Wonder how much of this is due to economics since computer vision tech never reached the expected performance + outsourced workers got (relatively) much more expensive after COVID.
[Disclaimer: Former Amazon employee and not involved with Go since 2016.]
I worked on the first iteration of Amazon Go in 2015/16 and can provide some context on the human oversight aspects.
The system incorporated human review in two primary capacities:
1. Low-confidence event resolution: A subset of customer interactions resulted in low-confidence classifications that were routed to human reviewers for verification. These events typically involved edge cases that were challenging for the automated systems to resolve definitively. The proportion of these events was expected to decrease over time as the models improved. This was my experience during my time with Go.
2. Training data generation: Human annotators played a significant role in labeling interactions for model training-- particularly when introducing new store fixtures or customer behaviors. For instance, when new equipment like coffee machines were added, the system would initially flag all related interactions for human annotation to build training datasets for those specific use cases. Of course, that results in a surge of humans needed for annotation while the data is collected.
Scaling from smaller grab-and-go formats to larger retail environments (Fresh, Whole Foods) would require expanded annotation efforts due to the increased complexity and variety of customer interactions in those settings.
This approach represents a fairly standard machine learning deployment pattern where human oversight serves both quality assurance and continuous improvement.
The news story is entertaining but it implies there was no working tech behind Amazon Go which just isn't true.
no idea how much they make on it, but it's a game changer in that small area.
If so, is the reason why it is not used related to cost?
If the customer is accidentally billed for an orange instead of a tangerine 1% of the time, the consumer probably won't notice or care, and as long as the errors aren't biased in favour of the shop, regulators and the taxman probably won't care either.
With that in mind, I suspect Amazon Go wasn't profitable due to poor execution not an inherently bad idea.
They trusted their tech enough to accept the false-positive rate, then worked to determine / validate their false positive rate with manual review, and iterate their models with the data.
From a consumer perspective the point is that you can "just walk out". They delivered that.
I think people just think that they must be misunderstanding something; that nobody could claim one thing while offering evidence of its opposite. 1/5 of purchases lose their significance.
Business Insider also reached out to Amazon at the time and a spokesperson denied that actually reviewed any transactions.
This "proven false" thing is just another anonymous source claiming that actually it was only 20%.
So you actually have no proof of anything, you just have three persons claiming three different things (0%, 20% and 70%).
[1] https://www.businessinsider.com/amazons-just-walk-out-actual...
It’s reasonable to expect a system like Amazon’s to use human feedback in training, and to quote the article linked on Wikipedia:
> Amazon said that the India-based team only assisted in training the model [and validating] a small minority of shopping visits.
It certainly felt like it could have been sent off to a lower paid country for a human to tot up.
Also consider you're in the store for what, 10 mins - that's a lot of video processing presumably using state of the art CV models. It's quite possibly cheaper to pay a human than rent the H100 to do it.
The supply/demand curve shifted and now those workers are becoming more expensive while domestic workers are becoming cheaper.
Scale is beyond comprehension though, there were 250 million people on strike one day last summer. This is not ever really covered in western media or mentioned on HN for reasons that are surely not interesting or worth pondering at all.
Theoretically if it was 99% computer and 1% human, that's enough to mess up the economics but it's not a bait and switch like some companies have done.
(Not to refute your point, of course, I am just curious)
Reports said the “AI” was largely 1000+ people in India watching the cameras.
If Amazon actually managed to build AI that worked well at a decent cost point it would have been great since nobody likes those silly self checkout machines.
What’s amusing about all of this is that before it got leaked that it was basically a bunch of people in India watching cameras Amazon folks spoke about the tech like there was some super secret AI they developed. Since that story broke nobody there seems to want to talk about “just walk out” anymore.
I have little trust that a corporate behemoth will do right by me and refund the discrepancy at an unspecified later time as it says it will on checkout.
This was totally fake news though. Those people were labeling training data and reviewing low confidence labels, after the fact. There wasn't ever live monitoring of shoppers.
¹ Individuals manning a labyrinthine system of cameras and sensor fusion, like hawks, logging the precise moment you plop a Twix into your basket! Praise Bezos!
"Do things that don't scale."
It was as if they expected everyone to know what to do, but when I’d watch 99% of people just sort of looked at the store, saw the odd gate things, and then just shrugged and walked off. The stores were almost always completely empty amidst a busy concourse.
Even if the tech worked (reports say it didn’t work well) they completely missed the boat on creating a clear customer experience that navigated the new tech.
If everything goes fine, self checkout is fine. But the exception handling process for any of those is thoroughly aggravating, as you wait and try to get the attention of the one overworked attendant dealing with a dozen of these machines constantly throwing exceptions, as the computer screams at you for whatever it thought you were doing wrong.
The issue with the first generation was that they were too strict with bag placement, weight sensors, etc. They were impossible to use without having to call a very grumpy shop attendant to unlock them. Sometimes multiple times. They were grumpy because all this was technically user error but when a largish percentage of users run into the same issues over and over again, it gets really annoying to deal with.
They fixed most of the glaring UX issues with the newer generation. No weight sensors. No prompts to put the item in the bag it is already in, etc. The new ones basically only need people to unlock things like alcohol purchases, but are otherwise fine. The first generation was over engineered and had way too many failure modes. They still have them in some super markets but they are getting replaced with better ones.
Anyway, it's getting harder and harder to hire staff for supermarkets. These are low wage jobs and most of these people can get better paying jobs. Self checkout creates some opportunities for shop lifting of course. But that is offset by the wage savings. They compensate with security, cameras, etc.
That's not to say the value of the convenience is never worth it. I exclusively use Sam's Club scan-and-go because the time I save is much larger than the publix/walmart/ect.
Optimus and Robotaxi are just as fake and Elon Musk never shuts up about them.
I guess Amazon never learned the important lesson that the OP meta for modern technology companies is just to consistently and blatantly lie.
I also found a loophole with the Amazon.com return grocery credit. The systems are separate for the $10 off $40 coupon and you just scan a QR code in the store to get it. It turns out you can just take a photo of their QR code and reuse it over and over again.
Wouldn't surprise me. I know a guy who invented a device for truckers that became ubiquitous in truck stops across the US. This would've been like 2014.
He refused to sell on Amazon, so Amazon duped his product and sold it at something crazy, like half price, until he agreed to list (at which point they dropped their competing product)
iirc that's exactly what Amazon did to destroy diapers.com over a decade ago
Phase 1: bankrupt the competition
Phase 2: ???
Phase 3: profit!
Wegmans is popular because Wegmansnis good. But if you have a local baker, a local grocer, a local deli, and a small grocery store within the same block, all within walking distance of your apartment, you don’t need to deal with the hassles of finding stuff within a massive supermarket.
You get the highest quality products from people who specialize in those products.
Further, when you don’t have to drive 20-30 mins to go to a grocery store but the stores you need are within a 5 min walk, or more likely, right by the subway exit when you’re returning from work, you buy stuff as you need it, rather than stocking up for days.
Thats why Wegmans opened a store in Brooklyn Navy Yards in an area that’s close to no mass transit, because supermarkets are valuable in car centric areas and not as useful in walkable dense neighborhood.
It was a good grocery store with decent produce, a good frozen section, some nice specialty items, and some decent prepared meals. I would put it at roughly the early-2010s era of Whole Foods with slightly better prices. Now that I'm no longer working near there, I don't miss it much.
So I've never understood the hype. But I've also been told that the Boston stores were pretty mediocre compared to the ones in NY and especially Ithaca.
Amazon Go, on the other hand, always seemed like a dead man walking. It's a fun novelty to check out and grab some junk food, but it must be far more expensive to build and run than a 7-Eleven, and it's not even meaningfully more convenient.
I should also add that we've been pretty happy Amazon Fresh delivery customers for a couple of years now (we resisted regular grocery delivery for a long time...until we had a child).
I'm going to miss those. Two nice things about them compared to a normal self-checkout: 1) you see things ring up as you shop instead of at the end, which is nice in case of errors or unexpected prices, 2) you can shop directly into a reusable bag or backpack instead of repacking everything at the end.
The prices are indeed pretty insane and the produce is always great, but the stores are ghost towns most of the time. The only people inside are those using it as a spot to drop off Amazon.com returns and those fulfilling pick-up orders
To be fair I've noticed this in multiple supermarket chains the last few years. Although they aren't usually employees, they are instacart runners or whatever.
I go fairly often to a Sprouts grocery store and there are times I need to avoid multiple people clearly doing an Instacart run with 2+ carts full of items.
Shelves are often emptier than they used to be also at these times.
This was a bit before the model of having Uber driver type delivery though. I am guessing that having the deliverers be close to the deliverees make it more economical to keep them in stores until a larger scale is reached. The dark store+ model was also predicated on a more factory floor like environment with only FTEs present. Think pallets moving about among the pickers- not too hard to work around IMHO but maybe the lawyers and insurers feel differently.
I still feel the overreaching factor is that in dense urban centers there is no cheap commercial/industrial space that is also in close proximity to customers.
That's cool. Which one? The cheapest one in Russia costs about $1.20 [0]
> a photo of their QR code and reuse it over and over again
Don't you think it's a wrong thing to do?
[0] https://5ka.ru/product/makarony-barilla-dzhirandole-n-34-450...
I moved away from Seattle a while back so I'm not sure if they ever closed that one. I really miss getting all those cheap groceries!
I guess Amazon pulled out of the project halfway through, since for the last ~2 years there's been a half-finished building just sitting there completely abandoned in our town center.
My hope is that more towns learn from your experience and don't tolerate this nonsense anymore.
What's the Matter with Kansas? (2004) by Thomas Frank goes into this in part of it. While a bit repetitive (because history) the book is quite good.
fyi since you may not have ever seen it spelled before it's milquetoast
The state and localities did a good job structuring incentives so everything was tied to milestones, some of which Amazon ended up not hitting. My recollection is that NY was offering to give a lot more away, which helped fuel the backlash, but don’t quote me on that.
It also helped in VA that Crystal City has been a commercial wasteland for many years thanks to DoD decentralization. A lot of offices moved down to Fort Belvoir and surrounding areas, leaving Crystal City with a lot of vacant office space. Nothing was being “lost” by Amazon coming in.
Local real estate agents put “HQ2” in their listings for a few years but it didn’t matter much because homes near Crystal City were already super expensive.
Given that a supermarket abandoned that location, and Amazon never opened on their either, perhaps that location or the lease price simply doesn’t work for a grocery store?
Wondering what the municipality’s responsibility there wrt zoning.
China has had 24/7 McDonald’s since forever, well, and lord of other things as well. But not grocery stores at least.
Ordinal : Cardinal
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5This would not be the first off-by-one error at Amazon.
They often put them in places, hoping that people would be attracted by marginally lower prices and brand extension, all while removing one of the primary appeals (for most people) of in person grocery shopping: impromptu community socialization, even if it is simply greeting the checkout worker.
I'm not surprised they failed.
And because they were relying on computer vision and Indian vision, they had to get rid of all their fresh meals because they were too hard to calculate prices for. So, it ended up being a half-assed 7-Eleven concept. The whole concept was made by someone who hates humanity.
I personally prefer stores with actual cashiers. What I don't like are lines, but that is very solvable. The organic grocer near me is super fast to check out.
I've gotten into the habit of stopping in to wander the aisles and check prices because of it (e.g. I stocked up on a bunch of canned soup when most (but not all) Progresso soups were $0.44 a month or two back, and I picked up some microwavable rice+quinoa pouches for my wife at $0.35 each a couple weeks ago, but the inconsistency and overall not great prices mean it can't be my go-to grocery destination.
I'm sure the one by me will be closing since there's a significantly larger Whole Foods just a few miles away.
That's one way to spin things I guess.
Edit: oh oops I see this is about physical fresh stores, we never had those in the first place. Here in Europe Amazon fresh is a weird service for quick small grocery orders. For the bigger ones they partner with a local supermarket ("dia" here in Spain). But I never do grocery delivery because I never make any plans, I just make my life up as I go along :)
But Amazon fresh here is expensive and still slow (2hrs) so really not good for anything.
Amazon go I'm not even sure what that is.
And now you don't have to!
Ba-dump-ching! I'll be here all week, folks! :)
I actually did find it saved me time. I would go in, grab a couple things, and leave, and it was actually a good experience to do that. I never did full grocery store runs there, though.
The aisles were always packed with workers picking/packing orders, which was frustrating to deal with.
One thing that was bad about it was that produce was all fixed price. At a normal store you pay per pound for an onion, but there every onion was $1 (or whatever the price was). Giant onion, tiny onion, all the same price. The produce got picked over in weird ways because of that.
Then one day they said, "okay, Just Walk Out is gone, it's just a normal grocery store now." Then it just became just a mediocre grocery store. There were definitely periods where aisles could be nearly empty, but lately it's been okay. Prices were great, though -- by far the cheapest in the area.
Their hot bar was extremely mediocre. I like the Whole Foods one, but theirs was just... not good. Half the time they didn't even have it stocked with food.
They had a little stand up front where kids could get a free piece of fruit, which mine liked.
It also had convenient returns for Amazon purchases, which was about half of what I went there for.
It was a convenient place for me, and I like it better than Safeway, but I can't say I'm too heartbroken that it's going away. QFC/Sprouts/Town&Country/Safeway are a few minutes in any direction, but they're more expensive. I doubt they'll turn this one into a Whole Foods either.
I remember WAY back in the day when Arby's implemented touch screen ordering (on CRTs!) and it was a very quirky process. An Arby's employee would sit behind the counter and stare at you while you spent 5 minutes poking a CRT display. Very slow and very impersonal. They discontinued them in a short period of time.
The Go stores were a great experience but they would certainly be uncomfortable for anyone other than early-adopter or tech-forward types of people. I would just walk in with my own bag, and put items directly from the shelf into the bag, and walk out the door. It was extremely convenient and fast once you got over how weird it felt.
I think they could have done a lot more in giving social clues on both the way in and way out.
The main thing I use it for is convenient returns, which is why I’m disappointed in this news. I hardly ever buy things there other than things like gum or chips.
I went there the first week they opened, and the whole store was a mess with shoppers standing still or walking slowly, completely unaware of their surrounds, while messing with the phone or their cart trying to figure it out.
I'm sure that with enough time, shoppers could figure out there system, but I was in a hurry so I just grabbed the few items I wanted and paid cash, which was just as fast as it is everywhere else.
I do want to stop by before it closes, and see if customers figured out their systems, in the year and a half the location has been open.
It's convenient; I only ever remember one problem where it thought I had purchased an item that I picked up and decided on something else. I disputed it online and it was resolved in a day.
one minor downside (especially since I don't live on campus anymore) is that in order to walk around and peruse the shelves, I have to give them my payment info just to enter
So no I didn't feel like I was a thief. But I felt like they assumed I was a thief. My guess is most stores are heavily surveilled nowadays, so it might be unreasonable for me to feel this way with Amazon but not Walmart or Target or Kroger, but that's how it felt.
Store was kind of bare, and poorly organized. But the kicker is they didn't accept any form of mobile wallet! They had an identical POS system to wholefoods which takes it just fine.
So I quickly put my items back and headed to Giant.. Haven't been back since
So even though these stores are closing, the tech is widely used and likely expanding and succeeding
I knew someone who worked at Amazon UK and they told me years ago they were doing very baldy and there was talk of them closing.
So I'm not surprised to hear this at all.
So you spend a few hundred thousand dollars extra on all the cameras, many millions on all the design, pay all the overseas contractors to manually review the transactions, and you still end up with twice the in-person staff than the average store in the airport.
Anyway, I didn't go back after the first time because it was more like a corner store than a grocery store. Bags of chips and sandwiches in plastic boxes and so on. Overall, the modern Whole Foods is a much better experience. Guards at the entrance to keep unpleasant people out, a fairly quick check-out experience, and the ability to scan your palm instead of having to pull out a credit card or tap your watch.
About the only improvement that I would personally like is a Fast Shopper bonus where you scan something that maps you to your Amazon Prime profile and if you finish checking out fast you get access to a faster lane. The only downside is when people with large bags of things insist on using the self check-out counters and then stand there having mis-scanned items.
Speeding up check-out is a personal life improvement but realistically it would not cause me to shop more, so I understand discontinuing the store.
It's a welcome change IMO, amazon groceries are super cheap online and now delivery is free. They have been removing the fresh name from products for a few months now and replacing with amazon grocery. Certainly less confusion for consumers, at least
Related: https://www.aboutamazon.com/news/retail/amazon-same-day-fres... https://www.aboutamazon.com/news/retail/amazon-same-day-fres...
They resisted implementing self checkout for years before eventually folding. No digital wallets though, you have to either use plastic or link it to your Amazon account.
The whole dash cart system was a solution in search of a problem IMO. I'm already able to check out about as efficiently as possible. Frontloading the scanning time isn't really an amazing improvement. The store was never crowded enough for it to matter.
My biggest problem with the store was that it was lacking random pantry staples and supplies that you would expect from your primary grocer. Several times I showed up in desperate need of something for a recipe or household task and they just wouldn't have it.
The produce was actually decent quality and competitively priced, but my alternative (the local Ralph's) I think just had some kind of curse or something on it because the produce at that specific location was a consistent level of awful observed over 5 years.
I hope they replace it with a whole foods, much better store IMO.
I suspect that they wanted to take a hail marry to see if somehow it was possible to get much greater efficiency compared to standard grocers, and it looks like that failed.
[1] it may come back. The technology is rapidly improving but they have bigger fish to fry ATM.
After covid, it was never the same. Open for shorter windows, closed on Sundays, reduced selection, no more meal kits etc.
I had many friends who worked on Amazon Go, so it's a bit sad to see that work come to an end.
What did they do?
Actually knowing Amazon they'll give the staff a huge gimme and nice bonuses. People crap on Amazon but they are a good employe for the little guy. They can't rewrite society so go for that take. Otherwise people are bummed they lost good jobs.
The interesting question isn't whether the tech was ready. It wasn't. The question is whether Amazon learned anything useful from the attempt.
Computer vision for retail checkout is a legitimate hard problem. Occlusion, similar-looking products, people changing their minds. I've worked on CV pipelines and the gap between "works in the demo" and "works at scale" is brutal.
My guess: they collected a ton of training data from those human reviewers. Whether they'll use it for a v2 or just write it off, who knows.
Knowing this, it was over before it ever started. Beyond the masses of people already having aversion to the oddness of how it worked and likely never wanting to try it, these and others would swear off the store forever the first time they ever got charged for something they didn't take. No one wants to monitor and fix erroneous purchase errors.
> We're reaching out because you have an active Amazon One account. Amazon One palm authentication services will be discontinued at retail businesses on June 3, 2026. You can continue using Amazon One at participating locations until that date.
> Amazon will automatically delete Amazon One user data, including palm data. No action is needed from you.
Now it seems Amazon is going to leave us a grocery desert yet again.
They were piloting smart carts at the location. The cart scans your items so checking out you just push the cart through a scanner that weighs it. But this invention was like a microcosm of Amazon’s whole fuckup with groceries. The problem with the store wasn’t that I couldn’t check out fast enough, it’s that it was a shit grocery store. They had popular products but they were missing all the unpopular, low margin products you need to actually cook (baking powder, shortening, tomato paste, soy sauce…). They only hire non-union jobs program people at the registers because Amazon believes that cashier is a sub-human role.
The previous store had an owner who would wander the aisles and chat with customers. The new store has Europeans with clipboards who watch you as you shop.
What grocery stores still have union workers?
For instance, the one in north Seattle (Shoreline) is within eyesight of a Safeway, a Sprouts, two international markets and a chef wholesaler.
The other three locations are similarly crowded with options.
What food desert are you referring to?
The implication being that humans who aren't in a union are "sub-human" in your opinion? If so, that's pretty messed up man.
I know at some point they got caught basically paying people to watch cameras to figure out what products people we're grabbing. I'm sure were either at the point or very close to the point where AI can successfully do this basically 100% of the time.
So I doubt it's the tech aspect of this, more just the grossness a person feels walking into a store with Amazon's name on it. Compare this to whole foods.
The Fresh stores are basically a conventional grocery store, with electronic tags for every item and quirky pricing. They also have "smart carts" with built in weight sensing and multiple cameras so you can basically put open bags in, say "ready to go" then shop by scanning a UPC before placing each item in the cart. Unscanned item? Error. Weight mismatch? Probably an error but I've never tried. The carts are running what looks like a Linux-based UI with some stuff in docker, I grabbed a picture of a shutdown screen on one not too long ago.
I guess it turned out that the need more human intervention than they hoped, so the cost is too high for regular stores. However at places where a premium can be charged for high throughput or a low friction experience then the cost of the human intervention can be recouped.
Absolutely zero interest in a physical version that lets me check-out easier, though. So, I can see why they're making this switch.
One of a my previous jobs had a grocery store on the way home. I took to stopping in pretty much daily. It allowed for a bit of decompression after work before coming home. It was very convenient to always have exactly what was needed for that night while being therapeutic at the same time. After switching jobs, losing that was probably the most noticeable thing about the new job
I will miss the grab and go tech in the Amazon stores. I was hoping they would successfully manage to sell that to other stores and make the tech wide spread in bodegas, gas stations and 711s
Amazon straight up wants to just become Walmart. Or maybe Sears is a more apt comparison given their mail-order beginnings.
Edit: it actually opened in August, so it was around for about six months instead of two.
All in all, it's a cool concept on paper with absolutely terrible execution.
Only went once, bought some snacks, and left.
In a world where anonymity is a thing there will always be at least one inherent shithead who ruins it for everyone. Even if you do have a community where it's true, that can change anytime someone has a kid or someone moves in.
I also know some Amazon warehouses had an entire Just Walk Out powered concessions area in their breakroom for purchasing snacks in partnership with one of their canteen vendors.
Been in one once for the novelty as they've never been useful.
We are closed
for the
remainder of
the day.
We apologize for any
inconvenience. Please come
back tomorrow during our
normal business hours.They almost seemed like an extension of Whole Foods to a more mainstream suburban market, and I thought they had solid foot traffic.
I show up at CostCo, on weekdays, like 30 minutes before closing time and it's _wonderful_. Few people, nobody blocking lanes while they consider their choices, etc. Same goes for Safeway, Fred Meyer, Trader Joe's, etc.
It doesn't work so great if you've got young kids, or you want to come home from work and just stay home (reasonable), but it's worth considering :)
But a 24/7 data driven store that is a last resort would have succeeded.
Put slop bowls at business areas, Junk food where there are clubs for drunk people.
Moving season in college town, put some stationary and some tools for DIY stuff.
In valentines cheap wine and chocolate for the singles, and flowers with top rated gifts for couples.
I honestly think in some ways, going to a store is about being around other people, the same as going to a cafe, not necessarily talking, but just being in the presence of others seems to be what many people crave. I largely think it's the appeal of shopping malls.
However, I love my local Amazon Fresh store because it's a super convenient Amazon return location...