Scarcity is pricing the logistics model out of the market. AI is your only way out

In the transport and logistics sector, artificial intelligence (AI) only delivers structural value when companies are willing to reinvent their business model. This does not mean they need to roll out AI broadly. On the contrary, they must make sharp choices, according to PwC expert Winand Doerga.

The Netherlands is grinding to a halt due to scarcity. Too little space, too little labour, tighter infrastructure and energy supply. Transport and logistics is, after agriculture, the sector with the largest spatial footprint, and relies heavily on low-paid labour and migrant workers. Policymakers are increasingly steering towards more value per square metre and per employee, and less space for logistics that primarily runs on cheap labour and many square metres.

This is not a trend you can wait out. It is a repricing of the conditions under which your company exists. And this is precisely where it goes wrong with how most transport and logistics companies look at AI. They see AI as an efficiency button: optimise routes, reduce costs per shipment, do more with the same fleet. Logical. But if you apply AI to your existing way of working, nothing changes about the underlying economics of your business. You save on a task, but the structure around it remains: the fleet, the square metres, the routes, the pricing logic per trip or per pallet.

So, you are paying twice. For the technology and for the business model you dare not let go of. Whilst that same model is being phased out by policy and scarcity. Those who deploy AI to make the current, space and labour-intensive model slightly more efficient do not reduce their risk. They spend money defending a model that the Dutch market is actively pricing out of the market.

AI is not your strategy. Reinventing the way you create, deliver and monetise value – in other words 'reinvention' – is the strategy. AI is your only realistic bridge from a capacity and execution model to a service and coordination model. So, the question is not to what extent you roll out AI. The question is what you dare to let go of.

Reinvention without a blank cheque: how to finance it from your own savings

At this point, most CIOs and CISOs rightly disengage. 'Reinvention sounds nice. But I cannot secure the agentic AI needed for this, and my board will not finance a gamble with the business model on top of what we have already spent.' This is not an objection to talk around. This is the core issue.

The answer is that reinvention is not an impossible project or goal and does not require a blank cheque. You finance the reinvention of your business model from the savings that AI is already delivering in domains with a low risk profile and proven applications. And you reinvest that money in coordination and data, where the real transformation lies.

Three domains deliver those first, measurable savings:

• Planning systems (TMS/WMS/YMS) with AI: dynamic allocation of orders to vehicles and drivers, prediction of waiting times, higher load factors. Result: higher productivity, lower costs per shipment.
• Customer service and exception management: AI assistants that detect deviations before the customer calls, generate recovery options and provide an estimated time of arrival (ETA) that moves with events. This ensures higher customer satisfaction and lower service workload.
• Supporting processes: order entry, tariff calculation, claims and documentation. Quickly measurable, scalable, monetisable. Precisely because of the low risk profile and proven applications outside the sector.

These are deliberately modest wins on demarcated tasks. This is not a weakness, this is the plan: they are the fuel, not the destination. With the room created here, you invest in supply chain coordination and ecosystems. In standardised data models and interoperability that create one shared view of reality, the basis for monitoring SLAs across the supply chain and planning with CO₂ awareness.

And the security? Agentic AI increases your attack surface. Autonomous fleets, purchasing agents and platform labour all bring new risk and governance burden. This is real. But security is not a reason to wait; it is a design requirement from the first domain. You should build human oversight, governance and security into your operating model, not bolt them on afterwards. You therefore deliberately start with the low-risk domains: there you build the governance muscle with which you can later handle the heavier, autonomous models.

The difference from the rest: humans remain in control

This is where the paths diverge from what you hear elsewhere in the market. The dominant narrative about AI in logistics is technology-driven: more capabilities, more autonomy, more agents. In this narrative, humans mainly feature as a cost item that disappears.

I believe the opposite. The companies that make AI work deploy AI agents as colleagues, not as replacements. An agent that strengthens the planner, the driver, the warehouse employee and the customer service employee, that removes repetitive drudgery and thus creates space for work where judgement and relationships count. In a sector that structurally struggles with labour scarcity, this is not a soft message, it is hard logic. Every employee that AI makes more productive instead of replacing is pure gain.

This does require a choice: investing in retraining, in new roles that emerge when people and agents work side by side, and in the human oversight that makes your governance credible. Reinvention that treats humans as a balancing item runs aground in the Netherlands on labour scarcity and on the legitimacy you need from employees and regulators.

Five AI business models that show how the economics are shifting

PwC researched nine new AI-driven business models worldwide that blur industry boundaries and create new value pools. Five of them directly affect how the transport and logistics sector will create value tomorrow. And each model is an illustration of the same shift: from capacity to service, from execution to coordination, with humans in control.

Four questions to translate this to your organisation

1. How would your fiercest competitor use AI to beat you? Imagine a well-capitalised entrant without a past, who reinvents your part of the chain with AI. Organise a scenario exercise with strategy, commercial, operations, supply chain and product development. Where must your strategy, investments and workforce move along, and what is the impact on your revenue, margin and personnel?
2. How does the customer journey change when AI assistants are the new buyers? Map the journey as if your organisation and the customer both operate via agents. If a shipper's AI system has no direct access to your real-time ETA, capacity or tariffs, you disappear from the shortlist before a human is involved.
3. What does your order processing look like in a hyper-personalised world? Use customer data to determine which personalisation (speed, flexibility, visibility, sustainability) actually delivers value. Then analyse the supply chain for obstacles: rigid planning systems, siloed IT, limitations with partners.
4. What does your workforce look like with people and AI agents side by side? Determine which agents strengthen your planners, drivers and warehouse employees, which new roles emerge, and which investments in data, security, governance and human oversight are needed. Invest in retraining so that employees see the agent as a colleague who removes drudgery and not as a threat.

Three choices you can make now

I do not end with a call to 'approach AI strategically'. You hear that enough already. Instead, three concrete choices:

1. Select one to three domains based on impact and feasibility (planning, customer service, supply chain coordination) that you genuinely redesign over the next twelve to eighteen months. Where you therefore do not just automate tasks.
2. Name what you are letting go of: square metres, a route structure, a volume-based pricing model, a department. No redesign without scrapping something.
3. Measure your AI deployment against margin, CO₂, service level and working capital, not against the number of tools you have rolled out.

Reinvention does not come from more AI. It comes from sharper choices and the courage to let go of square metres, routes and pricing models. In a country that is pricing the old model out of the market, this is not an ambition. It is the prerequisite for continued existence.