The world spends trillions of dollars on infrastructure every year, but a significant portion of this money is lost due to errors, delays, and lack of information flow. Experts from EY-Parthenon argue that agentic artificial intelligence can reduce processes that take weeks to just hours and limit costly rework. For Poland, facing an unprecedented wave of investment, the stakes are particularly high.
Global infrastructure productivity has stalled or even declined, while investment needs are growing rapidly. Mistakes that require corrections cost up to 15 percent of the value of infrastructure projects worldwide. This translates into hundreds of billions of dollars annually that could have been avoided.
Agentic AI can detect an error before it reaches the construction site and reduce costs. It also shortens decision-making processes and helps identify risks earlier. According to the Construction Industry Institute, direct rework costs amount to about 5 percent of project value, and when indirect costs are included, the figure reaches 10-15 percent.
Agentic AI as a solution for fragmented data
The EY-Parthenon report, titled “The intelligence layer,” points out that although the infrastructure sector has vast amounts of data, this data is fragmented and trapped in silos, reducing its usefulness for decision-making. The title layer is a system of cooperating autonomous AI agents that connect processes that were previously isolated: planning, construction, and operation of infrastructure. This allows for real-time data delivery for quick decisions.
The sector currently has the foundations for applying agentic AI, but adoption is still in its early stages. The experts argue that agentic AI improves investment predictability, reduces rework costs, and provides better insight into project risk. If AI agents take over the task of collecting and reconciling data from different systems, processes that now take weeks could be shortened to hours.
Poland’s unprecedented infrastructure wave
Poland is facing a huge infrastructure investment gap. Public spending on infrastructure has increased significantly in recent years. – Poland is now facing an unprecedented wave of infrastructure investment. The investment gap remains enormous, and state spending on infrastructure development has grown significantly in recent years. On such a scale, even small inefficiencies translate into very high additional costs – said Maciej Ziomek, partner at EY-Parthenon Polska and leader of advisory in transport, infrastructure and ESG Due Diligence.
Ziomek emphasized that what matters is not only how much we invest, but how effectively we implement projects. He noted that Polish infrastructure investments lack a layer that connects information from various sources and translates it into actions. – This is where agentic AI comes in, as a solution that operates above existing systems, complementing them rather than replacing them – he added.
Trust and autonomy in AI deployment
The report presents four levels of autonomy for AI agents used in infrastructure projects: from simply informing, through recommending and acting with human consent, to independently performing tasks within defined boundaries. The choice of autonomy level depends on the type of decision, risk level, and consequences of errors. AI should not independently decide on high-consequence issues where human judgment and legal responsibility are critical.
Decisions requiring human intervention include construction changes affecting bridge durability and any actions that may directly threaten workers’ lives. Greater autonomy is recommended for routine, repetitive tasks with low risk, such as reconciling data between silos, generating progress reports, or ordering materials within defined thresholds. – For Poland, where a significant part of investments is carried out by the public sector, these are necessary conditions for scaling such solutions. Without them, the potential of technology may remain untapped – concluded Ziomek.
Źródło: WNP.PL, Fot. Shutterstock
