By Glenn Brouwer – CEO & Founder of Inspech
What Road Agencies Are Really Looking for from AI
Last week, I travelled to Varna, Bulgaria, to attend the IRF Global Summit on Funding & Leveraging AI Applications for Roadway Innovations. Like many people working in infrastructure today, I have seen AI move from an emerging technology to a regular agenda item at almost every industry event. Whether the discussion is about asset management, inspections, maintenance planning or public procurement, AI is now part of the conversation.
That made this event particularly interesting. It brought together road authorities, concessionaires, contractors, consultants and technology providers from across Europe to discuss not only what AI can do, but also how it can be applied in practice.
Whenever I attend conferences like this, I try to look beyond the technology itself. New capabilities are always interesting, but what often matters more is the direction in which the industry is moving. What problems are organisations actually trying to solve? What concerns keep returning in conversations? And where is the gap between what technology promises and what practitioners need?
A Broader Challenge Than Technology
Over the course of two days, I noticed that the discussions consistently revolved around a common challenge facing road agencies around the world: how to do more with limited resources while maintaining accountability, safety and long-term sustainability.
The applications of AI being presented were diverse. Some sessions focused on predictive maintenance and asset management, while others explored digital twins, safety analytics, traffic operations and data-driven investment planning. Although the technical solutions differed, they all addressed a similar objective: helping organisations make better decisions about increasingly complex road networks.
Alongside the technology discussions, there was a strong emphasis on funding and governance. Many speakers highlighted the pressure road authorities face as infrastructure ages, maintenance demands increase and traditional funding models come under strain. Questions about investment prioritisation, performance measurement, leadership and organisational readiness appeared just as frequently as discussions about algorithms and data models.
One panel discussion in particular reinforced this point. While the conference was centred around AI applications, the conversation quickly shifted away from technology itself and towards leadership, organisational change and procurement. The panel explored how road agencies need to adapt internal processes, governance structures and procurement models if they want digital innovation to deliver measurable results. It was a useful reminder that technology adoption rarely fails because the technology is inadequate. More often, the challenge lies in integrating new capabilities into existing workflows and decision-making processes.
What I found particularly interesting was that the conference was not focused on innovation for its own sake. Throughout the programme, there was a clear emphasis on implementation. Case studies, public-private partnerships, interoperability and institutional readiness were recurring topics. The message was consistent: technology only creates value when organisations are able to integrate it into their processes and use it to achieve measurable outcomes.
That observation stayed with me because it reflects a reality we see every day in our own industry. Whether the discussion is about asset management, maintenance planning, road safety or inspections, the challenge is rarely a lack of technology. More often, the challenge is translating technology into practical improvements that help organisations work more efficiently, make better decisions and remain accountable for the outcomes.
The Industry Doesn't Have a Data Collection Problem
For many years, collecting road condition data was a challenge. Today, that is rarely the bottleneck. Inspection vehicles, cameras and mobile mapping technologies can capture enormous amounts of information quickly and efficiently. Most organisations are not struggling to collect data anymore. In fact, many are collecting more data than ever before.
The challenge starts once that data arrives on someone's desk.
Road authorities and inspection companies are dealing with increasing inspection volumes, growing reporting requirements and rising expectations around transparency and accountability. Yet staffing capacity is not increasing at the same rate. Experienced inspectors remain difficult to find, and the professionals already in place are expected to process more information in less time.
That reality changes the discussion.
The question is no longer how to collect more data. The question is how to process that data efficiently without compromising the quality of decisions.
That was the central theme of the presentation I gave in Varna.
Making Expertise Scalable
Rather than focusing on AI as a replacement for inspectors, I argued that the industry should think about AI as a way to scale inspection capacity while preserving expertise. The most valuable role for AI is not making decisions on behalf of professionals. It is helping professionals work more effectively.
Anyone who has spent time with road inspectors understands the challenge. A significant portion of their work consists of reviewing footage, searching for relevant defects, validating findings and documenting results. Much of this work is repetitive, yet it requires concentration and consistency. As workloads increase, so does cognitive pressure. The risk is not that inspectors lack expertise; the risk is that too much information has to be processed in too little time.
This is where I believe AI can make a meaningful contribution.
Not by removing inspectors from the process, but by supporting them throughout it.
At inspech, we use AI to help structure large volumes of road video data, identify potential defects and support consistent classification. However, the inspector remains responsible for reviewing and validating the results. The goal is not autonomous decision-making. The goal is to help inspection teams focus their attention where it matters most, reduce repetitive work and create a more efficient workflow.
In many ways, this is less about artificial intelligence and more about operational design.
Technology becomes valuable when it helps people perform their work more effectively. A defect detection model on its own has limited value. The real benefit comes when inspectors can review prioritised findings instead of manually scanning every frame of video. It comes when inspection data is organised in a way that supports reporting, planning and maintenance prioritisation. And it comes when organisations can build a structured inspection history that allows them to understand how assets are changing over time.
The Conversations After the Presentations
What struck me most after the presentations was that many of the conversations naturally moved in this direction.
Very few people were interested in discussing how inspectors could be removed from the process altogether. Instead, the questions focused on how organisations could maintain quality while handling increasing workloads, how they could improve consistency across networks and how they could make maintenance decisions with greater confidence.
Those are practical concerns, but they are also important ones.
Infrastructure management ultimately depends on accountability. Decisions about maintenance budgets, repair priorities and network safety need to be transparent and defensible. For that reason, explainability and professional judgement remain essential. Technology can support those decisions, but it cannot replace the responsibility that comes with them.
Looking Ahead
As I prepare for the journey home, that is probably my biggest takeaway from Varna.
The discussion around AI is maturing.
A few years ago, the focus was largely on what AI could detect. Today, the more interesting question is what AI enables organisations to do with that information. How can it help inspection teams work more efficiently? How can it improve consistency? How can it support better long-term asset management?
Those questions move the conversation beyond technology and towards outcomes.
In the end, road authorities are not investing in AI because they want more algorithms. They are investing in solutions that help them maintain infrastructure responsibly, manage risk and make better decisions. The organisations that will create the most value are not necessarily those pursuing the highest degree of automation. They will be the ones that successfully combine technology, process and human expertise into a workflow that is efficient, explainable and trusted.
That is certainly the future we believe in at Inspech. Not because it is the most ambitious vision of AI, but because it is the one that best reflects the realities of the people responsible for managing our roads every day.
Continue the Conversation
During the IRF Global Summit on Funding & Leveraging AI Applications for Roadway Innovations, I shared my perspective on how AI can help road agencies scale inspection capacity while preserving professional expertise.
If you would like to explore the topic in more detail, you can access both the presentation and the demonstration shown during the session.
What you'll find:
✓ The IRF Varna presentation handout
✓ A demonstration of Inspech in practice
✓ Examples of AI-assisted road inspection workflows
✓ Insights into how road agencies can move from data collection to better decision-making
