The reference demonstration for the delivery-robot engagement we ran through 2023 went like this.

A pharmacy on 92 Street in Dubai. The prospect places the order himself, through the client's interface. The rover comes to the pharmacy door and opens its hatch. Somebody closes it. The rover sets off, and the prospect follows it on a map link sent to his own phone. Sixty metres up the road it arrives, he presses a button on the link, the hatch opens, and the delivery is complete.

Nine minutes. No slide.

Roughly a third of the meetings in that engagement were demonstrations, and the demonstration was the sale. A pavement robot sounds like a gimmick in an email and stops sounding like one the moment somebody watches it wait politely at a kerb.

The question that reorganised the target list

Every conversation about autonomous delivery in this region hits the same wall in the first ten minutes. The technology works. The economics are arguable. Nobody in the room can say who gives permission for the machine to move.

Pavement robots cannot operate on public roads here. Which means the buyer is not really the restaurant chain or the pharmacy. It is whoever owns the ground: a master community, a mall, a business park, a university campus, a golf course.

So we stopped selling deliveries and started selling to landlords. The target list went from quick-commerce and logistics operators to those plus property: master developers, community managers, mall operators, and, unexpectedly productive, golf clubs with kilometres of private cart paths and a permanent shortage of staff to walk things around.

Ask the permission question in meeting one

Before a mall meeting our rep put it to the client plainly: are we asking this mall for permission to run the rover, for permission for the delivery operator, or for both.

That is the right question and the wrong time to be asking it is meeting four. Get it wrong and you spend a month with the correct proposal on the wrong desk, because the person who signs for a pilot on private land is rarely the person who buys the service that runs on it.

For any new category, the qualifying sequence is: who wants this, who pays for this, and who is allowed to permit it. Most sales training covers the first two.

What a demonstration actually costs

It is worth being honest about the other side of this. Demonstrations are expensive. Equipment has to travel, someone technical has to be there, and a failed demonstration is worse than no demonstration, because you have converted an open mind into a specific memory of a machine that stopped.

The economics only work if the demonstration is the closing step rather than the opening one. Our sequence became: qualify the permission chain, meet the sponsor, then bring the machine to the site where it would actually run, in front of the people who would have to live with it.

And a postscript that says something about the category rather than about us. Avride is now part of the Nebius group, and in 2025 Uber and Nebius announced an investment of up to 375 million dollars in the business. The rovers were not a gimmick. The Gulf just had not decided whose pavement they were allowed on yet.

book_a_call