model also provides a way to structure the four shifts, through its four seasons of change. Winter is for survival and naturally aligns with reconsidering the operating model. Spring is for shaping the team, where you re-examine how you organise. Summer is for growth, the season in which to address system architecture. Autumn is for harvesting, for reflecting on how work is actually done. All four shifts are taking place, but where you begin depends on the season in which your own team currently finds itself.

In other words: no big bangs. A gradual introduction, AI treated as part of the team rather than bolted onto it, and a culture sufficiently free of blame to allow problems to surface and sufficiently curious to enable people to keep learning from them. Building such a culture still leads to results, with or without AI.

Will I see you next time? For more events and for the unique NextGen Tech Executive Programme, please visit the NextGen Tech Leaders website.

Kind regards,

Anneke Keller
Group CTO QLS
Co-initiator and Member of the NextGen Tech Leaders Programme Board