A coffee shop is 33% drinks and snacks, 33% theatre and experience and 33% location and/or pleasant lounging space and such.

Maybe at the best end, a barista makes a better cuppa than a good machine but most of the places near where I am now wouldn't pass a blind test Vs the automated coffee machine at my office.

Automated coffee-making exist, and works just fine.

Still, €1 vending machines haven't killed the €3 coffee shops/stands/trucks. Empirically, people want the (subjective) experience, not cheap coffee.

This is a novel, theatrical way of doing the experience.

Besides those relatively banal points... it's kind of cool that they're putting robotic arms into public-facing consumer land. Not sure it leads anywhere, but I'd get a cuppa and watch it work as I sip.

This.

I keep making the point every time it gets raised in a "the robots are coming for our jobs" thread, but we had the technology to vend fast food to people putting coins in slots before we had McDonalds etc, whose laser focus on margins proved the business case for counter staff and killed off most of the Automats. Needless to say, the perception of a pleasant coffee shop offering good service worth paying extra for and superior quality coffee is more intrinsically linked to its staff than the fast food industry.

In terms of theatrics and novelty this super-expensive machine is a massive improvement on the vending machine and can probably justify a $3 coffee instead of a $1 coffee, but it isn't going to kill the traditional coffee shop either, because the novelty is directly proportional to how rare they are....

> McDonalds etc, whose laser focus on margins proved the business case for counter staff and killed off most of the Automats

McDonalds is returning to vending machines - not for the burger manufacturing, but for the order generation and payment:

https://eu.usatoday.com/story/money/nation-now/2018/06/07/mc...

in selected stores, in addition to the staff, some of whom McDonalds has concluded it makes financial sense to redeploy providing table service. And this despite McDonalds being pretty much the extreme outlier in the food industry for perceptions of its staff being charmless! Your article also points to a poll that suggests the majority of people are substantially less likely to buy from a kiosk-only restaurant, which I'm sure McDonalds has more scientific data on...

Actually, in Japan vending machine coffee is extremely popular, moreso than served coffee.

You can buy all kinds of weird things in vending machines in Japan: hot food, underwear, etc.

I think the robot might improve on the experience by not correcting me when I order a large.

In my experience, most vending machines produce far inferior drinks to most coffee shops. If they could match the quality I would happily get my drinks from them, whether they're in a coffee shop or not.

What's the other 1%?

A good commercial espresso machine of the size and type they're using can cost $15,000 and more, by the way. E.g. https://upscalecoffee.com/products/la-cimbali-s30-perfect-to...

I'm sure it can be done cheaper, but not $200 cheap.

Yeah espresso machines are nuts. How has YC not yet funded a coffee machine startup to "democratize" upscale espresso?

Kickstarter is a graveyard of many overhyped and then under-delivered (or non-delivered) espresso machines.

The problem is that Nespresso already exists, and is good for 95% of the market. The rest of the 5% of espresso aficionados demand tons of features that are not cheap to produce, and the market is inherently not that big. There are machines like https://decentespresso.com/overview (I believe they picked up the IP of an infamously undelivered Kickstarter) and https://www.breville.com/us/en/products/espresso/bes980.html

Yeah, they did pick up the IP of one of the most disastrous kick-starters ever.

It seems plausible, you know, that someone could make a machine like that, but at the end of the day it's all about the espresso head and quality of the build.

By the time they figured it out, I think, they were no better than well-established prosumer models in price and performance.

It wouldn't be "upscale espresso" if the machine didn't cost 10,000$.

If you look into the audiophile community, there are people that have gone crazy spending absurd amounts of money on equipment that's basically a placebo. It fails on blind tests all the time.

I bet it's the same with this coffee.

Do you have any hobbies?

I do, actually.

Luckily, none of them involve spending thousands of dollars on ultra-premium equipment whose advantages are completely imaginary.

I understand how an espresso machine works. If somebody is charging you five figures for one, they're trying to pull a fast one.

I have the impression you have been viewing this from a consumer point of view.

These are not for your kitchen, they are machines for commercial use. Even the fully manual ones easily cost mid-thousands.

I understand how they can cost that much. They're made in small numbers, by some artisans in Italy, presumably with the best of materials and to the highest standard. You can put an arbitrary amount of effort into anything.

The question is, is the end result coming out of that machine measurably better than what comes out of a no-frills machine?

My bet is that it isn't.

But if it can produce it faster, more reliably, in greater volumes, and with industrial grade parts to extend its useful life ...?

It's not just the coffee that's "the end result."

You should start making and selling espresso machines then, since you know more than coffee professionals about them. Same with high-end audio gear.

Do you also go to museums and say "I could have done that" to every painting?

> You should start making and selling espresso machines then, since you know more than coffee professionals about them.

That doesn't really follow. If I think five-figure coffee makers that perform a rather simple mechanical function are overpriced, I should start selling them myself?

> Same with high-end audio gear.

I'm talking about equipment whose qualities cannot be measured nor perceived in a double-blind study. Pure make-belief:

"Audiophile publications frequently describe differences in quality which are not detected by standard audio system measurements and double blind testing, claiming that they perceive differences in audio quality which cannot be measured by current instrumentation,[26] and cannot be detected by listeners if listening conditions are controlled,[27] but without providing an explanation for those claims."

https://en.wikipedia.org/wiki/Audiophile#Controversies

> Do you also go to museums and say "I could have done that" to every painting?

No, I've got better things to do. Having said that, modern art certainly is rife with pretense.

I think you are overlooking one important aspect with espresso machines. You can get much less expensive machine that will create same tasting cup. But only if you do few cups a day not few hundreds. They are expensive because they are heavy duty, they perform and you dont have to clean them as much because they are made with much higher precision and they last longer. This is highly valuable - you dont want you main source of income to die on you in middle of the day.

In the long run price difference is well worth it.

Really? Why are the espresso machines used in coffee shops so expensive then?

If ground-up rhino horn isn't actually a working aphrodisiac, how come these animals are being hunted towards extinction in parts of Asia?

Same answer: People are desperate to be fooled.

You might enjoy this article about "the Keurig model" for hardware startups [1]. I'd be surprised if Blue Bottle isn't tinkering with their own less wasteful version of a Nespresso. They have the perfect branding and reputation already for this kind of thing.

[1] https://medium.com/@BenEinstein/keurig-accidentally-created-...

I think that already exists in the form of Nespresso. I have one in my kitchen and I prefer it to what I get at a coffee place minus the experience.

The lower end of automatic espresso machines is quite affordable. You can buy a new DeLonghi Magnifica ESAM 4200 for under 300€ and with the right settings and good beans that'll make better coffee than you can get at Starbucks.

Better coffee than at Starbucks is extremely low bar. Any espresso machine with good quality beans will do better than starbucks because they use low grade beans (and mostly roasted to death). Unless US starbucks is somehow very differen from the european one.

You don't need to spend thousands of dollars to get a solid espresso machine for your home. I have a $400 Breville machine and it does everything I need. Most people probably wouldn't be able to tell the difference between my results and a coffee shop, except the latte art, mine isn't very good :)

When you start getting into the $1000+ range, it seems like you're mostly paying for the following:

1) more precise temperature control (PID control)

This is becoming less of an issue because the heat exchange systems on consumer devices are becoming a lot better. My machine is ready to go in under a minute, although they recommend you wait about 20 minutes for it to warm up. Some of the higher-end machines with a dual boiler even recommend a 45-minute warm-up time, though.

2) ability to steam milk and pull shots at the same time (double boiler)

This seems to be a big differentiator because then the dual boiler comes into play (dedicated boiler for espresso, dedicated boiler for steam wand). On my device, I simply pull my shot, then steam my milk after. This all happens in under 1.5 minutes. I simply don't need to do them at the same time, but I can understand why you would want this feature in a commercial machine where you need to be moving fast and serving a lot of customers at a business.

3) milk steamer is often quicker on $1000 machines (a few seconds instead of ~15-30 like on mine)

Again, we are talking about maybe 15 seconds of time saving here.

4) water-line hookup so you don't need to refill a water tank

My machine has a water tank, but I only need to refill it every two or three days. I usually have two or three double espressos each day, one or two of those are lattes. So the water hook-up is of no use to me.

5) fancy styling / italian design & looks

I think my machine has decent styling, but it doesn't look nearly as high quality as something like the Rocket machines: https://rocket-espresso.com/.

I researched this heavily before buying, and watched a lot of videos here: https://www.youtube.com/user/SeattleCoffeeGear. I was very close to buying a Rocket machine and it would have set me back well over $1000. Anyway, that youtube channel sometimes does blind taste tests between hundred dollar machines and ones in the thousands of dollars. They often cannot tell the difference.

This is a great one: $300 vs $3000 machine: https://www.youtube.com/watch?v=zuIOlsdDg_o

Agreed. The duty-cycle rating for the commercial machines is much higher and typically are field serviceable. $15k is still a bit steep in my opinion, based on peeking inside a few of those machines. I'm nosy that way.

Precision. They dont get dirty as often which is a problem. Consistency of pressure and heat that you can make every cup same quality. Longevity - if machine lasts 10 years instead of 3 than its big. You can find cafes with la marzocco machines from like 60s.

In fact, the first machine I looked inside of was from the late 70's. It was certainly built to last. I think the company that made it was out of business which is why we were tinkering with it in the first place.

>This is literally a $200 coffee vending machine with a $25,000$ arm to move the cup.

Having watched this (https://www.businessinsider.com/robot-barista-reveals-danger...) I tend to agree with you. Coffee making is a business of diminishing returns and quickly rising costs. You can do an acceptable espresso with a $200 espresso machine. However to make an amazing cappuccino/latte you need to go to at least $5000-$6000 price point. Separate boilers, precise pressure monitoring and all that jazz. The reason for that is that steaming milk is not a very easy job for a robot.

I agree with your larger point about diminishing returns but I’d have to disagree with the price point of making an “amazing” espresso or latte. My $800 home espresso setup makes drinks that are at least on-par with the local 3rd wave shops (although my latte art is pretty bland). To me, the reason for those very expensive latte machines at the shops is volume (pulling multiple espresso’s simultaneously) and durability— my gaggia definitely couldn’t handle hundreds of drinks per day.

Do you mind sharing which one you have?

Sure, its the Gaggia Classic. I bought it refurb a few years ago and it's been very solid. I modded the steam wand, which is done a lot and I would recommend if you go this route. Here's a tutorial:

http://www.gaggiausersgroup.com/index.php?topic=1484.0

Remember that a proper grinder is at least as important as a decent espresso machine. I've been through a few but have finally found the excellent Rancilio Rocky Doserless, which was also refurbished.

A few other accessories like a bottomless portafilter, a knock box, and a milk pitcher for steaming, some good beans and you're pretty much set.

I think my whole setup was less than $600 and I have never regretted it. It takes some time to dial in your grind but you'll be beating your local barista's shots within a few months (and at a significant discount).

I have to say I'm surprised at the accuracy at the title. It's indeed a coffee serving robot. It just takes it from the coffee machine and put's it on the counter.

Somehow I imagined the robot going through the classic steps you might see a barista perform: grounding coffee and putting it in the portafilter, tamping, using the steam wand, etc. The things that might make the difference between a handmade coffee and one from a standard coffee machine.

But why go the antropomorphic way of getting a robotized hand to copy the steps the barista makes ?

It usually makes more sense to make a coffee machine that does those steps internally. It's easier in the controlled environment internal to a machine. A espresso coffee vending machine seems to be doing some of them already, like grinding the coffee.

All it's lacking is a cup transport system like a conveyor belt to put the cup in from a cup stockpile and to later serve it. You could even automate recovering the cup, cleaning it and restoking it.

Is there strong customer preference for a robot arm serving the cup rather than a conveyor belt ?

> But why go the anthropomorphic way of getting a robotized hand to copy the steps the barista makes ?

I'd guess for the wow effect. Such a system is much harder to calibrate and maintain than a regular automation line. But a lot sexier also. Maybe the perception and experience of getting the cup from a robotic arm is more appealing to customers than getting it from a conveyor belt?

We went to a bar with an anthropomorphic robot bartender and it was fun to watch, especially since it “bantered” with the customers: it had a few jokes, did a little dance when certain music came on, and so on.

It's only a $25k arm if it doesn't need regular parts replacements, which it probably will.

In my opinion, in order for robots to be mass profitable they either need to be faster, better or run longer. This machine seems to do none of those things. Novelty factor only has value whilst it's novel.

That's a Mitsubishi RV4FR or RV7FR series arm[1]. It's a full on industrial arm, not an in-house solution. Industrial arms are some of the most well engineered devices on the planet; they are expected to last decades of nonstop, maintenance-free operation in factory conditions. They're supposed to go hundreds of thousands of hours without performance problems, while holding .005" precision (for this robot, +/- .0008"). This arm costs as much as a nice car because it has the absolute best components that can be built.

If it's the M class version then it's rated to IP67[2], and it can do all of the above while underwater or in a sandstorm.

[1]: https://us.mitsubishielectric.com/fa/en/products/industrial-...

[2]: http://www.dsmt.com/resources/ip-rating-chart/

RV4FR and RV7FR are IP40 in the manual? That would mean that it's not protected against dust and has zero protection from water.

In a controlled environment this is perfectly fine, where dust and moisture can be controlled - but we're talking about moisture from steam coming off of the coffee, the possibility of coffee spillage on the machine, organic dust from nearby humans, non-clean power supplies and loads which may fall outside of usage rating (i.e. obstacles or heavy loads or continued use close to its maximum).

Honestly I would be surprised if it's still working in a year in these kinds of conditions.

Another issue with smart motors that generally isn't discussed is that they wear out (can only handle lower loads) and lose accuracy. I've used magnetic encoded smart motors for a few projects and after two years you can't rely on them anymore - and this is easy work for the servos (occasional usage).

This seems like a classic case of massive overkill. The cheap solution would be to not have the arm at all, and just put the $200 coffee machine out for customers to operate themselves. It's not hard to put a cup in something and press a button.

120 cups an hour is one every 30 seconds, I'm not sure a barrista could keep that up, if they could actually reach that speed to begin with. I'm not sure what you mean by "run longer", it wouldn't get tired if that's what you mean.

Agreed its probably a novelty though.

Starbucks with a staff of 5-7 during peak mornings have a goal of ~40 second drive through times which they routinely achieve, while handling non-drive through customer load as well.

Also drive through includes multiple drinks and food items as well crammed into that 40 or so seconds.

When timing the serial steps of making a drink, it takes about a minute from start to finish, but their process is to batch and pipeline drinks so that they generally handle 3-5x drinks per batch in a pipeline (repeatable beverage routine), and this is for a lone barista pumping out hot beverages. The other 4-5 people are usually on cashier duty, and drive through duty.

Have you got any links. The research I've done suggests Starbucks are the slowest [1].

I haven't been able to find any figures excluding queuing, but these figures would seem to cast doubt. Unless Dunkin Donuts are doing the same orders in <30s.

I can assure you that a barista in Italy can easily reach that number. And do many other things in between.

(In the Netherlands it's probably three cups an hour, with looking at the ceiling or the floor in between).

Possibly, depends on the drinks.

Espressos? probably. Large lattes? Not so much. Steaming the milk would kill the throughput too much.

I would guess that the Italian would do a much higher proportion of shorter and smaller drinks.

Not too sure about the "large latte", which in Italy might only translate to "cappuccino" (and it's not that large). Anyway, apparently two baristas from Milan have the world record of 623 espresso cups/ hour. :)

I believe a "large latte" is more similar to what in Italy is a "latte macchiato" (though as you say is not very "large"[1]) than to a "cappuccino".

[1] "normal" bars in Italy only have two sizes of cups, one (smaller) for the espresso and one (larger) for the cappuccino (used also for tea, chocolate, ginseng, etc.). Large mug-sized cups normally exists only in homes where traditionally you have a caffelatte with bread or biscuits at breakfast (or sometimes instead of dinner)

Yeah they could save $25K by just putting down a coffee vending machine. Like at truck stops.

I don't get the "they took our jobs" vibe from this video. Especially considering these things probably come with support contracts (and troubleshooting, maintenance, etc) that cost more than an employee, because it's specialized hardware.

To be fair, it looks like a commercial espresso machine not unlike what Starbucks would use. Those alone are well above $25K.

The 25K is likely just the robot arm itself, and not even the espresso machine, grinder, enclosure, payment system, and bean hopper/dispenser.

Real commercial espresso machines are still a very manual process for many reasons. This is something you’d see in the micro kitchen of an office or an airport lounge, not a Starbucks.

I disagree. Starbucks and similar needs to be automated, at least partially, while maintaining the premium it charges on a per cup basis. That means that it would need to keep the traditional expresso machines and process that people believe they are paying for -- if they just replace those with large blockly traditional coffee machine people will stop going to Starbucks because they are no longer a premium differentiated solution.

You need to keep the premium that Starbucks customers pay for and that is using traditional methods with lots of customization options. Making coffee at Starbucks is sort of a performance, and that is remaining because that is what people are paying for.

You are misunderstanding Starbuck and the premium coffee experience completely.

I think that something like this will succeed at the medium-high end. But "artisan" premium coffee that is hand made at an espresso machine will remain but it will likely get even more expensive.

Remember that we have already automated pretty fully the low end coffee market with Nespresso, Kurig, etc.

I think you are completely wrong, there is a market for this.

all those people who ordered using Starbucks app and walk in hoping to see their cup ready for pickup absolutely don't care about the 'performance' or which machine was used to make it. some people just want their coffee and gtfo of there

Indeed people trust the brand. As long as it keeps tasting the way it does and meets expectations, you could have it coming out of tubes from some automated mini factory in the back and 9 out of 10 wouldn’t care. They just want their drink.

Not to make it equivalent but in Japan and other places people get their coffee from cans. All they want is a little pick me up routine in the morning or whenever.

The majority of your argument seems to be Starbucks provides a sort of performance art, which can't be replaced by machines and, whilst partially true, the general idea of automation is "fire and forget".

I pay €29K (gotta love the exchange rate, right now) for a machine which will replace a worker that I probably pay for year the same service. Sure it's not "artisanal" but, given enough time to tweak the machine's behaviours, it could be.

For the lower-end markets, definitely, it would be a god-send (in the ROI sense); however, I think your arguments of dissuasion that Starbucks would never do this are akin to the arguments of high-end car manufacturers never moving to automation to build vehicles.

Put succinctly: If the end-goal of business is to increase the profit-marigin (which it is), then someone, somewhere will find a way to make that happen and this would be, in general, no exception to that rule. Eventually, they'll hit that sweet spot to make it work; even if it's slightly sub-par from the quality of today's product.

The question is whether or not Starbucks is like a moderately high-end car or a luxury handbag, where the whiff of anything not 100% assembled by hand makes consumers flee.

Starbucks really muddled the waters during their rapid expansion about which customer they‘re targeting, and at some point they will be forced to pick.

Well, I disagree thats its a $200 coffee machine, but theres no way on earth this needs to be an arm emulating a human user on existing machines. Im pretty sure you could buy a "premium" commercial automatic coffee machine for less than the cost of the arm - in fact, thats what the arm appears to be operating!

Looks like I misunderstood GP. He's pointing out that all the arm is doing is moving cup and pressing buttons on automatic coffee machine too. I agree, this isn't anything special, this is just a robotic arm performing simple manoeuvres.

> “Starbucks needs to be automated.”

That’s a pretty bold assertion. It seems to be doing just fine without having been automated.

Starbucks is premium? In London I consider it bog standard.

I think "premium" means in relation to office provided filter coffee/pod coffee, not in relation to cafe coffee

It's been years since I worked somewhere that didn't have bean machines. Figured it was the norm these days.

They started off that way. It was one of the places in America that treated workers nicely and probably one of the first chains to source beans that were fair trade and sustainable.

Then they decided that moving into every big box store and rest stop was important.

I'm hoping the $25k is more than just the arm and is more for the entire retail ready unit as displayed on their website... which is not bad value

Even at 30k for the coffee machines and 25k for the arm plus 5k for the enclose. 60k investment for $300/hr. Like $200 margin running even 12 hours a day. That's over 800k a year in the right location.

Add the cost of coffee, water (usually filtered/depurated), electric power, sugar, washing, ...

... and good luck to find a location that has customers needing 120 coffees/hout for 12 h each day of the year.

Robot arms are surprisingly expensive. This seems like a low-end robot arm.

> Today, an industrial robotic arm can cost anywhere from $25,000 to $400,000. [1]

[1] https://insights.globalspec.com/article/4788/what-is-the-rea...

Low-end arms are ~5k. This is just quite small, a Mitsubishi RV-FR series, payload ~5 kg. 300k+ robots are things like the KUKA KR1000 ("Titan"), which can literally toss around cars. It's got 200x higher payload for <20x the cost; if anything small arms are overpriced (they are- the only current use cases are pharmacies, low-end semiconductors and medicine).

Anyway, Mitsubishi is well respected and this is a very good arm.

[1]: https://us.mitsubishielectric.com/fa/en/products/industrial-...

> $25,000

That's an annual salary of a coffee server

I don’t know how good the vending machine’s coffee is, but there’s a cafex close by and it’s pretty dang good.

In most of those chain coffee shops what you get is a coffee vending machine with a human arm to move the cup.

That’s the goofy part here: that arm is completely unnecessary. You could just have the customers put the cup under the spout themselves.

Yeah, but what about the experience?

You could already have pure vending machines serve what baristas serve you at Starbucks, Costa, whatever, you would get exactly the same product and it's close to what happens in reality. But people want a coffee shop experience, not a vending machine experience so I think moving robots will be a thing (unless people really insist on interacting with humans)

I’m not sure I agree. We went to a lounge with a moving, anthropomorphic robot bartender and had a lot of fun interacting with “him” (and watching other people do the same). However, it was mostly the novelty of the experience; we had a great time but we’re not dying to go back ASAP.

Sometimes, the service is part of the coffeeshop/bar experience, but seeing stuff being done is a fairly minor part of that, at least for me. Instead, it’s a quick chat or a recommendation, neither of which a moving arm really provides. You might be right that the apparent labor, either by a human or moving arm, also masks the (large) margins on a coffee drink. I hadn’t really thought of that as “the experience” (vs comfy chairs, decent music, and a pleasant murmur of other conversations) but you might be onto something there.

However, I think the main reason vending machine coffee isn’t popular is because it’s historically been pretty bad. The newer ones (Nespresso, etc) do make decent coffe, but you can still get some vile sludge at hospitals and rest stops.

I'm not aware of any major chain coffee shop that uses coffee vending machines like that.

Well, they can't make it too obvious, can they. But it is basically it.

Employees do no more than employees at say, McDonald's. In fact probably less. Push a button, right amount of coffee pours into cup. Move cup. Push button, right amount of milk pours into cup. Move cup.

Replacing people by robots will make no difference to the final product.

What we need to wait and see is whether consumers want to be served by a human or whether they don't care. We'll have an answer very soon because for fast food and coffee shops there's just a couple of years before humans become operationally optional.

Personally, in a high end place where staff adds value I will care, but I couldn't care less in a Costa Coffee or McDonald's.

> Employees do no more than employees at say, McDonald's. In fact probably less. Push a button, right amount of coffee pours into cup. Move cup. Push button, right amount of milk pours into cup. Move cup.

I don't know what places you're talking about. Everywhere I've seen they use an espresso machine which isn't a case of 'push a button, right amount of coffee pours into cup', even places like Costa.

I observed in Costa and others...

Even when they use en espresso machine (because perhaps you ordered an espresso), it makes no difference to the final product if that is automated entirely. A standard espresso machine requires a human only because it cannot automatically reload coffee.

The simple fact is that at this point humans are really only there for the 'experience', not to create the final product.

Well I was in a Costa myself earlier this week and I literally watched them use an espresso machine, and it wasn't just pressing a button. I can see that with my own eyes so I don't know what parallel universe you're living in.

(I can see you editing your comment out from underneath me by the way.)

What do your comments add to my actual point? I obvious know that it is not an actual vending machine at the moment... But it is in all but name.

Humans in Costa don't add any value to your cup.

They might make you feel better, though, so we'll see if people are fine with being served by a robot (as I already wrote).

But, please keep focusing on the literal.

When pointing at the Moon...

Your original point was

> In most of those chain coffee shops what you get is a coffee vending machine with a human arm to move the cup.

What I'm adding is that anyone with eyes can go to a chain coffee shop and see that they aren't using a vending machine.