AI owned the air at UNLEASH Paris 2025. “GenAI.” “Copilot.” “Assistant.” And the ever-present: “Buy or build?” You couldn’t queue for coffee without hearing at least one story of a pilot gone right – or wrong.
But this year felt different. The success stories sounded more like lessons learned. The big theme wasn’t capability; it was choice. Yes, the well-worn line about AI pilots failing to take off still held true. But the question everyone’s now asking is, what happens when they do? How do you move from pilot to practice? How do you fold AI into the day-to-day, keep people engaged, and make it part of how the business really runs?
The shift was clear: people and culture were centre stage; tech was the fabric that brings it together. The conversation moved from what AI can do to what people should do with it safely, fairly, and with intent.
Bolting on vs. starting fresh
You could feel the tug between old and new. Some companies are still bolting AI onto existing systems, hoping it’ll fix tired processes. It never does. The stronger examples came from those willing to start again, to ask harder questions: If AI changes how work gets done, what roles still matter? What kind of leadership do we need when a machine starts to share the decision-making? That shift, from add-on to redesign, was the real undercurrent of the few days.
The ‘AI air gap’
Across sessions, one idea kept surfacing: the AI ‘air gap’. Plenty of big companies now have sweeping AI strategies. Fewer can explain what those strategies mean for day-to-day work. And at the same time, there’s a flood of small pilots that aren’t linked to anything bigger.
The ‘air gap’ is filled with the need to get the right governance, support, value metrics and continuous improvement models in place. An adoption playbook that is fit for the organisation and often a version of the enterprise AI strategy that can be translated for HR specifically.
Until top-down ambition meets bottom-up reality, value will stay theoretical.
Mindset over mechanics
Upskilling dominated the agenda. But the most honest speakers said it plainly: it’s not just about tools. It’s about mindset. People need trust and safety before they’ll experiment, fail, and learn alongside AI. That trust starts with leadership; transparency, fairness, consistency. When leaders act with openness, employees follow as a key part of the change management process. Put simply: AI maturity is about behaviour, not technology.
Automation → Augmentation → Human
The rhythm of progress feels familiar now: automate what you can, augment what you should, and leave the rest to people. The real challenge is not building the tool but knowing where to stop – when the task needs human judgment, empathy, or creativity. Those who treat AI as a co-worker, not a controller, will go further and faster.
Buy or build… or both?
The big question this year wasn’t whether to use AI, but how to get it. Do you buy into a single platform that runs across your organisation, or build smaller tools that solve specific problems?
Every vendor had a view, most arguing their platform should be the hub for all your AI agents. And there’s truth in that; capability is improving fast. But most businesses are messy ecosystems. The reality will be a mix:
- Buy where scale and control matter most.
- Build where flexibility and speed matter more.
- Get your data in shape first without it, none of it works.
By 2026, deciding how these layers connect, and what data fuels them, will be one of the biggest challenges companies face.
A market expanding and contracting at once
The expo floor was buzzing. Start-ups, scale-ups, and the enterprise giants all showing their latest. The landscape is growing and shrinking at the same time. Small AI-first vendors keep popping up, built around niche problems. Meanwhile, the established players are still retrofitting intelligence into old architectures. Neither side will win outright. The future lies in mixing the two, AI-native tools where they shine, and upgraded enterprise platforms where trust, governance, and scale matter.
The humanoid moment
And then there was Ameca. A humanoid robot on the main stage, holding a fireside chat with David Green. It was fascinating; equal parts impressive and uncanny. Ameca’s responses were quick, mostly coherent, occasionally interrupting or missing the point entirely. Which, if we’re honest, isn’t so different from a few human colleagues we’ve all known.
Still, seeing two humanoids discuss the future of work – one organic, one engineered – was a jolt. A reminder that what once felt like science fiction now stands at the podium, shaking hands and cracking jokes. The future of work isn’t coming. It’s sitting under the stage lights, waiting for a coffee break.
What it all means
The excitement around AI has matured.
Businesses are done chasing the next shiny demo; now comes the hard work of making it deliver. And that won’t happen through another round of pilots or another layer of automation. It’ll happen through people; how they think, trust, learn, lead and figure out how to design and harness AI.
So yes, AI dominated UNLEASH. But the loudest insight wasn’t about algorithms. It was this: AI doesn’t transform organisations, people do. AI is the plumbing and culture is the operating system.
Our job now is to design the house around both.
Interested in hearing more about sustainably embedding AI into your business? Author Martin heads up our AI team, so reach out for a chat.
