Davos 2026: Leaders on why scaling AI still feels hard - and what to do about it

“When you are going to adopt new workers into your workforce, you need to rethink how the team is going to play together to do the same tasks.”

These words from Roy Jakobs, President and Chief Executive Officer, Royal Philips, during the World Economic Forum’s Annual Meeting 2026, weren’t part of a discussion about how to welcome new people to a team at work – Jakobs was explaining how to integrate artificial intelligence (AI) into a business to optimise the impact on operations, employees and the entire company.

During the Scaling AI: Now Comes the Hard Part panel, Jakobs talked about the need to reimagine current work processes to scale AI. He sees this as AI's biggest challenge because it means redefining how we work.

But many companies are still struggling to work out how to plot a route to reach this point. While $1.5 trillion was invested in AI last year, according to Gartner, a McKinsey global survey of almost 2,000 companies found that nearly two-thirds have not yet scaled their AI projects across the enterprise. Corporations want to accelerate AI use to capture the full economic impact of this technology – but how?

Scaling AI, and overcoming the organizational challenges that come with it, surfaced across numerous high-profile CEO sessions at Davos 2026. Companies at the forefront of AI adoption discussed how they’re making bigger bets and creating pathways to scale AI beyond early successes. According to these frontier AI users, successfully moving beyond AI pilots means creating new strategies, capabilities and organizational designs. This is how they have addressed the significant challenges of scaling AI.

Unsurprisingly, many firms’ AI adoption journey starts with technology and data. “AI has been a catalyst for companies to really look at their technology,” said Julie Sweet, Chair and Chief Executive Officer of Accenture, with many companies investing in data too. “The companies that have done this early, like Saudi Aramco – or McDonald's, which created its data foundation very early – are surging ahead in how they're using it.”

Members of the World Economic Forum’s AI Transformation of Industries community have also been using AI to change how they benefit from institutional knowledge – the expertise and judgments of frontline workers and experts. This kind of information is often left untouched as unstructured data that’s not fully integrated into a company’s processes, losing valuable insights.

For example, equipment manufacturer, Allied Systems, uses real-time optimization and embeds operator knowledge to enable high volume lines to reach significant targets. With AI, processes that were once dependent on an employee’s intuition have become repeatable and teachable.

Other members are using AI to uncover new signals that go beyond what traditional analytics can detect. S&P Global, analyzed 192,000 earnings calls to identify patterns that correlate unstructured communication data into forward-looking financial insights. And Claryo supports continuous enterprise learning through a “glocal” model that adapts to each of its sites’ unique operational DNA while drawing on structured intelligence from across the company’s network.

Data and technology are important, of course, but a good rule of thumb is to spend at least as much time thinking about adoption as tech development, Jakobs said during Davos. In other words, considering how AI will be used in practice by people throughout the organization. “Adoption is ultimately where success is measured,” he said. “You need to design that in from the get-go. And that is much less about technology, much more about understanding the practice that it will actually serve.”

It's important to think deeply about how AI is going to change the workflows of the organization and also the nature of its work, Meta's Chief Global Affairs Officer, Joel Kaplan, said during another session. Every organization is going to have to think about this over the next couple of years,” he added. “And the ones that [will] succeed are the ones who start thinking now.”

Focusing on AI uses and users will also create feedback and show the tangible benefits of AI. For example, JLL Technologies redesigned its product development lifecycle by automating requirements for gathering, code generation and testing. It says this reduced its development cycles by 85% and its resource needs by 30%, allowing its senior engineers to focus on more complex tasks.

Google developers use AI as a coding partner to generate 30% of new code and to support code reviews, testing and migrations. This has resulted in an estimated 10% increase in engineering velocity, according to the company, freeing its engineers to work on other priorities.

SandboxAQ’s AI Co-Researcher uses a hierarchical multi-agent system to automate complex scientific workflows that were previously handled by experts. These human scientists can now focus on strategy and oversight, reducing project completion time by 50% and doubling project capacity, the company says. Nestlé Purina has also redesigned its plant operations, using Boston Dynamics’ Spot robots to automate routine inspections. It says it achieved full ROI on this project within a year across 23 facilities.

It’s not just companies, some governments have been scaling up AI use too. During a panel discussion on AI regulation, United Arab Emirates Minister of State, Maryam bint Ahmed Al Hammadi, explained how her government uses AI to develop regulations. But safeguards for AI use is crucial, she warned. For the UAE, guardrails prevent bias, ensure AI outcomes can be traced directly to laws and protect data privacy, quality and consistency.

Transparency and human accountability are also important principles. “We believe that, at this stage, AI can advise, but a human is still in command,” she said.

Self-regulation is important, especially since regulators also struggle to keep up with the speed of technology development. “We need to have our own rules – how we test, how we validate, what level of rigour we apply, what kind of practices, but also biases that we take into account,” Jakobs said.

Keeping humans in the loop in this way will also help to foster more trust in AI in general. In turn, that may encourage people to try to understand and use this technology even more. If people and AI work together, each can play to their own specific strengths. Humans can provide judgment, oversight, creativity and spot exceptions, while AI can handle execution, optimization, prediction and pattern synthesis. This shift is already starting in some organizations.

Microsoft Research, our community member, uses embodied AI systems that learn from interaction with physical environments, adapt to dynamic conditions and translate perception into action for everyday tasks. Indian IT firm HCL equips its service agents with real-time insights and sentiment aware guidance, allowing its people to focus on nuanced issues that require empathy and complex reasoning.

Advertising company WPP uses an AI-enabled operating system to combine creative and media workflows, increasing creative capacity and productivity. And investment firm Vista Equity Partners applies AI to instantly resolve a large share of its customer inquiries, reducing operational pressure while maintaining high levels of satisfaction.

Scaling AI successfully means redefining work, but for those at the forefront of corporate tech development it also means keeping a human touch – “human in the lead, not human in the loop,” said Accenture's Sweet at Davos 2026. “We will inspire people and we will run companies with people, and they will have a greater technology landscape," she added. "But we need to completely change the narrative to inspire people to paint the future.

With contributions from the AI Transformation of Industries worksteam. Additional input by Natalie Marchant.