If a robot making and serving you lunch sounds fancy and futuristic, Alex Kolchinski says you're falling behind the times.
This weekend, Kolchinski's company Mezli is set to open a fully autonomous restaurant in San Francisco. The eatery, also named Mezli, will serve a variety of Mediterranean-themed grain bowls conceived by Michelin-star chef Eric Minnich. Kolchinski says it'll be the world's first fully autonomous restaurant, with a customizable menu of more than 60,000 possible variations, depending on the sides, add-ons or substitutions you choose.
Other robotic food-making machines, and even simple-concept autonomous restaurants — like this pizza shop in Paris — already exist. But instead of an expensive novelty, Kolchinski says Mezli's goal is actually to make your lunch cheaper, healthy and convenient.
"We realized that if we could do food service in an autonomous way, then we could actually bring the price point way down and start serving better food for lower prices without people having to cook it themselves," Kolchinski, Mezli's CEO and co-founder, tells CNBC Make It.
The restaurant's savings on labor costs and square-footage are reflected in its menu prices. Meals start at $6.99 for a roasted carrot and cauliflower bowl, which comes with red rice, hummus and other veggie garnishes, and top out at $11.99 for more protein-heavy options with chicken or lamb.
That's less than some of the more expensive options at popular fast-casual lunch chains like Sweetgreen or Chipotle, where you could end spend upwards of $15 per lunch, Kolchinski says. Chipotle's least expensive burrito bowl starts around $9, depending on your location, according to a CNBC Make It analysis of the chain's website.
Housed in a blue-and-white box the size of a shipping container in a San Francisco food truck park, Mezli contains a high-tech oven and robotic machines that heat up prepared ingredients and mix them together in a variety of combinations.
As a customer, you make your selections — like a bowl of falafel, or spiced lamb and tzatziki with vegetables, turmeric rice, hummus and a cookie and drink on the side — from one of the digital screens on the side of the box. Inside, machines heat, mix and plate the ingredients associated with your specific order.
Minutes later, the restaurant dispenses your order from a pickup window at the end of the box. Mezli can currently churn out roughly 75 meals an hour, which Kolchinski says is on par with most large-scale fast-casual restaurant chains.
There's still an important human element in Mezli's process. Minnich, the founding chef who devised Mezli's adaptable menu to work with the restaurant's robotic tech, leads a team of cooks in an off-site kitchen who prep ingredients that get delivered and loaded into the refrigerated Mezli restaurant.
In the off-site kitchen, Kolchinski says ingredients are "cooked to food safety standards." Some get additional cooking time in the restaurant "for freshness and quality," he adds.
Ahead of Mezli's grand opening on August 28, Kolchinski says the automated restaurant had a "soft opening" that included a few private events and test days with the general public. "Having this kind of new experience, I think people don't know what to expect," he says. "And when they end up getting served good food, it's nice for them. People have generally been a combination of surprised and pleased."
At first, Mezli employees will be nearby to explain the restaurant's concept to curious customers and look for potential improvements. Eventually, the restaurant shouldn't need any human supervision, Kolchinski says.
That has the potential to leave some customers uneasy. After all, what happens if a machine messes up your order — or you simply want to send something back because it's not what you expected?
At a typical restaurant, you could voice a complaint to a human staffer, and potentially get a refund or substitute meal. At Mezli, the customer service element looks more like calling or emailing the company to give your feedback, though that system brings no guarantee of an immediate response.
"There are ways for people to reach out to us, even if there's no one on site at the time," Kolchinski says.
Kolchinski co-founded Mezli in 2021 with fellow Stanford engineers Alex Gruebele and Max Perham. The trio came up with the idea as graduate students, struggling to find healthy, convenient lunch options for less than $15.
They researched restaurant economics and discovered that food ingredients cost very little compared to the costs of labor and renting or building a restaurant's physical space. When you buy a bowl from Mezli, about half of your money covers the meal's cost, Kolchinski says. At most restaurants, cost of ingredients accounts for roughly a third of a meal's price, he adds.
Kolchinski says using machines may eliminate some human jobs, but it allows the company's human chefs to focus more on the work they most enjoy — actual cooking and prepping — rather than the repetitive, time-consuming steps of taking orders, plating them and serving them to customers.
Mezli may serve food, but it's a tech startup at heart: The company developed prototypes in late 2020, and was accepted to startup incubator Y Combinator by the end of that year. It has raised roughly $3.5 million in seed funding from venture capital firms Share Capital and Metaplanet Holdings.
That money has gone toward designing and building the current version of Mezli that will launch this weekend, Kolchinski says: "All in all, end to end, it's taken us just over a year and a half to get from starting out to having this thing up and running."
The soft opening's results were promising, but Mezli will have to prove it can attract enough hungry customers to turn a profit and attract more funding before it can think about launching additional locations, Kolchinski says.
New locations would first sprout up elsewhere in California's Bay Area: Kolchinski notes that his company could reduce its margins and even lower costs with new locations stocked by the same cooks prepping food in a central off-site kitchen.
The company could also expand to other cuisines, like Korean or Thai. "The nice thing is that these [machines] are agnostic to what they serve," Kolchinski says. "So it really is a matter of updating the ingredients that are delivered in the morning, and the software that lets people order, and you can have a whole new menu."