Posts in category: AI & Deep Learning

Mike Talbot, CTO at SFG20, explains how AI is shaping fire safety compliance by streamlining maintenance tasks, improving accuracy and reducing reliance on manual processes 

Artificial intelligence (AI) is transforming the way buildings are maintained, with a growing impact on fire safety and compliance. SFG20, a leader in building maintenance, is leveraging AI to improve efficiency, accuracy and regulatory adherence. While AI is often associated with automation, its role in fire safety goes beyond simple processes, helping to manage risk, streamline inspections and enhance preventative measures.  

In this interview, Mike Talbot, CTO at SFG20, explains how AI is being applied in building maintenance, its relevance to fire safety, the challenges of adoption and the future potential of AI-driven compliance solutions. 

Can you introduce SFG20 and your role at the company? 

SFG20 is the go-to standard in the UK for planned preventive maintenance. It has been around for more than 35 years and evolves with changing best practices and legislation. The standard is a way that organisations can outsource the complex tasks of keeping up to date with best practice and legislation, ensuring that their facilities are efficiently maintained and compliant with the law. We do this by providing detailed schedules on how and when to maintain building assets.  

My role focuses on technology, specifically on leveraging it to enhance compliance and safety.

Mike Talbot

We do this by combining the detailed content in our standard with a forward-looking modern software platform that uses AI tools —like those from OpenAI. Our aim is to simplify the complex job of upkeeping the standard, increasing understanding of our maintenance schedules and matching building assets to the right maintenance tasks. 

How does SFG20’s work in building maintenance relate to fire safety and compliance? 

At SFG20, we make sure that maintenance tasks critical to fire safety are not missed by maintenance professionals. Our schedules clearly outline what is required to keep fire safety systems working properly and compliant with regulations. With the help of AI, we simplify complex documentation into straightforward steps, providing overviews of what each maintenance schedule is for. Most recently, we have automated the matching of schedules to assets within a building, including those that encompass fire prevention, fire management and evacuation. This automation makes sure that all asset maintenance stays on track, without having to rely on best guess matches and subjective associations. 

What specific ways does AI enhance building maintenance and safety? 

At SFG20, AI significantly streamlines the tricky job of figuring out exactly which maintenance tasks are needed for each asset. Instead of sifting manually through lengthy lists and complicated asset registers, our AI can quickly identify what’s important, ensuring no vital safety task slips through the cracks.

We employ AI to help our customers find the correct schedules manually too, allowing them to ask questions about particular equipment and manufacturers and provide a comprehensive summary of the correct maintenance schedules. 

Mike Talbot

We’re increasingly seeing AI combine with sensors and the Internet of Things (IoT) to help predict maintenance issues before they arise. AI has the potential to act as an expert remote worker, capable of taking on tasks traditionally handled by people. Thanks to recent advancements in Large Language Models, many of these tasks can now be completed quickly and efficiently. AI is particularly useful as a virtual team member, identifying and reporting issues for human review. For example, it could monitor sensor data alongside user complaints and notifications—something that would typically be too costly to have a person manage on a daily basis.  

How can AI improve fire risk assessments and compliance monitoring? 

AI helps by automatically identifying the key maintenance tasks required by law and best industry practice, ensuring these critical tasks are always scheduled and completed on time. Employing AI technology will mean we are less dependent on individual engineers, facilities managers, or surveyors guessing the right tasks to keep a building safe and compliant. 

Beyond what we do at SFG20, AI can also be used to sketch out risk assessments, by probing into key factors and combining information from many different sources to have an “expert on hand”. 

What are the biggest challenges in maintaining fire safety and how can AI help? 

The big challenge is making sure nothing important is overlooked and that records are accurate and up to date to ensure the situation on the ground is clearly understood and issues addressed. Without complete visibility, there’s a risk that critical fire safety tasks, like equipment checks or remedial actions, slip through the cracks—potentially putting occupants at risk and leaving building owners exposed to legal consequences. 

AI can help this by automatically aligning tasks to building assets, identifying gaps and making sure all legally required maintenance tasks are properly documented and tracked. It can also flag overdue actions, monitor patterns of non-compliance and provide real-time insights, helping duty holders stay in control. 

Are there concerns about AI’s accuracy and reliability in safety-critical areas? 

Of course—accuracy and reliability are always concerns, especially in critical areas like fire safety.  The best way to address this is to use AI to review, summarise, or collate information rather than create it.

AI is getting smarter, but as of now, human review is still required and we always ensure to use expert oversight to assess how AI tools perform.

Mike Talbot

AI is accelerating tasks at a significant rate but is yet to fully automate them.  It is possible to imagine a time when this won’t be necessary, but SFG20 will always use humans to ratify and validate output, as we do for content written by anyone. 

How does SFG20 ensure that its AI solutions are built on accurate and relevant data? 

We don’t train AI models ourselves. Instead, we provide them with information and design prompts, along with frameworks and functions, to make them effective. By combining the broad knowledge of Large Language Models (LLMs) with their ability to execute tasks, we build agentic systems that solve the problems faced by us and our customers. At the core of these systems is the SFG20 library of schedules—our unique content guides every decision and output. 

What regulatory or industry challenges exist in adopting AI for fire safety? 

Adopting AI for fire safety faces several regulatory and industry challenges. Current fire safety regulations often lack specific guidance for AI-driven systems, creating uncertainty around compliance and certification.

There are also concerns over liability and accountability if AI fails to detect or respond to fire hazards. Industry-wide standards for testing and validating AI fire safety technologies are still emerging, slowing widespread adoption. Additionally, integrating AI into existing infrastructure can be complex, requiring alignment with traditional safety protocols.

Finally, data privacy and security are key considerations, particularly when AI systems process sensitive building or occupancy information to enhance risk detection. 

Do you see AI playing a role in predicting fire risks before they occur? 

Absolutely. AI is great at spotting patterns and anomalies in maintenance data or sensor information that humans might miss—such as noticing that smoke detectors are triggering false alarms more frequently or identifying a pattern of overheating in electrical systems.

Catching these issues early means we can often predict and address potential fire risks well before they become serious, significantly boosting safety.

Mike Talbot

Over time, as AI analyses more data across multiple buildings, it can even start to identify wider trends and risk factors, helping inform smarter maintenance strategies, prioritise high-risk areas and support more proactive decision-making to prevent fires before they start. 

This article was originally published in the April 2025 Edition of International Fire & Safety Journal. To read your FREE copy, click here.