During the workshop we work with use cases that are familiar from the construction logistics chain. We don't present them as ready-made success stories (for some we are explicitly cautious), but as concrete situations to which we apply the legal, technical and organisational analysis. That is more instructive than an abstract conversation about "AI possibilities": participants discover for themselves where the line lies between a useful assistant and a high-risk system, and where the data foundation has to be sorted out first.
A first classic is route optimisation to construction sites. AI helps plan transport rounds from a hub to multiple construction sites, taking into account zero-emission zones, time windows, access restrictions, vehicle type requirements and driver availability. Technically that is feasible: it builds on classic routing algorithms with an AI layer for exceptions and demand patterns. But the real challenge is in the embedding: what does the planner see, what does the driver see, and who is ultimately accountable when an AI route gets something wrong.
A second is stock forecasting for the construction hub. Based on BIM 4D planning, current orders and historical deliveries, AI predicts what a hub will need in the coming weeks in terms of space, vehicle capacity and manpower. The legal bar is low here, but the data bar is high: without a tight integration between BIM, ERP and the hub WMS, the model produces rubbish. We go through the data foundation, the evaluation framework and how to measure a forecast without it starting to steer the planning through its own feedback.
A third is delivery estimation & just-in-time coordination. The system sends a reminder to a subcontractor, flags when a delivery doesn't match actual progress, or proposes a different delivery time. Useful, provided the communication is accurate. We discuss how to prevent AI from amplifying miscommunication instead of resolving it.
A fourth is vehicle clustering and last mile. AI clusters deliveries from the hub by vehicle type, route direction and time window to reduce the number of trips. This is where the measurement question gets interesting: how do you know clustering really saves CO2 and costs, rather than just producing a better schedule on paper.
A fifth is BIM 4D → delivery flow integration. An AI layer reads the BIM 4D timeline and automatically translates it into a delivery sequence, flagging when the planned activity deviates from actual progress. This is one of the most promising applications for reducing waiting hours and buffer stock, but it requires a data foundation that many organisations don't yet have in order.
A sixth is environmental zoning & zero-emission-zone-compliant planning. With zero-emission zones expanding in Dutch city centres, route planning is becoming more complicated. AI can help with vehicle choice, alternative delivery streets and cross-docking via a hub. We go through what that means for your TMS architecture and how to record that compliance evidence.
A seventh, for organisations going through a Wkb process, is a Wkb-compliant AI log. When AI contributes to decisions that end up in a quality file, you must be able to reproduce what the system said, with which input and which version. We discuss a logging architecture that enforces this without slowing down your operations.
These use cases serve as analysis material. We go through each case in terms of AI Act classification, GDPR, data foundation and feasibility. By the end, your team will share a view of what is sensible to investigate first in your chain, and what is not (yet).