AI depends on the data foundation

AI is now part of almost every supply chain technology conversation, but its value depends on the quality and accessibility of the data behind it.

Jethro Borsje’s session at DELIVER Europe 2026 focused on a practical challenge: many AI initiatives fail not because the models are weak, but because the data foundation is not ready. If organisations cannot trust the data feeding AI-enabled workflows, the outputs become difficult to trust as well.

For supply chain leaders, this means AI readiness is not only a technology question. It is a data integration, governance and process question.

Supply chain data sits beyond the organisation

One of the central points from the session was that supply chains are not contained within one company’s systems.

Critical information often sits with suppliers, carriers, logistics service providers, customers and other trading partners. If organisations only connect the data inside their own four walls, they may still be missing the external signals needed to make better decisions.

This is especially important for use cases such as claims management, order processing, fulfilment, inventory updates and exception handling. AI can only act on what it can see. If external data is fragmented, delayed or manual, automation becomes less reliable.

Existing systems can become AI-enabled resources

The session challenged the idea that organisations always need to replace core systems before they can benefit from AI.

Instead, the focus was on creating an integration layer that can connect existing ERP, CRM, point-of-sale, supplier and logistics systems, then make those systems available to AI-enabled workflows in a controlled way.

This approach allows organisations to use the systems they already have while improving how data flows between them. It also helps avoid creating a fragmented AI landscape where different tools connect separately to different sources without governance.

Context is what makes AI useful

AI tools can process documents, emails and natural language quickly, but they still need business context.

For example, an AI tool might read an order document accurately, but if it does not understand updated article numbers, price lists, discounts or internal product data, it can still produce the wrong outcome. Accuracy in extraction is not the same as accuracy in business decision-making.

That makes real-time contextual data essential. AI needs to work with current information from the organisation’s own systems and trading partners if it is going to support reliable automation.

What this means for the DELIVER community

The session positioned data strategy as the foundation for AI strategy.

For retailers and brands, the priority is to connect external partners and internal systems in a way that is automated, governed and scalable. For logistics and technology teams, the challenge is to reduce manual work, improve data quality and keep visibility over how information moves across the supply chain.

AI can support faster decisions and more efficient workflows, but only when the underlying data is trusted. The organisations that build that foundation first will be better placed to turn AI experimentation into practical business outcomes.