AI in Building and Public Works: the Challenge Is No Longer Technological, but Informational
A project-based sector where knowledge is easily lost
The construction industry has a structural particularity: it operates through temporary projects. A construction site is set up, brings together teams and partners for several years, then comes to an end. Teams are reshuffled, organisations change and, all too often, part of the experience gained along the way disappears with them.
Unlike industry, where a permanent site makes it possible to embed tools and processes in a continuous improvement approach, the building and public works sector finds it more difficult to retain and reuse information from one project to the next. Knowledge still largely relies on the individual experience of employees rather than on shared systems capable of making it accessible and usable over time. This is one of the main reasons for the gap between the promise of AI and its actual deployment on construction sites.
Data: the true raw material of AI
To generate relevant recommendations, AI needs access to available, structured and usable information. Yet in construction, data still remains widely scattered across companies, software platforms, documents and projects. When a construction site is not sufficiently represented within a shared digital environment, it is difficult to expect a system to optimise its organisation or anticipate risks.
This is where the real challenge lies: not simply in digitising, but in capitalising on data and knowledge. The development of common data environments, better circulation of information between stakeholders and the preservation of lessons learned are becoming prerequisites for moving from isolated applications to AI that is genuinely integrated into operational processes.
Start with business needs, not technology
There are already many potential use cases: optimising schedules and task sequencing, monitoring construction progress through cameras, analysing regulations and contracts, predictive maintenance of machinery, optimising energy consumption, or simulating scenarios to reduce the carbon impact of technical choices and material sourcing
One condition remains essential, however: starting from a concrete operational need. At Leonard, AI projects are first assessed according to their feasibility and business value before being developed and, where relevant, industrialised. This pragmatic approach is all the more necessary because moving from a prototype to a solution genuinely deployed across construction sites requires significant investment.
From experimentation to transformation at scale
The next stage could be agentic AI: systems capable not only of generating an answer or assisting with a task, but of carrying out a sequence of actions within a complete process. On a construction site, this could eventually mean continuously analysing project data, identifying potential delays and suggesting operational adjustments.
But this evolution makes the initial requirement even more critical: without a data strategy, there can be no scaling up. The AI revolution in building and public works will therefore depend less on the emergence of new tools than on companies’ ability to organise their information assets, share lessons learned and train teams to replicate successful use cases from one project to another. The technology is mature; the pace of transformation will now depend on how effectively the sector structures and shares information.
See you at INTERMAT 2027
From 21 to 24 April 2027 at Paris Nord Villepinte, INTERMAT will bring together the entire building and public works industry around the solutions and technologies transforming construction sites. Among the show’s five areas of expertise, the New Technologies & Energies sector will notably showcase innovations related to artificial intelligence, IoT, BIM, drones, robotics, 3D printing and new energy solutions. A key space to discover innovations, compare real-world applications and accelerate their adoption in the field.
