[croatially]

Their logistics solution is just not as advanced as it can be. That's just common startup logic, the hardest problem will be solved with more money and good people and SV network has good people.

They need combinatorial search and some travel time data, live re-search when new orders come in. For this to work well they need to predict new orders in advance, this means they need data on the spatial distributions of orders through time (which now they certainly have).

They practically need to solve a variant of a problem that Uber has. Their delivery drivers load much more "passengers" but have static pickup points (known restaurants so there's only variance in the delivery point, pickups are pretty much static).

It's an interesting problem to solve, given that this kind of logistics research is practically invisible and I can't name any papers that dealt with this problem in any advanced way.

If done properly, then they've solved the logistics problem for on-demand companies that have enough data.

If solution is some ad-hoc mutated monster then it is just a chance for someone else to get the other pieces of cake.

[robk]

Your dismissal of a non SV company's innovation is embarrassing for you. The tech here is as good or better as anyone in the Valley.

[croatially]

Sorry if I was unclear, I had no intention to imply that their innovation is weak, or that SV has anything to do with that. It's pretty clear they have access to SV network, despite being based in London.

[ChickeNES]

When I left there a year ago there was not much innovation going on, just kowtowing to marketing/design.

[gwenzek]

> For this to work well they need to predict new orders in advance

Calls for deep learning! Or using data from last week…