Post by Nadine Soyez

 

End October 2025, I attended the World AI Summit in Amsterdam, a truly global gathering of more than 5,000 people from around the world exploring what’s next in the AI-driven business era. For two intense days, the event felt like a snapshot of the future: executives from every industry sharing lessons, best practices, future perspectives on what will come next, and conversations that cut across strategy, ethics, and transformation.

As I did for past AI conferences, I sum up my key insights in a series of LinkedIn newsletters. I also did additional research for the study and numbers. What made this year’s summit stand out was the maturity of the discussion. The question is no longer:

“Should we use AI?”

but rather

“How do we integrate AI into the core of how we operate, decide, and grow?”

We’ve moved beyond curiosity and hype into a phase where organisations are actively building the structures, policies, and cultures that make AI work at scale. The atmosphere was one of urgency mixed with optimism, a sense that we’re all part of a massive learning curve, redefining what work and leadership will look like in the years ahead. If one theme captured the essence of the day, it was this:

Agentic AI, data, and mindset are no longer separate topics. They form the foundation of the intelligent enterprise. We are entering a new phase where AI is not just a tool or an assistant, but an intelligence infrastructure that connects people, data, and decisions.

 

 

Agentic AI – Transparency, Openness & Connection

 

Agentic AI represents a fundamental shift from task automation to goal-driven intelligence. These AI systems can act autonomously within defined parameters, collaborate with humans, and execute multi-step processes, making them the next evolution beyond chat-based assistants. At the summit, the message was clear: transparency, openness, and connection must be the foundation of this new architecture. Without transparency, trust collapses; without openness, innovation slows; and without connection across systems, teams, and data sources. Agents simply cannot function.

A striking comparison was made:

“Imagine a world where companies refuse to use agentic AI. It will soon be as unimaginable as businesses without email.”

However, to enable this, organisations need robust data foundations, well-documented processes, and interconnected knowledge systems. Agentic AI cannot create value in isolation. It depends on structured information and clear governance.

According to Capgemini’s 2025 study on AI agents:

• Only 2% of organisations have deployed AI agents at scale.
• 12% operate at partial scale, 23% run pilots, while 61% are still exploring.
• Trust in fully autonomous agents has dropped sharply from 43% to 27% in one year — reflecting concerns about transparency and control.

These numbers show a market full of intent but short on readiness. You can find the study here.

 

What companies should do now:

• Start with small, transparent agent pilots that show value and build trust.
• Map out your current processes and data flows, agents depend on clarity.
• Involve compliance, legal, and data protection teams early to design governance frameworks.
• Communicate openly with teams about what AI agents can and cannot do.

Transparency is not a “nice to have”. It’s the operating principle of agentic organisations.

 

 

Data Gets Closer to the Business

 

“Data is the new differentiator” is no longer a cliché. It is an operational necessity. At the summit, the consensus was clear: data must move closer to the business. It’s not enough for data to sit in isolated silos managed by IT. Business units need direct access to reliable, well-prepared data to experiment, simulate, and train AI systems.

This shift from IT-owned to business-embedded data marks one of the most important cultural changes in AI adoption. Companies that empower their teams to use and interpret data in context will move exponentially faster than those waiting for central analytics teams to deliver insights. Capgemini’s research reinforces this urgency: while 15% of business processes are expected to reach semi- or full autonomy within 12 months, fewer than one in five organisations report high maturity in their data and technology infrastructure.

The bottleneck is not AI, it’s data readiness. Without clean, well-structured, and accessible data, even the best AI agents will underperform.

 

What companies should do now:

• Conduct a data readiness audit, assess where your critical business data lives and how accessible it is.
• Establish data stewardship roles within business units, not only in IT.
• Standardise how knowledge and documentation are stored (policies, procedures, project data).
• Prioritise integration platforms that connect structured and unstructured data sources.

The goal is not just to collect data, it’s to activate it for intelligent collaboration.

 

 

Mindset – Start with the Goal in Mind

 

Technology follows purpose. The organisations that succeed with AI are those that start from the business outcome and work backwards. Every AI project should begin with a simple but powerful question: What business result are we trying to achieve?

This “goal-first” mindset prevents technology from becoming an isolated experiment. It forces alignment between leadership, business needs, and technical design. Strong compliance, data safeguards, and ethical guardrails do not limit innovation — they enable it. Clear frameworks reduce uncertainty, allowing teams to explore AI use cases confidently and responsibly.

The pace of change is accelerating. The call to action was direct: “Lean in or be left behind.”

 

What companies should do now:

• Anchor AI initiatives in measurable business goals (e.g., customer satisfaction, efficiency, risk reduction).
• Create cross-functional AI taskforces with clear accountability for results.
• Foster a learning culture — allow teams to experiment, fail fast, and share lessons.
• Regularly assess and adjust your AI roadmap to keep it aligned with evolving goals.

The learning curve starts now, but so does the opportunity curve.

 

 

Trustworthy AI – A Living Practice

 

Trust is the true currency of AI adoption. Trustworthy AI isn’t about perfect algorithms; it’s about continuous ethical alignment between humans and machines. Different panels emphasised the need for living policy documents, frameworks that evolve as technology evolves. They serve as dynamic guides on how AI is used, monitored, and improved across the organisation.

The MIT Initiative on the Digital Economy captured this challenge perfectly:

“Our understanding of how to work with AI agents to maximise productivity and performance — as well as the societal implications — is still nascent, if not nonexistent.” Here you find the MIT source.

This means organisations are designing their own playbooks as they go. Trust will not come from regulations alone — it will come from leadership, transparency, and values-driven culture.

 

What companies should do now:

• Define AI ethics principles grounded in your company’s core values.
• Write an AI usage policy that is reviewed and updated quarterly.
• Establish AI governance committees with representatives from HR, IT, legal, and business units.
• Train employees on responsible AI use — from data handling to human-AI collaboration.

Trust is not static; it’s a practice that must be maintained daily.

 

 

Two Clocks, One Urgency

 

One of the most striking metaphors of the day came from a panel on transformation: “AI has two clocks.”

• Clock 1: the AI clock is running exponentially fast, driven by innovation cycles measured in weeks, not years.
• Clock 2: the organisational clock moves slowly, constrained by structure, skills, and culture.

To thrive, leaders must learn to accelerate Clock 2. This means aligning organisational design, decision-making, and governance speed with the velocity of AI progress. AI is no longer a side initiative, it’s becoming the intelligence backbone of customer and workforce experiences alike.

 

What companies should do now:

• Assess where your “Clock 2” is slow: decision-making, procurement, compliance, or skills.
• Introduce agile governance — faster review and experimentation cycles.
• Invest in leadership training on AI fluency and decision-making with AI insights.
• Align transformation timelines with technological cycles — don’t plan AI as a 3-year project.

The faster you align both clocks, the faster your organisation will move from exploration to execution.

 

 

Humans as Orchestrators

 

Even as AI agents take on more work, humans remain at the centre — as orchestrators, interpreters, and sense-makers. We are moving towards hybrid teams, where human intelligence and machine intelligence complement one another. By 2028, 38% of organisations are expected to have AI agents embedded as team members, according to Capgemini (see study link above). These blended teams will drive productivity and innovation through continuous collaboration and decision support.

The human role will shift from “doing” to designing and directing intelligent systems — a profound change in what it means to lead, collaborate, and create value.

 

What companies should do now:

• Redefine roles: identify where human judgment adds most value.
• Equip employees with AI collaboration skills, validation, and decision support.
• Design hybrid workflows where humans review, guide, and improve agent output.
• Foster a culture where people see AI as a partner, not a threat.

The most successful organisations will be those that master the art of human–AI orchestration.

 

 

Conclusion: A New Era of Intelligence Infrastructure

 

Day 1 at the World AI Summit made one thing undeniable: We are standing at the intersection of Back to the Future and Business Reality — where Agentic AI rewrites the rules of time, talent, and transformation. The opportunity is enormous. Capgemini estimates that by 2028, AI agents could generate $450 billion in economic value through revenue growth and cost savings across industries. Yet as MIT reminds us, our understanding of how to work with these systems is still at an early stage.

If you would like to learn more about what this might mean for your company, just reach out for a conversation.