I spoke at the Financial Times Future of AI summit today, joining global leaders to explore what’s next for industrial AI.

The event brought together some of the field’s most influential voices. Interviewed by Jonathan Eley, Correspondent for Financial Times, I shared a simple message today: trust, speed, and real-world outcomes now separate winners from the rest.

At IFS, we work where complexity and criticality intersect such as aerospace & defense, energy & utilities, heavy manufacturing, construction & engineering, and telecommunications. These industries keep economies running. Getting AI right here isn’t a headline; it’s operational resilience and performance. Today, our customers are deploying what I call operational AI. Operational AI is not experiments or pilots, it’s production-grade intelligence embedded in day-to-day workflows. That shift matters because it is where value finally starts to compound.

At the FT Live summit, I saw how closely this mirrors the direction industry leaders are taking. From our own scaling efforts at IFS to the strategies shared on stage, the message was consistent: AI must deliver real outcomes, fast. Here are five key takeaways from my conversation at FT Live; insights that reflect where industrial AI is heading, and what it takes to move from pilots to production.

1) From “Transform Everything” to “Prove It Fast”

A few years ago, many companies approached AI as a grand transformation project. The intent was admirable, but the result was often slow and unfocused. What we are seeing now is a pivot to precision: targeting specific, measurable problems that deliver payback in months and proof in weeks.

Too many organisations have been stuck in what I call “AI prototype purgatory”. The technology works, but the business case does not land. Our role at IFS is to break that cycle by anchoring everything to P&L impact.

Take inventory optimisation as one example. A decade ago, when I worked in a project at a large regional utility, it took around two and a half years and several disconnected systems to reduce spare-parts stock by 40 per cent – a benefit that reduced over time due to the approach and available technology. Today, with new AI tooling and better data foundations, similar results are being achieved in around six weeks. That is not theory, it is happening now, and it is attracting serious attention, including from government and highly regulated industry circles.

2) Why AI Is Finally Speeding Up

Three factors are making this acceleration possible.

First, data foundations have matured. AI-powered master data tools reduce the painful manual work of cleansing and standardising information. Industrial AI cannot afford to be wrong so systems of record and trusted contextual data are everything.

Second, agentic frameworks now allow systems to evolve as businesses change, securely and at scale.

Third, integration barriers are falling. Modern AI can work across mixed legacy environments without forcing wholesale replacement. You still need trusted systems of record and deep domain expertise, but the new AI stack is a genuine multiplier, and it is evolving daily.

3) Safety, Trust, and Governance by Design

In industries where safety and uptime are non-negotiable, we cannot afford AI that hallucinates or obscures accountability. That is why our approach centres on verifiable data from systems of record, industry-specific models, permissioning, traceability and deliberate and transparent human oversight and accountability.

Our AI Trust function ensures every AI deployment is governed through pragmatic, case-by-case principles, focused on enabling innovation, underpinned by pragmatic scalable principles while managing risk with transparency and control.

4) Scaling Without the Chaos

Anyone who remembers or has been told about the 1990s era of enterprise integration knows what happens when innovation outpaces discipline. To prevent history from repeating itself, we have built processes and engagement models that productise what works.

Our IFS Nexus Black team develops next-generation AI capabilities, while IFS Loops turn high-impact digital workers into scalable, transparent and auditable components. A proven win in inventory, for example, can be extended to fleet or maintenance with no sprawl and full visibility.

And all of this is reinforced by the fact that IFS applications are already AI-native. Each month, new optimisation, generative and agentic capabilities are embedded directly into our products.

5) What the Best Are Doing Differently

The most successful organizations share a few common traits. They come to the table with a defined problem, a clear understanding of their data and realistic security expectations. They demand working prototypes in four to six weeks. And they validate partners not by promises, but by demonstrated domain expertise, credibility, commitment and early results.

At IFS, many of our people have spent years inside the industries we serve. That matters because context shapes outcomes. AI maturity is not built from the top down or bottom up, it is built through a sequence of tangible wins that build confidence and momentum.

The Bottom Line

It’s clear the headline from the FT event was: industrial AI is no longer an experiment.  It’s moving decisively from pilots to production. Across sectors, the winners are focusing on concrete jobs to be done, grounding their AI in trusted enterprise data, keeping humans in the loop, and scaling success through productised patterns. The message to every business and every vendor is simple: show measurable impact fast. Not in years, but in weeks and months.