Its September research moves the conversation from pilots to operational deployment across four layers, from optimisation and generative AI to semiautonomous agents and physical AI in picking and packing.
For vendors, generic AI claims will not cut it. Buyers will ask which proven use case comes first — labour forecasting, slotting — how decisions stay transparent, and where human oversight remains.
Vendors should rebuild their pitch around sequence and evidence. Lead with the first deployment — the specific use case, the site profile it fits and the operational result — and publish a credible path from that beachhead to the wider operation, because buyers are now evaluating rollout risk as much as capability. Transparency is a feature to market, not a compliance footnote: show how recommendations can be inspected and where a human can override. Sales discovery should start with labour constraints and the buyer’s existing automation layer, then map the first use case to it. A vendor still selling a pilot is answering a question buyers stopped asking.
