HiQ: The report is clear: almost all organizations use AI, but very few manage to scale it. Why do so many get stuck at the start?
Sofie Perslow:
“This is a classic trap. Companies implement AI solutions, but they don’t change anything around the technology. According to the report, two-thirds of organizations are still in the experiment or pilot phase.
The problem is that even if the technology is in place, processes, ways of working, responsibilities, and data flows look exactly the same as before. In that scenario, AI becomes an ‘add-on tool’ rather than an integrated part of the business. I often compare it to installing state-of-the-art solar panels on your roof but still heating your house with an oil furnace because no one bothered to redo the piping system.
This is also why McKinsey sees so few companies achieving EBIT impact at scale. Only 39% report any impact at all – and for the majority, that impact is under 5%. Without structural change, the real gains never materialize. Another challenge is measurement itself: for example, how should companies quantify a general productivity increase and attribute it to AI without having to break down and time individual work tasks?”
Agentic AI has arrived – but organizations aren’t ready
HiQ: One of the report’s major themes is “Agentic AI” (AI agents). What is your take on this development?
Sofie:
“Agentic AI is still new territory for most organizations, even though the curiosity is enormous. 62% are experimenting or running pilots, but only 23% are scaling agent solutions – and usually within a single function.
This tells us two things: First, the technology is mature enough to create value. Second, organizations are not mature enough to absorb it. Those who succeed typically apply agents in areas like support functions, knowledge management, and software development. What these areas have in common is well-structured processes and data – which is a prerequisite for success. To work with agentic automation, you must understand your processes from the outset, which becomes a much larger task for complex process landscapes.”
The difference comes down to ambition
HiQ: One of the most striking numbers is that only 6% qualify as “AI high performers.” What are these companies doing differently?
Sofie:
“McKinsey is very clear: it comes down to ambition and leadership. High performers are far more likely to use AI for transformative change rather than just chasing cost savings (3.6x more common).
They are almost three times as likely to redesign their workflows from scratch, and senior leaders are significantly more involved in driving and owning AI initiatives. They are also willing to invest – they are almost five times more likely to allocate over 20% of their digital budget to AI.
One of the most interesting paradoxes is risk. High performers actually report more risk incidents – errors, IP issues, compliance challenges. Why? Because they do more, test more, and apply AI in business-critical processes. The lesson is that to be at the forefront, you must accept calculated risks and build an organization capable of managing and learning from them.”
It’s the structure, not the technology, that slows progress
HiQ: Looking at executive leadership – what is required to avoid getting stuck in “pilot purgatory”?
Sofie:
“The report highlights McKinsey’s ‘Rewired model,’ where success factors span everything from strategy and data to talent and operating model. I usually boil it down to three questions for leadership:
- Have you defined how AI will create value?
Most have not. Those who succeed have both a clear vision and a concrete roadmap directly connected to business goals.
- Have you redesigned workflows – or just layered AI on top?
This is the hard part. But workflow redesign is one of the strongest drivers of real business impact.
- Is leadership genuinely engaged?
Leaders must act as both sponsors and users. You cannot delegate an AI transformation to a team far down the organization.
Only when these three elements are in place can the technology do its job.”
Competence needs aren’t disappearing – they’re shifting
HiQ: There’s a lot of discussion about how jobs will be affected. What does the report say about the labor market?
Sofie:
“We see two forces at play simultaneously. On the one hand, 32% expect a reduced workforce. On the other hand, 13% expect an increase. This clearly shows that AI doesn’t automatically ‘remove jobs’ – it reshapes them.
Looking at recruitment, the numbers speak for themselves. Larger companies are aggressively hiring for new roles: Data Engineers, ML Engineers, AI Product Owners, and AI Architects. Competence needs aren’t disappearing – they’re fundamentally changing.”
Three recommendations for decision-makers entering 2025
HiQ: To summarize, what are your three concrete recommendations for leaders reading this?
Sofie:
- Make AI a strategy, not a tech project.
The winners aim for transformation, not just efficiency. That requires direction, budget, and executive ownership.
- Redesign workflows – that’s where value is created.
Right now, this is the biggest bottleneck. The technology exists, but processes haven’t caught up.
- Invest in infrastructure and people.
Data quality, MLOps, and new competencies are the foundation. Organizations already doing this are far ahead of their competitors.
So where are we headed?
HiQ: Var landar vi då, när vi blickar framåt?
HiQ: When we look ahead, where do we land?
Sofie:
“We’re entering a shift where the wheat is separated from the chaff. AI is no longer an innovation – it’s a production factor. The big difference in 2025 will be between companies that merely use AI and those that change with AI.
McKinsey shows that all arrows point in the same direction: high performers are pulling ahead – fast. We see the same pattern among our customers. Those who dare to think big move forward quickly. Those who wait don’t just lose momentum – they risk losing talent.
This year isn’t about experimenting more. It’s about scaling.”
