Is your operations leadership fit for the future?

In our work, we notice that most organisations focus on and optimise use cases within existing workflows (e.g., how can Claude, ChatGPT, Lovable, or *insert your tool here* help with this dashboard or clean up that report). Individually, these use cases often deliver incremental improvements and are valuable, but they rarely move the top line or change the overall cost picture of the company. They remain islands of improvement, too isolated from the next value pool to really add up into anything larger. That is causing many operations leaders to question themselves, the technology, and whether the investment is worth it at all, even though few will say so publicly.

Our observations – and Ethan’s argument – are reinforced by what a set of frontrunners across industries point out: moving the needle requires a different approach compared to what has historically been the case. We argue that this is not a tech question but rather an organisational one.

A recent HBR piece draws on data and interviews from frontier firms, i.e., a select group of companies leading AI adoption and proving value at scale, to explore what is really holding back scale and ROI in this AI age. The authors outline seven key frictions these firms had to overcome to achieve the impact they did – an uplift ranging from 10 to 20 percent of EBITDA, substantially higher than most companies pursuing AI have achieved.

Only one out of the seven frictions is mainly technical, ‘architecture complexity’. The other six ultimately stem from challenges related to people and organisation, ranging from performance systems and governance to process debt and portfolio management. Some of them are also very tightly connected to leadership, where tribal knowledge (the undocumented institutional knowledge that is notoriously difficult to transfer), the efficiency trap, and the productivity gap can all create resistance to change.

In other words, according to those firms, the primary obstacle to progress is rarely model quality, data availability, or pilots, but rather the 'last mile' of transformation, where technical capability must meet organisational design. But what does it take to cross it? The frontier firms already point to some answers, so by combining their findings with the World Economic Forum's work on human-machine collaboration and our own portfolio of projects, we have tried to distil them into seven directions below.

The solutions themselves vary in their degree of novelty – the focus on portfolio management can hardly be considered revolutionary, but it still needs to be executed – and many do not yet have an obvious ‘how’ behind them. To use access and governance as an example: How do you actually govern a digital workforce that is constantly changing? Do you consider agents part of teams, moving them as part of organisational redesign, or do you embed them within workflows, making them more independent of changes to the organisational structure?

These seven shifts change the conversation from technology procurement to a much more nuanced discussion about organisational design and the operating model. They are rarely on the agenda in everyday management meetings and instead require a significant willingness from the organisation to change. They are harder, as well as riskier, to address, but they may also be among the most important ingredients for improving the odds of a successful AI transformation.

The conversation we need to have now

Our prediction is that as early as 2029 the gap between the top and bottom quartiles of operations organisations, measured by EBIT contribution from AI, will be larger than any technology-driven gap we have seen previously. That gap will be largely explained by the factors described in this article. If this comes to be true, the next discussion you need to have is whether your operations organisation is working on the right things. 

To assess that, we would put six questions to any operations leadership team:

Starting the conversation around these questions is also the beginning of AI transformation that will deliver tangible impact, and the quality of the discussion that follows will probably start pointing to the areas you will need to focus on. Coming back to Ethan Mollick’s point, that is how you find out where you are still navigating by the old maps, and how you change course in territory that keeps moving.