The AI wave is crashing, now comes the real work

After three years of hype, businesses are expecting AI to start returning on their investment – and the resulting shortcomings have prompted talk of a ‘bubble’ to reach fever pitch. Rhys Merrett, senior vice president with The PHA Group, argues that to make the most of the technology, firms need to start focusing on the practical ways it is making an impact, which might be less “sexy”, but is “more realistic”.

If 2025 began riding the crest of the AI wave, the year is ending with that momentum crashing hard against the rocks of reality. This might seem like a cynical take, however, the entire narrative around AI has become repetitive.

News outlets, social media, and industry blogs all echo a similar message, that businesses must adapt to AI or risk falling behind. While the specifics vary, the narrative is consistent, because it resonates with audiences and signals relevance in a world changing faster than most companies can keep up with.

Therein lies the problem: many voices want to be seen as stakeholders in AI’s future, aligning themselves with a technology that promises permanent change. However, too often, these attachments offer little insight beyond signaling credibility. The conversation can be noisy, with real expertise buried under repetition and hype, reflecting more about the desire to be “in the game” than about the substance of AI itself.

Substantial messaging

The message is relentless, creating AI fatigue. Businesses are either leaping headlong into AI adoption, or pausing, wary of committing to a technology that is still evolving. That’s why the narrative needs to change. The focus needs to be the practical and tangible ways that AI is transforming businesses and organisations. Perhaps not as sexy, but more realistic. And one way this is happening right now is in the breakdown of work silos.

For decades, companies have been structured around functional divisions, PR, marketing, sales, recruitment, growth, each with its own data, objectives, and internal logic. These silos have made conventional sense in a world of fragmented information and linear decision-making. But AI does not operate in compartments. It thrives on integration, pattern recognition, and holistic analysis.

Large Language Models like ChatGPT are a clear indication of this - suddenly, professionals can quickly create content and research documents after entering a few quick prompts, instead of relying on a separate research function to deliver.

As AI systems ingest and interpret data across departments, they do more than just streamline operations, they expose inefficiencies, redundancies, and missed opportunities that were previously hidden in siloed workflows. According to McKinsey, 78% of companies now use AI in at least one business function, up from 55% just a year earlier. This widespread adoption is revealing how disconnected strategies across recruitment, PR, sales, and growth can be harmonised through shared intelligence.

The implications for organisational structure are profound. Roles are converging. Teams are becoming cross-functional by necessity, not design. A Deloitte study found that companies with strong collaborative cultures are five times more likely to be high-performing. The traditional org chart which was once a tidy hierarchy of boxes and lines, is changing. I would expect job titles and job descriptions to also change as part of this.

Challenges ahead

This evolution is not without friction. It challenges legacy mindsets, entrenched power dynamics, and outdated metrics of success. But it also presents a rare opportunity, a chance to build organisations that are not just AI-enabled, but AI-aligned. Where insight flows freely, decisions are made collectively, and value is created across boundaries.

It brings me back to my earlier point on the growing sense of AI fatigue. We’ve spent much of the year obsessing over the technologies themselves, what they can do, how fast they’re evolving, and the sheer scale of their capabilities. But in doing so, we’ve overlooked a critical truth. That is that the success of AI will not be defined by the sophistication of the models, but by the quality of human input that guides them.

All the excitement around models, capabilities, and lightning-fast innovation has distracted us from the real story: AI’s success hinges on people, not code.  Organisations must move beyond a tech-first mindset and begin reimagining the role of human professionals in this new landscape. In this paradigm, leadership needs to evolve from command-and-control to orchestration. It’s no longer about directing tasks, but about enabling teams to work with intelligent systems.

Success will depend on how well companies adapt their structures, empower their people, and embed AI into the very fabric of how they operate. That means investing in new skills, fostering cross-functional collaboration, and designing workflows that are fluid, not fixed. Human insight is what will define the next chapter of AI.

This is the approach we should see leading organisational discussions on AI, and which will give new life into a conversation that runs the risk of falling more into hype.