At Valtech’s recent Intelligent Enterprise: Live event in London, leaders from Google Cloud, UBS, AXA, and easyJet all shared a blunt perspective: AI is creating a level competitive field across industries. An organisation’s proprietary assets — its data, its context, its people — will create the leverage needed to build an advantage.
So, the question becomes: What must change inside an enterprise to turn widely available AI into durable, defensible value?
What we discussed at The Intelligent Enterprise: Live in London
Valtech and our guests hosted several sessions during the evening. Across every session, four themes emerged:
- Efficiency with AI is table stakes. Real differentiation comes from capabilities a competitor can't simply buy.
- Discovery is fragmenting. Customers now find brands through AI assistants, video and social, not just search.
- Context only works if it's earned and explainable. Data has to come with permission and a clear reason for use.
- Governance has to be executable, not just written down. And none of it lands without people who trust the tools enough to use, and question, them.
Building AI advantage beyond efficiency
Alex Rutter of Google Cloud opened with a crucial distinction: efficiency AI versus Opportunity AI. Efficiency AI optimises what you already do. Rutter cited productivity gains of 10–20%, but that's becoming commoditised quickly. Opportunity AI builds capabilities that were previously too expensive or too hard to attempt, and that's where the real value sits.
It's also where your brand's own history, relationships and ways of working become an asset rather than overhead. These are the things a competitor can't buy off the shelf.
The test Alex offered leaders: If a competitor bought the same foundation model, would your advantage disappear? If the answer is yes, you're not differentiated. You're renting.
Rutter pointed to four capability unlocks worth chasing: decision simulators for live what-if exploration, playable engagement that replaces passive marketing, codified expertise that makes institutional judgement always available, and instant narrative production at a scale that manual methods can't match.
His five-question scorecard for leadership teams:
- Is AI creating new revenue or only cutting cost?
- Could a competitor replicate this with the same tools?
- Does the system get better from your unique data?
- Are you running enough experiments to learn quickly?
- And is AI part of every P&L owner's rhythm, or is it still boxed inside an ‘AI team’?
Reimagining the workforce: The confidence gap is the real bottleneck
Valtech’s Deborah Womack led the audience debate on the workforce, and the headline number should worry every leadership team in the room: Ipsos data shared during the session found only 21% of workers feel confident using AI at work, and just 28% feel confident using it in daily life. Nearly half sit neutral or unsure.
While the technology gap is closing, the confidence gap is getting wider, and that’s a binding constraint on scale.
The panel's central argument was that every AI decision is also an organisational design decision. Dropping AI into yesterday's workflows won't produce tomorrow's productivity. You have to redesign roles, decision rights, governance and skills alongside the tooling, or AI just becomes another layer of complexity.
There's an upside buried in this, though. The skills that get more valuable in an AI-enabled organisation are the ones we've spent years dismissing as ‘soft’ — communication, empathy, judgement, collaboration.
Automation frees capacity. What you do with that capacity, and whether people are equipped to use it on framing problems and owning outcomes, is what actually determines the payoff.
That freed-up time is exactly where Opportunity AI lives. It's capacity to spend on the human judgement, creativity and relationship-building that efficiency gains alone can't buy. What you do with that capacity, and whether people are equipped to use it on framing problems and owning outcomes, is what actually determines the payoff.
Context as commercial advantage
The Race to Context panel tackled a particularly hard problem: Customer discovery is no longer a predictable funnel.
UBS is watching conventional search metrics decline as activity shifts to LLMs and AI-generated summaries. easyJet described the same erosion. The familiar path from organic search to landing page to conversion is breaking down.
easyJet's response is to stop measuring keyword rankings alone and start tracking its share of a defined ‘prompt universe’ — the destinations, products and needs customers actually ask AI about. It's directional, not exact, but it's providing more honesty about where visibility is actually won or lost now.
Two disciplines matter here:
- Evidence beyond your own site. AI-generated answers draw on media coverage, reviews, third-party references and consistent brand signals, not just your website copy. Search, PR, social and content teams need a shared view of customer need and the evidence you can credibly offer.
- Trust as a hard boundary. Context must be earned and used transparently. Customers should know what's being captured and when AI is involved. AI doesn't create a new right to use every signal you happen to have.
Where this pays off commercially: UBS is exploring agentic workflows that help wealth advisers prepare for client meetings by pulling together calendar, email, CRM and relationship signals — all while preserving the discretion financial services demands.
In travel, the parallel opportunity is meeting customers in new interfaces with live pricing and availability, not just more content.
Data readiness is an ownership problem, not a technology problem
Julian Dicker of AXA and Daniel Cano of easyJet made a point that should reframe how most organisations think about their AI readiness. Most enterprises don't lack technology. They lack clear accountability for the meaning, quality and permitted use of shared data.
Cano's example was the word ‘customer’. Marketing counts one booking as one customer. Airport operations count four travellers on that booking as four. Finance sees one transaction. Ask an AI agent a simple customer question, and you'll get a different answer depending on which dataset it reaches — and the AI can't resolve that ambiguity for you.
The fix is explicit ownership: who owns each term, how it's used, and enough context for people and AI to pick the right meaning for the job at hand.
AXA's approach is to move governance from documents into the platform itself. Computational policy automatically enforces compliance and produces evidence of it, which matters in a regulated environment. And because real-time AI shrinks the data quality window from weeks to seconds, easyJet is deliberately keeping AI in a decision-support role for now rather than handing over autonomous operational control.
Both panellists agreed on where to invest first: the semantic layer. Definitions, lineage, KPI meaning, synonyms, intended use have all moved out of people's heads and into managed systems. That looks like unglamorous foundational work, but it’s what makes every other AI investment reliable.
10 principles that define an Intelligent Enterprise
Valtech’s vision is for organisations to orchestrate their tech — and their data, and their context, and their people — around the experience, which is the biggest force-multiplier for growth. That’s what it takes to become an Intelligent Enterprise.
In practice, this is Efficiency AI and Opportunity AI working together. Efficiency AI unlocks the time and capacity, and Opportunity AI is where that capacity gets reinvested — most visibly in the experience itself. And none of it works without human intelligence. Data and tools only become advantage when people apply judgement, context and creativity on top of them.
Following our conversations in London, 10 principles emerged that reveal what steps must be taken on the path towards becoming an Intelligent Enterprise.
- Build capabilities that are hard to copy. Differentiate through proprietary data and workflows, not the model itself.
- Balance efficiency with opportunity. Fund both deliberately.
- Anchor every use case in a real decision or customer need.
- Treat context as managed enterprise data, not a byproduct.
- Make ownership explicit. Technology can't fix an accountability gap.
- Turn governance into working controls, not just policy documents.
- Design data and content for reuse, with named owners.
- Keep trust visible. Use data with permission, and disclose AI's role.
- Organise teams around outcomes, not silos.
- Experiment quickly, and feed the learning back into the system.
What this means for your next quarter
The organisations that get the most from AI will pair pace with discipline. That means:
- Separating efficiency bets from capability-building bets.
- Running the ‘Could a competitor replicate this?’ test on every initiative.
- Mapping the handful of business terms and decisions that matter most.
- Investing in the confidence and judgement of the people who'll actually use these systems.
Put AI progress into business reviews and P&L conversations, not just tool-adoption dashboards. The technology is available to everyone. What you do with your own data, context and people is what turns it into an advantage.
Could a competitor replicate your AI strategy?
Bring us an AI initiative you’re considering. In a no-obligation workshop, we’ll explore what a competitor could easily copy, where your own data and context could give you an edge, and what to do next.
Session contributors
A massive thank you to everyone who helped make this event possible:
- Alex Rutter, Managing Director, AI Go to Market Europe Africa Middle East and Gulf at Google Cloud
- Liv Marit Brahin, Managing Director, Digital Platforms at UBS
- Ian Chambers, Digital Director at easyJet and easyJet holidays
- Julian Dicker, Head of Data Platforms at AXA
- Daniel Cano, Director of Data and Analytics at easyJet
- John Cunningham, Global CTO Valtech
- Deborah Womack, Executive Director, Strategy & Consulting at Valtech, who led the audience debate on reimagining the workforce
- Tizzy Philp, Global Head of Marketing at Valtech, who moderated the Race to Context panel
- Sophie Bennett, Google Partner Development Manager at Valtech, who moderated the Is Your Data Ready panel
