Via Salis begins piloting AI tools on the Czech Republic's 48km D4 motorway under a 28-year PPP

Via Salis - the concession company behind the Czech Republic's first PPP motorway - has begun testing AI-based tools on the 48km D4 highway between Příbram and Písek, according to comments published by Global Construction Review in mid-2025^[1]. The road opened to traffic in December 2024^[1] under a 28-year concession valued at approximately €530 million, signed in February 2021 between the Czech Ministry of Transport and Via Salis.

The project and its partners

The D4 is the Czech Republic's first highway procured through a public-private partnership^[1]. It was delivered by a 50/50 consortium of Meridiam and Via Salis, the latter comprising Vinci Concessions and Vinci Highways^[1]. Construction work - 32km of new greenfield road and 16km of refurbished brownfield sections - was completed in under four years, employing as many as 1,600 workers at peak.

The tolled road cuts travel time between Prague and Písek by around 20 minutes^[1]. Via Salis selected German AI, BIM and built asset management firm Thinkproject to manage all project information and workflows across delivery, handover and operations^[1].

AI in a pilot phase, not yet core

Via Salis communications director Daniela Pedret told Global Construction Review that AI has not yet been widely integrated into the core systems of the D4 project^[1]. Testing is limited to selected systems, with the goal of evaluating potential efficiency gains before any broader rollout^[1].

Thinkproject chief executive Renzo Taal framed the situation as typical of the wider infrastructure sector: the priority has been building trusted information foundations before layering AI on top^[1]. His argument is that project data is often siloed across disconnected systems and documents, and that AI cannot deliver value until that underlying information is connected and governed^[1].

Areas where Via Salis sees AI potential over the life of the concession include:

• Common Data Environments and model checking
• Document management and version control
• Operational dispatching and maintenance scheduling^[1]

Thinkproject's own position is that the next phase of AI should move beyond information retrieval toward supporting decisions and automating repetitive tasks - helping teams understand what matters and act on it, rather than simply surfacing what happened^[1].

Information management as the prerequisite

Thinkproject's role on D4 has centred on providing the Common Data Environment that underpins collaboration and contractual compliance^[1]. A CDE functions as a single governed platform for collecting, managing and distributing project information across all parties. On a 28-year concession, that continuity of data from construction through to operations is commercially significant: asset records created during the build phase must remain accurate and accessible for maintenance teams decades later.

The D4 concession agreement runs to 2049. Whether the AI pilots currently under evaluation on the highway mature into operational tools - and whether the Czech Republic's Ministry of Transport moves forward with further PPP procurements in the region - will be the clearest indicators of how seriously the sector is treating digital infrastructure as a long-term asset management discipline.