Drone delivery in Ireland: the tech worked, but scaling it was another story.

At HumanX this week I heard the story of Manna Air Delivery, the Irish drone delivery company, and it's stayed with me.

It started with a craving. The founder was sitting in his garden in suburban Dublin one evening and wanted a bag of chips, but getting them delivered wasn't easy. The economics of suburban delivery can be difficult, so he decided to take the driver and the car out of the equation.

Manna started around 2018. After two years of building drones and getting approvals, they began delivering in a small Irish town at about 100 deliveries a day. Since then, they've completed around 380,000 deliveries, most of them in Dublin's suburbs. They've raised $110M and say each flight already covers its own cost.

This summer, Manna paused all deliveries in Ireland, the country where it was born.

The drones worked, aviation regulators had approved them, and customers were using them. The trouble came when scaling met the local environment. Residents raised concerns about the buzzing over their homes and about privacy, while local planning decisions made expanding the network of hubs increasingly difficult, particularly around noise. Manna, for its part, says Ireland has no clear national framework for scaling drone delivery.

We usually treat regulation as the big hurdle for deep tech, and Manna cleared the aviation part years ago. Winning over the people under the flight path turned out to be harder. The customers got coffee in three minutes, but everyone else on the street got the noise.

I don't see this as only a European problem. Drones, robotaxis and delivery robots all operate in shared space, so their "users" include everyone nearby, not just the person who ordered.

Manna is now betting on the US. It first tested the waters in 2023 in the Dallas/Fort Worth area, launching around Halloween by dropping candy to trick-or-treaters. Tulsa is the real bet: its first full-city rollout launched this summer.

I'll be watching what happens in Tulsa. For physical AI, adoption may depend as much on the people living alongside the technology as the people actually using it.