Not long after the pizzeria Zume opened for business last year, its kitchen staff noticed a problem with some of its pizzas: they had holes in them.
It wasn’t the fault of the workers, who rolled out intact dough bases. There wasn’t a kitchen mole prodding holes. It wasn’t even the recipe — a Zume pizza base can handle its fair share of toppings. It was the robots.
Josh Goldberg, 38, is the chief technology officer of the Mountain View, Calif., pizza joint. Although most pizzerias don’t have an engineering staff, let alone a CTO, Zume prides itself on its use of automation to make operations more efficient.
It estimates its kitchen can make 10 times more pizzas than a pizzeria with a comparable staff. It has a robot that squirts tomato sauce onto its pies. It has a robot that spreads the sauce, mimicking the movements of Zume’s head chef. There’s a robot arm (similar to those found in auto manufacturing facilities) that puts the pie in the oven. And, as of this month, there’s a robot that presses the dough into a perfect circle.
So if the company has a non-human problem, it’s Goldberg’s problem.
Observing operations in the company’s lab-like kitchen, Goldberg watched as the human cooks spun glossy blobs of dough and placed them on the conveyor belt. He watched as a camera hovering above snapped a photo of the dough so it could inform the other robots of the pizza’s size, shape and precise location. Another camera detected the center of the pie and instructed a nozzle to squirt sauce, and a delta robot — the kind used on assembly lines — used a spiral movement to spread it. Humans topped the pizza with pepperoni, fresh basil and cheese.
And just as the pizza was about to be put into the oven, Goldberg found the problem.
There was a small gap between two conveyor belts. If both kept moving, then the pizza would glide along without any problems. But if the conveyor belts stopped for as little as two seconds, part of the pizza would sink into the gap, creating a tear when the machines started up again.
Fixing the problem was easy enough. Goldberg just had to program the conveyor belt to not stop when a pizza was passing over the gap.
But there are always new problems.
Robots need calibrating, code needs to be updated. Whenever there’s a change in the dough or sauce recipe, the robots must be taught new ways to work with different textures and consistencies.
That’s why Zume has a team of 20 software engineers. And its 20-person kitchen staff doesn’t just prep ingredients; many have been trained to work with the robots. Its entire culinary team uses project management tools such as Jira and Kanban, which are typically used by software engineers at tech companies for managing projects and fixing bugs. And other staff, such as delivery drivers and line cooks, are being trained in data science.
“We have fewer people, but we pay them much higher rates with full benefits as opposed to having a proliferation of lower-skilled workers,” said Julia Collins, Zume’s co-founder and co-chief executive.
The goal isn’t end-to-end automation, Collins said. There are still things that humans do better than machines, such as prepping ingredients, making sauce, developing recipes and knowing when something isn’t right with a pie. But automation and software enables Zume to reduce costs, make more pizzas, predict what pizzas people want before they order them and, eventually, take on the big pizza chains.
At least that’s what investors — including the venture capital firm SignalFire — are betting. They’ve put more than $23 million into the company.