Hacker News Reader

Comment by narrator | Hacker News Reader

I used to use Instacart, but it was a hassle to manage the people doing the shopping. They would get bad produce. They would mistake zucchini for cucumber. They would make nonsensical replacements. Constant screw-ups. Sure if you're buying toilet paper or some other generic commodity then it's fine, but you can already do that with Amazon.

View on HN

At the height of the Instacart craze in my city, I was at a grocery store (as a customer) when an instacart shopper walked up holding bok choy and frantically asked “Is this collard greens??”

This is one place the UK seems to be massively ahead - all the major supermarkets here offer delivery, and it's nearly flawless in my experience.

We just tried Instacart on a free trial; the only substitution was very reasonable and all produce was excellent. It will be interesting to see if that experience was anomalous.

I was trying out Instacart a few days ago and thinking whether I should purchase their Express program. This is what I realized:

- You pay them $150 a year. - On every order, they add a 10% service fee that helps them keep the service running. This is optional, but you have to remove it every time. - The delivery guy expects his tip.

I am wondering if this is cost-effective this way or not.

I haven't used them in ages so I didn't believe this. It seemed too ludicrous. I logged in to check and actually laughed out loud when I saw it. How much trouble must this company be in?

WF 2 hour delivery makes InstaCart toast not just because it will be a 2 hour delivery, but because pickers will be WF employees who are trained to know what is what and because one would not be ripped off via Instacart markups.

The next blow will be to BlueApron/Plated/HelloFresh/etc of the world when WF starts doing meal kits in 2 hours.

I believe a robo-picker would not mistake the clearly coded cucumber for zucchini, nor would the well-trained AI allow non-ideal produce into the sorting bins. No need to worry.

A robo picker would pick up a cucumber and not notice it's rotten on one side though, unless we assume that the robots will scan the entire item and somehow figure out what is good and what is not using AI - but I think that would prove to be an extremely difficult problem and ultimately not worth the effort.

I believe it would though. It's already happening higher up the supply chain where conveyor belts of produce are graded using computer vision (and other techniques, I'm sure).

This is about moving these techniques further down the chain. A robot picker arm with a few cameras attached that can tell whether something is rotten is well within our grasp. (Awful pun intended!)

Reply on news.ycombinator.com