From complexity to clarity: How AI is enabling maintenance

Given my role as EMEA lead for maintAI, Wood's maintenance optimisation solution, it may come as a surprise that I don't class myself as an expert on artificial intelligence (AI).

I'm ultimately just an end user - someone who's had operations experience and worked with teams maintaining complex assets in high-risk environments. Moving into this role at Wood, I've seen firsthand the impact AI can make. The opportunities are endless and exciting developments lie ahead - and that's exactly why I continue to speak about the power AI has for maintenance.

But to convey this impact in energy and material sectors, we need to talk about it in a way that makes sense to people.

Five years ago, when we first started using AI at Wood, the conversations were tough. It felt like a "magic box", something mysterious and untested. Thankfully, that perception has changed.

Today, industries are more onboard with AI than ever before. They've seen what it can do, and they're starting to trust it. It's a bit like when oil and gas operators first began outsourcing work to contractors decades ago. At first, there was hesitation. Teams wanted oversight and reassurance.

But over time, trust grew as results spoke for themselves. AI is on a similar journey. We still need people to validate its output, but confidence is building and soon, we'll trust it to take on more. That's when the real transformation begins.

And that transformation starts with how we talk about it.

Keeping it simple

Messaging matters, and AI can sound intimidating. It's often wrapped in technical jargon, presented at conferences in ways that can feel disconnected from the day-to-day reality of maintenance teams and asset owners.

Here's the truth: AI isn't magic. It's a tool - and a powerful one - that helps us make better decisions, faster.

At Wood, we've been using AI to support smarter maintenance strategies for years. Not because we invested millions in flashy tech, but because we know how to combine technology with subject matter expertise, as well as shaping our offering around what our clients needed.

When it comes to the world of maintenance, AI is leveling up strategies to unlock what I like to call maintenance 5.0.

What does maintenance 5.0 actually look like?

Today, maintenance 5.0 means using data and AI to assess asset performance, optimise maintenance routines, prioritise work based on risk and impact and embed insights directly into work processes.

Asset owners can make smarter decisions, faster - without overhauling their entire way of working. It's designed to be scalable, intuitive and embedded into existing workflows, making AI adoption seamless and sustainable.

A key part of maintenance 5.0 is global benchmarking. Because we work across sectors and locations, we're able to compare performance data from different types of assets and industries.

This gives our clients a unique advantage - the ability to learn from others, spot gaps and adopt best practices that might otherwise stay siloed. It's a powerful way to drive continuous improvement, even in mature operations.

Smarter doesn't always mean more

One of the more surprising insights is that sometimes, doing less maintenance is actually the smarter move. Traditional approaches often default to "fix it before it breaks," but AI helps us understand how assets really fail, and when intervention might cause more harm than good.

Take the bathtub curve: equipment tends to fail more often early in its life (infant mortality), then stabilises and eventually fails more frequently again as it ages. If you maintain too aggressively, you could reset that clock and cause more failures.

AI helps us spot these patterns and make smarter choices - across all industries.

In one of our early projects, AI helped reduce maintenance effort by 15% - without impacting asset reliability. By analysing four years of historical asset data, we streamlined processes and avoided unnecessary maintenance interventions. Three years on, the results show no drop in performance, but a significant cost saving: £1.5 million annually.

AI for asset maintenance transcends sectors

The value of AI in maintenance isn't limited to one industry. The challenges faced by offshore oil and gas operators - ageing assets, limited resources, pressure to do more with less - are mirrored in mining, renewables and other heavy process industries.

At Wood, we've applied the same AI-driven techniques across upstream and midstream oil and gas, lithium processing in the U.S. and mining operations in Australia. The equipment may differ, but the principles are the same - and the opportunity to share learning across sectors is huge.

By collecting and comparing data from different industries, we help clients learn from each other - even if they'd never normally be in the same room.

Based on the benchmarking results shown above, Client C (Refinery) demonstrates the optimal maintenance strategy, showing an improving reliability trend despite a longer task frequency of 24 months, outperforming the reliability of Client B and closely matching Client A. By making a simple change to their maintenance strategies, Clients A and/or B can reduce maintenance effort whilst also remaining confident that they will not see detrimental impacts to uptime & availability. This is where AI is bringing best practice and supporting people to be confident in making the change; true data-led decision making.

The real challenge? Change management

AI doesn't work unless people do. And in industries like oil and gas or mining, change is hard.

No one wants to be first to adopt. The risks are high and the default is often "do what we've always done." We've seen it firsthand - teams sticking to legacy processes because it's comfortable and familiar.

That's why we focus on persuasion through proof when convincing asset owners and maintenance teams on the value of AI. Show people that others have made the change, and that it worked. That's how you build trust.

Beating perceptions - AI isn't replacing humans, it's amplifying them

There's a fear that AI will take jobs. But in reality, AI isn't in the decision-making seat, we are. What AI does is give us better information, faster. It helps us prioritise limited resources, spot risks earlier and make more confident choices when it comes to maintenance.

In fact, the most exciting development in AI right now is the rise of small language models - tools trained on specific, trusted datasets that act like niche experts. Imagine having a maintenance SME in a box, ready to answer questions, benchmark performance and guide strategy in real time.

We're starting to explore this at Wood, and it's super exciting.

Combine that with agentic AI - systems that can iterate and improve without constant human input - and you've got a future where AI becomes a true partner for maintenance.

Getting started: looking for one simple step?

If you are an asset owner or reliability engineer just starting with AI, here's my advice: experiment. Play with tools like ChatGPT in your own time. Get comfortable. Understand what it can and can't do (without sharing private company data in open environments of course!).

That curiosity is the first step toward building trust - and toward seeing how AI can help your team, not replace it.

The road ahead: smarter AI, stronger maintenance teams

The next five years will be defined by AI tools that are smaller, smarter and more specialised. They'll be embedded in our workflows, helping us plan, schedule and execute maintenance more efficiently. And they'll be accessible - not just to a select group of people, but to the people on the ground too.

I see Maintenance 5.0 as an integration of physical technology, agentic AI, the best language models and human expertise to create a collaborative, intelligent ecosystem that enhances decision-making, automates routine tasks, and drives predictive, adaptive asset care.

But ultimately, maintenance 5.0 isn't about technology. It's about people. And AI, when used right, empowers them to do their best work.

Interested to hear more about maintenance 5.0? Drop me a message to learn more or join me at ADIPEC where I'll be presenting how AI can improve safety barrier management in partnership with A. Rampaul from Adnoc Sour Gas- using a real life case study from an onshore site.