Lukas Olofsson is Product Manager at Centuri, working to develop the platform with the latest functionality – in a value-creating, responsible, and secure way. We asked him to share his key insights on AI in quality management: how it has changed the industry, where it makes the biggest difference, and when you actually shouldn't use it.
Hi Lukas! You meet many companies and their quality departments in your work. How mature is the industry when it comes to AI?
It varies enormously. Some have come a long way and are actively working with automated workflows and data-driven follow-up, while others are still at the starting point.
What I see most right now is a great deal of interest combined with some uncertainty about where to begin. Many know the potential is there, but aren't quite sure how to take that first step without it becoming a large and expensive project.
When we talk about AI in quality management – what do we actually mean?
There are primarily two areas where AI can make a real difference. The first is about creating information. Many companies have already made good progress here. Using models like Claude and ChatGPT, you can easily produce documents, routines, and process descriptions.
The second is about managing information: using AI to streamline analysis, generate insights, and drive improvement work. That's where the really significant potential lies, and it's also where most companies still have a lot to gain.
AI is everywhere today – but far less so in quality management specifically. Why is that?
AI is already used to some extent in quality work, but nowhere near as much as in many other areas – and there's a clear reason for that. In quality management, you often handle sensitive data: deviations, audits, process documentation that shouldn't leave its context. This means you can't leverage external AI services in the same way as in other functions.
To truly harness the potential of AI in quality management, the functionality needs to be built into the management system itself, so that all processing takes place within a controlled environment.
But it's not just about technology. In quality management, control isn't just desirable – it's a fundamental requirement. And that places demands on how you think about AI use. It's not enough to ask "can we use AI here?" – you also have to ask "is it appropriate to use AI here?"
How has AI changed quality work?
Quality work has shifted from being about creating critical information to reviewing and quality-assuring it. Developing processes, routines, and similar materials is easier today than ever before – but the challenges now arise in ensuring that information is correct, approved by the right people, has gone through the right steps, and ultimately reaches the right recipients.
This places new demands on governance and control, but also opens up a more structured, consistent, and value-creating approach to quality work.
What risks do you see with AI in quality management?
In regulated industries with high demands for safety and control, the central question is how to best capture the opportunities that AI opens up without compromising governance and compliance. It's not about choosing between innovation and control, but about finding a responsible way to combine them.
What is your best advice to quality managers who want to start using AI in quality work?
Start with a concrete problem, not with the technology. It may sound obvious, but it's the most common mistake I see – starting by looking at tools and then trying to find a problem to solve with them. That almost never works out well.
Instead, find a process that is repetitive, takes unnecessary time, and where you already have some data. Deviation reporting is a classic example. Many teams spend a lot of time writing reports, assigning cases, and following up on actions manually – and that's exactly the kind of work that can be automated quickly with clear results.
What do you see companies missing most when introducing AI or automation?
Data quality. It's almost always data quality, and that's the entire foundation.
AI and automation can do fantastic things – but only if the input data is reliable and consistent. If deviations are classified differently depending on who reports them, if measurement data is missing for certain periods, or if documentation lives in email chains and Excel spreadsheets – then the analysis will be misleading regardless of how good the tool is.
My advice is to set aside time to structure and quality-assure your data before starting with advanced analysis. It may feel like tedious work, but it's the investment that determines whether your AI initiative pays off or not.
Can you give an example of an area within quality work where AI makes the biggest practical difference?
Predictive quality control is what impresses most when you actually see it in practice. That a system can flag that a process is about to produce substandard quality based on trends in process data – before the problem appears in the product – that's a fundamentally different kind of quality work.
Traditionally, we react when something has already gone wrong. With predictive analysis, you can act before it happens. That's a significant difference, both financially and for the teams who are spared the reactive stress.
But it requires good data and some patience with training and validating the models. It's not plug-and-play, but the potential is real.
What happens to the quality manager's role as AI takes over more of the operational work?
In a world of AI, quality work and the role of quality professionals becomes more important than ever. That is my firm conviction.
When systems handle the repetitive and rule-based tasks, time is freed up for what actually requires experience and judgment – understanding deeply why problems arise, driving improvement work strategically, building a quality culture within the organisation. These are things AI cannot do.
How do you see AI in quality work in three to five years?
I believe predictive analytics and automated workflows will be standard, rather than a competitive advantage. Companies that haven't digitalised and automated their core processes will struggle to keep pace – both in delivering quality and in meeting the demands of auditors and customers for traceability and documentation.
What I find most exciting is the connection between quality data and other business data. Today, the quality system often lives in its own silo. Going forward, I believe we'll see much closer integration with production, procurement, and customer data – and that will provide a completely different ability to understand where quality problems actually originate and what they cost.
Finally: one piece of advice for the quality manager who feels overwhelmed by all the talk about AI?
You don't need to do everything at once. Choose one process, structure your data, automate one workflow, and measure the effect. Then the next one. That's how this kind of change actually works in practice – step by step, not as one big technology project.
And don't forget that the fundamentals still apply. AI improves quality work, but it doesn't replace well-developed routines, clear standards, and a well-functioning structure for information management.
Want to know where AI can make the biggest difference in your organisation's work with information management and quality? Book a meeting with us – we'll help you find the right starting point.