Launch HN: Mezli (YC W21) – Robotic restaurants that serve healthy fast food
The three of us met as grad students at Stanford where we were all working on different things – I was doing AI research before dropping out of my PhD, Alex G was in a robotics lab (and just finished his PhD!), and Max was in aero/astro. We worked on a variety of classes, research, and side projects together, but we wanted to start a company and none of our ideas were looking particularly commercially viable. Then, as I was winding down a project building an autonomous weeding robot, it crossed my mind that one of my own biggest daily frustrations was something that was worth building a company to solve.
That frustration was that eating well in America requires spending a lot of time cooking or a lot of money buying meals. In grad school, I didn’t have enough time to cook every meal, but I also couldn’t afford to spend $10 or more at Chipotle, Sweetgreen, etc. It turned out that most of my friends, in and out of grad school, had the same problem. So, with Alex G and then Max as well, I started looking into why good/healthy restaurant meals in America are so expensive.
It turns out that a lot of it comes down to costs that are passed down to customers. An average Chipotle restaurant costs a million dollars to build and runs up a $600K/yr bill for on-site labor. That all gets passed on to customers, so that a $10 burrito bowl has only about $3 worth of ingredients in it, but also $3 of restaurant labor and $4 to cover things like rent and profit margin – which for most restaurants is quite thin. We realized that reducing the cost of building and operating a restaurant could unlock much cheaper great-quality meals. So Alex G and I, soon joined by Max, started talking to people all over the restaurant and automation spaces and brainstorming how to solve the problem.
It turned out that if we constrained ourselves to bowl-style meals (grain bowls, salads, soups, curries, etc.), we could use a lot of existing automation equipment off-the-shelf, put it in a shipping container and integrate it with a few pieces of custom hardware to make an autonomous restaurant-in-a-box. The hardest part turned out to be the dispenser technology – putting ingredients in a bowl reliably is not trivial! We came up with a new approach for that that we’ve recently filed a patent application on and we'll be able to talk about more publicly once the patent is granted.
Like most restaurant chains, we do the bulk of our prep in a central kitchen and then the auto-kitchen itself uses a variety of heating and finishing steps (e.g. applying sauces and dry toppings) to make bowls to-order. Unlike some food automation companies, we’re focused on creating a fully automated “restaurant in a vending machine” rather than human-in-the-loop partial automation. Getting our tech to work reliably enough to not need a human to monitor it is a challenge, but comes with benefits like being able to make more meals, faster, out of a smaller space. It also gives us food safety advantages because there’s less room for human error, and we can also do things like bathing the insides of our boxes with high-intensity UV light that kills germs but would not be very employee-friendly!
We’re also taking the point-of-view that solving food automation requires leaning into special-purpose hardware, rather than just trying to program a robotic arm to do everything a human cook does. As a former AI researcher, I can speak to the difficulties of programming arms to do even simple tasks like pick-and-place, let alone ...
128 comments
[ 2.5 ms ] story [ 197 ms ] threadAre there any videos of the robots working? Or none yet due to the patent work?
(But: I can see why you might want to wait to make sure any images provided also put potential food-purchasers at ease.)
> Prepared at KitchenTown by Chef Eric Minnich, former Chef de Cuisine of Michelin-starred Madera.
I thought robots were preparing the food? What's this about?
>The Mezli founders teamed up with Eric Minnich, a classically trained chef with experience at Michelin-starred restaurants, to develop a menu that a robot can confidently execute 24 hours a day, seven days a week. Because there are no humans involved or expensive rent to pay, that means you can get a cauliflower and turmeric rice bowl for $4.99.
I guess he's the chef that created the menu
I think people underestimate that Americans have a hunger for homogeneity when they go out for fast food. Everywhere in the continental United States, a hamburger from McDonald's tastes the same. If I find a few things I like from a restaurant, that's what I am going there for. If I want to be surprised, I will go somewhere new.
Edit: just saw it's mediterranean food. Cheers
It's also funny to see the Spyce creators complaining about $10-$12 meals, when the prices on the website range from $10.40-$11.90.
The business model and approach does look pretty different, but the types of meals look vaguely similar. I'm curious if that's just because when you start thinking about what types of meals can be automated (partially or fully) it pushes you in that general direction. If you don't mind sharing, what was the process like for figuring out what types of meals could be cooked by robots?
Good luck!
Your goal of bringing healthy food more cheaply to people is great, but based on the other advantages this tech has, I think the price should be much closer. Eg if chipotle is selling it for $10, yours should be $8 or $9 not $5.
If you hit about $6-7 a meal, you'll be very close to competing with home-cooked meals on cost. Sure people can go a bit cheaper, but it isn't gonna be really a significant reduction, but there will be a large time cost to it for home-cooked.
I am rooting for you. :)
Yeah for our broke selves the goal has been to basically get the price point down to the point where it's a no-brainer to skip cooking.
Not if you value your time. Cooking raw beans can easily take 45 minutes.
Plus, I don't get the fascination of cooking "single" portions the last decade. The amount of time it takes to cook one portion of nearly everything, you can knock out 4-8 in the same 30-40 min time span. Leftovers next day, lunch or freeze it for a little while. You spent $3/serving and you technically saved more time in the long run and hassle just because you learned to utilize the whole pan instead of a small corner.
The problem was taste. At some point he was getting literally depressed. I told him to do 1 normal meal once a day, and his mood improved.
Taste is important. Variety is the spice of life.
Turned out things didn't work out too well for Eatsa, but I think it was more due to errors in execution (hyper fast growth with nation-wide expansion, and lots of time and money spent on developing custom hardware) rather than a lack of product/market fit.
Best of luck to Mezli!
(Disclosure: former Eatsa employee)
McDonalds announced a big automation push years in 2019. Again, not robotic kitchen automation. Automating order taking and supervision.[1] A previous try at that was called "Hyperactive Bob".[2] That's like Amazon's warehouse automation system with Kiva robots.[3] Or the Chicago Dryer "Cascade" system for folding linens.
These things use computers to tell the people what to do, because unskilled human hands are cheaper than robots. The human part is reduced to simple eye-hand coordination tasks, which robots still can't do very well. "Machines should think. People should work" automation.
[1] https://www.foodserviceequipmentjournal.com/mcdonalds-plans-...
[2] https://www.techdirt.com/articles/20040908/0156256.shtml
[3] https://youtu.be/CWNuaPE4DTc
[4] https://youtu.be/k-DSd2o-mP8
Obligatory dystopian short story on this theme:
https://marshallbrain.com/manna1
Oh, it's in process. I consulted on a early prototype of a new automation project for them and that was 3 years ago. McD just refines and tests the hell out of things for years with their vendors before they launch it.
I worked in a McD kitchen 30 years ago as a teenager. The ironic thing is that the amount of people needed to run a kitchen has stayed essentially the same.
Most run with 2-3 people during an average period. It's just that over those years McD has already automated a huge portion of the grunt work in the kitchen, the rest is just tending to the machines and final assembly + customization, which no robot will get right for another decade or two.
Humans are more versatile, easier to train, quickly replaceable when one goes offline, and keep a much cleaner workstation. The robots do the JIT work on everything else.
This is a biassed viewpoint.
The problem is that humans do dozens of tasks. When we take one of them and give it to a machine, we view that as a new tool, and we imagine the human hasn't had its work reduced, because 19/20 of the things they were doing last month they still are.
Fry machines are now automatic. So are cash registers. Many locations' burgers are auto-flipped. Those ice cream machines automate 5 different human jobs. Lots of orders are taken on machine now. Etc, etc, etc.
.
> Humans are more versatile, easier to train, quickly replaceable when one goes offline, and keep a much cleaner workstation.
No they don't. If you made your own ice cream, you'd understand.
It's just that when a machine takes over one specific job, we stop thinking of that as having been a human job, and let the human messes go out of our heads.
There was a time when fast food restaurants sliced their own cheese; ground their own beef; cut and cooked their own french fries; mixed their own sodas.
You don't live long enough to understand the progress of piecemeal automation.
I relate 100% to your problem statement. It's even harder if you have dietary restrictions you're trying to accommodate, for example non-dairy, gluten-free, or low-carb. You spend a lot of your life prepping food, paying a ton of money for the small number of fancy restaurants that serve what you want, or else compromise and eat food that isn't what you want to be eating.
If there was a low-carb robot kitchen in my neighborhood serving $5 meals I'd probably be visiting that 10-20 times a week.
I could see tech like this also helping in underserved communities where healthier food is not only not affordable, it's simply not available. Lowering the capital requirements and unit cost could mean a better supply of healthier meals in neighborhoods that currently have few choices.
That being said, it's hard not to also feel a tinge of concern when reading announcements like this. Automation is coming for a _lot_ of jobs (in this case the ~14 million Americans who work in restaurants). I'm an optimist about such things, but I do also sense a concern that for a lot of people there aren't many "good jobs" left (where "good job" is defined by something that you could learn/train on the job without requiring special skills or higher education, and eventually make median income or better).
100% agreed on the underserved communities. I remember going to Burger King with my family as a kid because that's what we could afford and was nearby.
Restaurant jobs are an important question. We think we're more likely to displace more home-cooking than restaurant demand (we don't do sit-down at all), but there's also a big labor shortage in the restaurant world right now because fewer people want to work in kitchens than used to. I'm hopeful that automation will help people be able to cook less but keep demand for dine-in restaurants largely untouched (people want to go out to eat with friends and family).
You would still need proper permits/licenses and real estate/rent. Insurance. Utilities.
The prep and cooking of the ingredients still needs to be done, and then those items transported in a chilled/heated manner for food safety.
And if there are perishable goods like produce or meat, the machine will need to be thoroughly cleaned and sanitized regularly, which would require labor, a sink, etc. Unsold food at the end of the day needs to be stored/chilled properly back at the central kitchen. Ingredients will need reloading, machines may get jammed, etc. So it doesn’t seem like it will be labor-free.
Even in the best case scenario, like the sandwiches or muffins at Starbucks, or even deli items at the grocery store, the prices aren’t all that low, so I am curious to see if this can be done profitably.
E.g. ingredients are pre-prepped in the central kitchen and then transported refrigerated to the auto-kitchens. The mechanism is designed for easy cleaning and refilling upstream and reliable operation in the auto-kitchen. It's one of the only parts of our machine that's totally custom; we've filed a patent on it because we think it gives us a significant leg up over previous approaches.
Food safety is also a huge deal to us; we've been working closely with regulators from almost day 1 to make sure that our approach is even safer than a traditional restaurant (less room for error.)
And rent is also a plus for our approach: a parking spot costs less to rent than commercial real estate!
How does last mile supply chain to deliver the pre-prepared ingredients work?
How does fresh/stale ingredients management work?
How cleaning of the insides of the machine work?
How frequently will it require servicing? Will the entire container go to service center (swap out the container without downtime for customers) or will a service crew visit to fix/service things at night?
Have you thought of doing this as a food truck? It might make things easier during the early days.
One huge frustration I've had as a customer is the unavailability of healthy fast food. I spent five months in Berkeley and ate at Sweetgreen all the time.
Now, I'm back in Columbia, SC where I work at a large university. If there was a Sweetgreen, or something similar, near campus then I would eat there 1-2x a week.
So I'm hoping that your efforts will lead to healthy and fast restaurants opening in places where they aren't currently viable.
It sounds like you have really analyzed your cost models. It's really fascinating to see the breakdown of labor and rent.
Walmart and Amazon have proven that people respond to cost more than anything, even if there is evidence those purchasing choices weaken their local community businesses and tax collection.
But, having said that, if there were a way to pay $7 for a bowl that had an innovative way to bring innovation to local employment AND had robots doing things in trucks that didn't have to pay rental in downtown, I would happily pay for that bowl over a $4 one that removed the labor entirely. Just my $0.02
But we might displace some demand and therefore labor from fast-food options like McDonalds. That said, they're having a lot of trouble finding workers right now (~200% annual turnover in the space from what I've seen) so I don't think we'll be creating unemployment there either.
It's good for providing employment to young people without a track record and inconsistent schedules. It's bad for making any kind of career. Even managing is only barely a living wage, despite being the overwhelming task of dealing with those fungible workers.
It's kind of useful that such a thing exists. But it's disappointing that it makes them the third largest employer in the US -- right behind Yum! Brands, who do pretty much the same thing. They turn out a product that people should consume sparingly but is instead one of the main food providers.
In other words... a lot is broken. Poorly paid, disposable employees is just a tiny piece of it.
My take on the employment situation is that if currently ~10% of people are employed in the food industry, and we replace 20% of those with robots (so 2% of the whole population) then that should be an unequivocally good thing.
If there's not enough jobs to go around, but we're still producing the same amount of value then we should be supporting those who don't have work (in this case, probably by everyone in other industries doing 2% less work but getting paid the same as they are currently, and employing the people replaced by robots to take up the slack, along with some sort of subsidy).
The problem here isn't that we're working out ways to do things without human labor, it's that we're still stuck in this mindset of people having to work, even though there's plenty of resources to go around (and there should be even more with automation, not less).
Think of a society where there is a collective bank account and everyone shares their wealth. There are 100 people and there is 800 hours of work to be done. 100 hours of work in fast food. If you automate fast food then only 700 hours of work have to be done for an economy of the same size.
The real problem is fairness. Now that there is enough work for 88 people it would be unfair for the working people if those 12 people simply get to stop working. The remaining working members would go on strike. So we expect those 12 unemployed people to work regardless of whether there is enough work for everyone.
Globalization is another productivity booster that has the same effect.
The answer is to give people work, even if the work has to be created through artificial means. Infrastructure is a common example. You can just let these unemployed people build roads or houses and lay fiber for internet and so on. Infrastructure doesn't discriminate and tends to benefit everyone so ultimately even those who didn't lose their job will benefit. If that infrastructure results in a greater tax income in the future you can even finance it through debt.
I don't agree with this, but even if I did, we all know that in practice automation leads to a transfer of wealth from the working class to the very richest, that own the automated companies.
Just a decade ago, I would've scoffed if someone told me my phone could have as powerful a GPU as my desktop. And yet, here we are today!
On that note, is nutrition accessibility something that's priority on a short-term to medium-term basis on your strategic roadmap? I have no idea how something like this would work and integrate into programs like SNAP et al, but I can only imagine that once this technology matures a bit, it could seriously impact nutrition for at risk populations which are the ones who are most vulnerable to the vicious cycle of malnutrition anyways.
1964 technology for automated fast food: AMFare: [2]
[1] https://www.alibaba.com/product-detail/Commercial-Wok-machin...
[2] https://youtu.be/1Xop9py8zBY
"The food was really good. The hamburgers were great. The patties were transported on an open steel belt as they passed through the gas burner that cooked them top and bottom. They had a great char-broiled flavor. But the Amfare system never got anywhere, because it took a crew many hours each day to clean the machines. A normal fast food chef will clean the grill after each burger is made. You can imagine how bad the hamburger machine got after running continuously for 12 to 16 hours every day."
#2 is an interesting comparison that raises the question of why now is the time to build automated restaurants. The biggest reason in my mind is actually labor cost and people's increasing unwillingness to work fast-food style jobs. When AMFare came out it was easy and cheap to hire teenagers to flip burgers, but that's less and less the case and people still want cheap, convenient food. There are also other factors like the fact that we have a lot more modern electronics to work with and learned from examples like AMFare to make our system easy to clean.
How have you made your system easy to clean?
Now combine it with something like this for the outlet/dispensing/sales:
https://en.wikipedia.org/wiki/FEBO
http://blog.brillianttrips.com/2009/07/febo-dutch-fast-food-...
https://www.alamy.com/stock-photo-fast-food-snack-automat-re...
I remember them from decades ago in the Netherlands, while being there, stoned, and having the munchies.
I felt like I've been transported into the future!
Food in the wall! How cool is that?
I'd guess the Japanese have something similar in their cities.
The Automat was invented in 1895 and they were common (largest national chain of eateries common) in the US in the past :)
Besides the obvious difference that you're fully automated, I mean how will you avoid the trap of "Hey it's a lot easier to sell this food tech to others than to be in the restaurant operations business!"?
1: https://sf.eater.com/2019/7/23/20706270/eatsa-closed-tech-co...
Other startups that went this route found they still had to have a person on-site.
A lot of our IP is actually around reliability so I can't share everything but the high level of it is that: 1) We design every system with reliability as a top priority. 2) We're building in several layers of fault recovery mechanisms. 3) Most of the steps of our process have redundancy built in, so that even if something fails everyone still gets their lunch until a tech shows up.