The Future of Industrial AI is Execution, Not Just Insight
Suliman Gaouda, Regional Vice President, AI – APJ MEA at IFS, discusses how agentic AI is moving industrial organisations beyond insights to execution, while addressing workforce shortages, data sovereignty and the growing need for governed AI across Saudi Arabia and the wider Middle East.
What are you showcasing at LEAP 2026 that is genuinely new, and what problem does it solve for customers?
The genuinely new part is the execution layer. Plenty of vendors will show you an anomaly detected. Very few will show you the work order raised, the technician scheduled, the part reserved, and the record written, without a human stitching those four steps together.
Three products do that. IFS Loops digital workers are pre-built for industrial work rather than assembled per project, so the first one is live in ten to fourteen weeks rather than a twelve to eighteen month build. Nexus Black Resolve deploys in ninety days and works above whatever the customer already runs, whether that is IFS Cloud, SAP, Oracle, Maximo, or Hexagon. It reads asset data, work order history, parts catalogues, and contracts, and writes back completed work orders and structured fault reports. IFS Cloud is not a prerequisite. And Operational Intelligence sits above the historians, SCADA, and DCS systems already in the plant, applying anomaly detection without replacing any of them.
The problem it solves is blunt. Industry average first-time fix rate sits at seventy to seventy-five percent, which means roughly one in four jobs needs a return visit. Thirty percent of a technician’s day goes on admin rather than wrenching. Analytics that stop at insight do not touch either number. Somebody on the frontline still has to translate the recommendation into work, and that somebody is exactly the resource our customers are shortest of.
Saudi Arabia is investing heavily in AI and digital transformation. What are you seeing from Saudi customers that differs from demand in other markets, and how is this influencing your product strategy?
Three differences, and they are real rather than diplomatic. First, the timelines are shorter. Mandates in this region carry dates, and those dates sit inside two years, not inside a five-year plan. That rewards configuration over bespoke build. It is why our strategy is pre-built industrial agents rather than a platform plus a consulting engagement. Building an agent from a framework is delivery cost. A pre-built industrial worker is inventory.
Second, sovereignty is a requirement, not a preference. Customers want to know where the data sits, where the model runs, and who holds the keys. We support running IFS Cloud in a customer’s own data centre or private cloud with full access to our AI capability, identical to cloud-hosted customers, from release 25R1 onward. That has moved from a niche request to a standard one, and we are honest about the boundary: fully air-gapped deployment is a different engineering problem and we do not pretend otherwise.
Third, national capability transfer is a scoring criterion, not a courtesy. Customers are not only buying an outcome, they are buying the local capability to run it. That shapes how we structure delivery, not just what we ship. We sent the MEA team to London for training on Nexus Black for exactly this reason.
What is the biggest technology challenge your customers in the Middle East are asking you to solve right now?
Workforce, and it is not close. Half the skilled industrial workforce is expected to retire within five years. Half the skilled maintenance workforce across industrialised economies is already over fifty. There is an estimated 2.6 million worker deficit globally, and seventy-three percent of field service leaders name an ageing workforce as a strategic threat. Nobody in this region is hiring their way out of that.
Meanwhile the asset population is detonating. Connected, instrumented, increasingly operator-free assets are multiplying faster than any organisation’s capacity to manage them, and every single one needs a lifecycle owner. Those two lines cross. That is the challenge. Not “give us better dashboards” but “grow output without burning our people out.”
It is also why the framing matters. Efficiency arguments do not move an executive who is already running lean. Hours given back to the frontline do. And there is a second-order benefit people miss: when a senior expert retires, their judgment does not walk out of the door if it is already encoded in the agent. A new hire can be oriented against that accumulated base in a fraction of the time. Institutional knowledge stops being something you lose every time somebody resigns.
LEAP brings together global technology companies, startups, investors, and government organisations. What partnership or ecosystem opportunities are you looking for at the event?
Three, in priority order. Sovereign cloud and compute operators. Our software is the layer that turns their capacity into industrial outcomes. We do not compete with them and they do not build what we build. Systems integrators with genuine industrial delivery capability. Our constraint in this region is delivery capacity, not demand, and we would rather say that out loud than pretend otherwise. Twenty-nine Middle East customers is a real base, not a dominant one.
Government and government-linked asset owners. Ports, rail, utilities, defence sustainment, aviation MRO, and now data centres. These are the organisations where the asset is the business, and where an operating layer either exists or the AI ambition stalls.
What we are not looking for is another analytics partnership. There is no shortage of insight in this market. There is a shortage of execution. PwC’s global chairman, Mohammed Kandi, made the point better than I can: large language models on their own are not enough, and companies need Industrial AI products that change how work gets done rather than making old processes more efficient.
How important is Saudi Arabia to your regional growth strategy, and are you using the Kingdom as a launchpad for expansion across the wider GCC, Middle East, and Africa?
Saudi Arabia is where the industrial diversification capital is actually being deployed, so it is central rather than important. On the launchpad question, I would be careful with the word. The Kingdom is not a beachhead we are using to get somewhere else. The industrial base here, energy, resources, manufacturing, and the infrastructure programme around it, is large enough to justify the investment on its own terms. We have been in-Kingdom for years: STC since 2023, ARO Drilling across onshore and offshore operations, International Maritime Industries at Ras Al-Khair, BAE Systems.
What is true is that the operating patterns transfer. A turnaround is a turnaround, whether it is in Jubail, Ruwais, or Richards Bay. An engine shop is an engine shop, whether the tail belongs to Saudia or to Air France KLM, who run more than five hundred and seventy aircraft on a single instance of our maintenance platform. So the reference architecture we prove here does travel across the GCC and into Africa, and that is a real commercial advantage. But the sequencing is Kingdom first because that is where the assets are, not Kingdom first as a stepping stone.
Looking ahead three years, which technology do you believe will have the biggest impact on your customers in the Middle East, and what are businesses still underestimating today?
The biggest impact will come from governed agents that execute, rather than from larger models. The models are already good enough for most industrial work. What is missing is the governance, the audit trail, and the integration into systems of record that lets an organisation actually let an agent act.
Two things are being underestimated. The first is the data foundation, and specifically where it comes from. AI is only as good as the operational data system feeding it, and in industry that data is generated by the frontline: the shift log, the inspection, the operator’s note, the sensor on the pump. Organisations are buying compute and models while leaving that layer untouched, then wondering why the pilot did not scale.
Our own research found industrial AI adoption heading from thirty-two percent to fifty-nine percent inside twelve months while only twenty-nine percent of organisations trust AI for strategic decisions. That gap is not a model problem. It is a data provenance problem.
The second is governance. The agentic mandates in this region will not fail on model quality. They will fail on governance, because nobody could answer who authorised the action and on what basis. Noble Corporation’s CIO framed the bar correctly for our world: in an industry where safety is paramount, you have to be 99.999 percent accurate, which is why having a human in the loop remains absolutely critical.
