An easy, novel alternative
Customers liked the novelty, found it easy to use, and appreciated the option in hot weather or when they wanted less interaction.
noon minutes × Yango · 2025
Designing robot delivery for noon minutes
noon minutes’ first autonomous grocery delivery, piloted with Yango in select Dubai communities.
TL;DR
noon minutes partnered with Yango to pilot robot delivery in select Dubai communities. The goal was to learn whether autonomous delivery could work inside noon minutes, and what it would take to run it.
I designed the end-to-end customer experience, from choosing robot delivery in the cart to collecting groceries from the robot. The pilot shipped in 3 weeks: 27% of eligible orders went out by robot, and 95% of robot deliveries were completed.
Visible, not disruptive.
Robot delivery started in the cart, the page where most of noon minutes’ business decisions play out. For a pilot, that set a clear limit: keep familiar flows familiar, and only introduce something new where the robot changed what customers needed to know or do.
It couldn’t interrupt checkout, but it couldn’t be so subtle that customers missed it. It had to be visible, not disruptive.
Discovering the robot
With around 60% scroll depth on the cart, many customers would have found robot delivery on their own, but not all of them. Three decisions guided the rest without taking over checkout.
Make it discoverable
A cart nudge pointed customers to the delivery widget without taking over their checkout flow.
Make the choice clear
A dedicated widget separated robot and rider delivery, with clear visuals for each.
Depth on demand
An education card covered how it works, what to expect, and what customers would need to do.
Tracking the robot
Robots weren’t always available, and customers wouldn’t keep the app open. Tracking had to set expectations early and reach people outside the app.
Setting the fallback early
A robot wasn’t guaranteed, so we said a rider might deliver instead while the order was still being packed. A fallback felt expected, not broken.
Communicating actions
Once the robot was on its way, two things were highlighted: bring your phone, and come to the building gate. A moving robot on the map made it feel tangible.
Beyond the app
Push notifications and Live Activities kept customers updated as the robot approached and arrived, so they knew when to head down.
Collection
The most unfamiliar part of the journey, and where the hardware showed most. The lid takes seconds to open, there’s a fixed window to collect, and the app can only reflect what the robot does. Each step had to explain the machine before customers had to wonder.
Before the lid opens
Where to find the robot, the 10-minute collection window, and that the lid might take a few seconds to open.
While the hardware responds
After “Open lid”, a 5-second button animation reflected the physical delay and prevented repeated commands.
After collection
“I’ve collected my order” confirmed pickup and closed the lid. A cancellable grace period let customers recover from early taps.
From screens to the street.
The pilot went live in select Dubai communities. For the first time, customers could complete a noon minutes order through a robot, from checkout to collecting groceries at their building gate.
Early pilot signals
27%
Went out by robot
Share of eligible orders during the early pilot.
95%
Successful completion
Of actual robot deliveries, excluding rider fallbacks.
What we heard
Hardware and app weren’t always aligned
Some customers saw timing mismatches between the app and the robot, or robots stopping in unexpected places. That pointed to supply and backend logistics.
Reflections
The biggest milestone was making an unfamiliar delivery experience real in 3 weeks. If I did it again, I’d design for the moments the app and robot fall out of sync from day one, with clear states for a late robot or one that stops somewhere unexpected. And I’d talk to the customers who switched to a rider, to understand what held them back.