Flipping burgers. Frying potatoes and chicken wings. Assembling pizzas on a conveyor belt. Self-driving robotic food runners and bussers. There is no doubt that robots are “invading” the food industry at a faster pace after the pandemic. They are increasingly seen as a competitive advantage in addressing short labour supply and relieving staff of the tediousness and boredom of doing repetitive tasks in quick service restaurants (QSR) and fast-food chains.
With increased adoption, we are seeing the future of food robotics evolving before our very eyes. It is no longer a question of robots becoming a mainstay in the industry but of how fast it will shape the industry – moving from automating repetitive and batch-like tasks towards fully autonomous robotic kitchens that can produce food from scratch.
Barcelona-based REMY Robotics is banking on this shift in the industry, producing autonomous robotic kitchens that do not mimic humans but use robots for what they do best— precision, speed, repetition— to deliver entire meals with food from different cuisines, dish types and recipes.
We spoke with Yegor Traiman, Founder and CEO of REMY Robotics to find out more.
How did you get started with REMY Robotics?
I have a background in cybernetics and have always loved tech. I had worked in IT for many years and when the opportunity arrived to work with robots, I jumped on it. I was on the executive team of Arrival (NASDAQ: ARVL), the British electric vehicle manufacturer. Arrival was my first experience in the robotics industry: from that I understood the power of robotics and how robots can make a lot of things easier, safer, better. I realised that cars were not my thing, but I wanted to use the exponential potential for change that robotics can bring to our everyday lives. I figured we eat three times a day, so that’s the biggest impact you can have: tackle food. And the food industry desperately needs an overhaul. Food waste, terrible working conditions for kitchen staff, low margins for food operators- all these major problems can be solved with smart use of robotics. So, I founded Remy.
I wanted Remy’s HQ to be based in Barcelona, which for me is the best city in the world. I recruited the best of the best from 15 different countries across Europe, Asia, and the Americas. Being in Barcelona means we all have our dream job and the best quality of life. The sun, the sea, great food, and a job in tech that doesn’t ask you to move to grey, cold, expensive cities. This is the secret recipe to our environment of creativity, trust, and ambition.
Where does REMY Robotics stand in the automation vs. autonomy spectrum in restaurant food robotics?
There are a lot of food tech solutions out there, but their application is very limited because they are based on conveyor-type robotics – a legacy of industrial robotics (think of car manufacturing systems, for example). This approach means solutions are hardware-defined and therefore inflexible. They can only work in a fixed isolated environment, can only produce one type of product at a time and are extremely expensive to develop. With this approach, if you want to cut a pizza one day, but you want to build a burger the next, you need to create two entirely different systems.
If you want to truly disrupt the food industry, you need to build a flexible system around the robots- not just automating one task in a fixed environment. Our approach to deliver autonomy- not just automation- is based on adaptive robotics and culinary engineering. With adaptive robotics we can cook any type of food and our kitchens can be deployed and scaled fast. We cook by combining the latest technology- robotics, IoT and AI- with food science and human creativity. We put robots first and design around their strengths, rather than their limitations. For example, we don’t try to make a robot handle food (robots are bad at dexterity), but we let it work with data and neural networks to cook food with precision, consistency, and speed.
On the food side, our culinary engineering team developed a new cooking method with algorithmic cooking at its core. Data-driven and precise cooking is possible because we built a whole ecosystem to sustain it: proprietary ‘food capsules’, robotic ovens, ultra-precise temperature control, and AI-powered quality control. Flexible robots and Artificial Intelligence orchestrate everything so that cooking can happen 24/7 and 100% of our orders are accurately assembled.
Tell us about your robotic kitchens. What are the capabilities and limitations?
Our robotic kitchens are fully autonomous kitchens where smart equipment uses robot-to-robot communication to assemble, move, cook and check food, as well as refill stock. Software binds everything together and makes each of the purpose-built appliances and smart modules do what they need to do for kitchen and restaurant automation.
The kitchen is made up of programmable robotic arms, interconnected equipment like robotic ovens, fridges, machine-based quality control stations, and smart shelves. These work seamlessly together thanks to the kitchen’s ‘brain’- our end-to-end software system and neural networks. Of course, this sounds terribly complex- and it is. But to the naked eye, our kitchen looks like a very compact unit with a robotic arm in the middle. You can see the food being picked up from our fridges and placed in the ovens to be cooked. Once it’s cooked, the robotic arm takes it out and places it on our machine-learning based quality station, where the system decides if the dish is good to go to the customer or not based on key parameters like, among others, moisture loss or temperature. All decisions- to move the food, how long to cook it for, etc- are made by neural networks and software. The magic of working with robots is in all the things us humans cannot see. Robots don’t need eyes; they have computer vision. They don’t need hands; they have robotic arms. And they do incredible calculations in a fraction of a second. They are never wrong- the level of precision is unattainable for a human being. The beauty of truly adaptive robotics is that there are no real limitations: you can cook almost anything you want and have the exact same result every time.
Can you walk us through a specific example of how a meal is prepared from the cooking process until it reaches its final customer?
We cook with digital recipes that are dynamic, i.e., based on algorithms that we feed to our robotic ovens. Our recipes go from our most sold item ‘lemon and dill salmon’, to barbecue ribs, Indian curries, pasta, dumplings, desserts- you name it, we can cook it. We develop formulas for each dish, detailing all the variables to guide our smart oven to perfectly cook food inside our capsules and to allow AI to check that it’s done properly.
So, let’s say we’re doing salmon: It starts with our Culinary Engineering team who classifies the dish and then develops formulas for how to cook it to different states (medium, rare, well done). They then input variables and create an equation that allows the robots, which are permanently connected to the ovens, to precisely execute the dynamic recipe. The recipe algorithms define weight, density, water content, desired texture, time to eat (how far a delivery customer is) and many more variables that ensure food is cooked to the desired level and quality.
In our central kitchen, humans prepare the food, doing what robots are not good at: chopping vegetables, handling food, and putting it into the food capsules. In the case of our salmon, they prepare it ‘sous vide’ and then place it in our food capsules. Once the preparation work is done, QR codes are automatically generated detailing all the information about the dish, so that when a customer orders the salmon on Glovo or UberEATS (or any other delivery platform), the robots know exactly what to do.
The online order system automatically communicates with the robots in our autonomous kitchen, where in a matter of seconds they start cooking the dish. It’s usually ready to be picked up by the rider as soon as they arrive (we cook 40% faster than usual kitchens and 95% of our orders are cooked precisely on time). The dish will then continue to cook during delivery (usually between 12 and 20 min) so that it arrives at your doorstep perfectly cooked.
With dark kitchens, you are currently offering delivery food of in-house virtual brands through platforms like Deliveroo and UberEATS. How is it going?
It’s going great. We’ve sold 70,000 dishes and have great reviews on all platforms- on Glovo we have an average of 97%, which is huge. We’re very proud that we’ve managed to create food that people love and come back for (we have a repeat customer rate of 50%). Our flagship virtual brand is a flexitarian delivery only restaurant called OMG (you can try it if you are in Barcelona), but we’ve also operated other brands of Indian curry, chicken wings, pizza, vegan Chinese food and much more. All these brands can be deployed at the drop of a hat, as we have done all the work of developing brand concepts, menus and digital recipes, as well as tested them with consumers in the last 12 months. We don’t just do what is currently available for sale- we can do (and have done) much more. We have over 100 recipes in our arsenal of food from all over the world.
Are you currently working with other restaurant brands offering food deliveries?
We’re in conversation with a number of food operators and have a couple of pilots under way, but we cannot say much more for now. At the height of the pandemic, we started collaborating with physical restaurants, too, to help them create a delivery system that worked quickly- and it was a success.
So, we run our own food delivery brands, but also collaborate with other types of food operators: catering services (at universities, hospitals, airports, etc) and restaurants who want to find a solution to the labour shortage crisis. We can help you set up your robotic kitchen in 24 hours in as little as 15 sq. m and develop a food concept for you in 6 weeks.
We love the delivery space- it’s growing, it’s exciting and there’s innovation ready to be deployed, but the end game for us is to power the commercial kitchens of the food industry in general- all segments. We have successfully passed the product-market-fit stage. Our robotic kitchens and the economics behind them are tried and tested.
What are the biggest misconceptions that people have about food robotics?
The first big hurdle for us is that when you say ‘robotic’, people think of a humanoid robot with big eyes and walking around doing things that humans do- in our case, chopping vegetables and stir frying with ‘wrist-like’ movements. For us, that’s not just a mistake in terms of what a robot actually is (a humanoid robot is just one type of robot), but also a misconception that creates fear, rather than trust. If you think a robot looks like you and will replace you, it’s only normal that your first reaction is to oppose it.
It’s also unhelpful for two more reasons. Firstly, food robotics that mimic how humans cook are inherently flawed as a solution to the most urgent problem of the food industry- the labour shortage crisis. The current level of dexterity, i.e., the ability to handle food, of a robot is still very limited and development costs very high, so solutions where a robot handles the food are just long-term R&D projects that won’t see commercial application in decades. In short, a robotic chef is not going to steal your job- not just because it is not able to sauté your vegetables, but also because this is likely a job that you don’t want. Robotics can mean better kitchen jobs, where people are not doing boring, repetitive tasks at ungodly hours and, instead, can focus on the more creative parts of working with food. This would solve the current labour shortage crisis in a way that helps food operators have better margins and people better working conditions.
Secondly, thinking of robots only in terms of anthropomorphic machines misses the point on why we use robots: they do certain things much, much better than us humans. They are precise in a way we simply cannot be. They are faster, can work any time and they can take care of the boring tasks we don’t want to do without complaining. In food industry terms, this means kitchen staff can work normal office hours without stressful peak times and go home to their families, have a social life and enjoy working with food creatively.
What food trends do you think will emerge in the next 2 years?
So far we’ve seen food tech focusing on single-task automation- with robots that can either flip burgers or fry potatoes- or on vending machines-style solutions, with automated dispensers creating poke bowls or pizza-in-a-box restaurants. These are great innovations, but they have limited impact on the problem we need to solve: labour shortage. I think in the next couple of years we’ll see a shift from automation to autonomy. By that I mean we’ll see more solutions that are truly autonomous, where robots execute the entire cooking process without human intervention. It’s possible- we’re doing it. I think we’ll see more of that, as the market focuses on solution longevity and better ROI.
I also think that a new type of quick service restaurant will emerge: the MacDonald’s of the future will be more focused on delivery and will push the boundaries on digitalisation- much more than just screens to order food on site. The QSR restaurant of the future will be a tech company first and foremost: end-to-end automation (or autonomy) will be built-in from day 1, and digitisation will not be just an add-on to existing antiquated and intricate processes. Tech will be the key driver to create new avenues for true innovation. With tech at the core, delivery and digitisation will be an integral part of a restaurant’s business model, leading to much better service, customer experience and ultimately, revenues.
How do you define “culinary engineering” and its role in the products you produce and the direction in which REMY Robotics is heading next?
When we started creating food for robots we realised our approach was unique and there was no word that would describe how we do it. So, we created a new term to describe it: Culinary engineering. It’s a new discipline which maximises the main advantage of using robots: precision. It combines cooking creativity, food science, food engineering, food design and technological engineering. It looks at the culinary world as a whole.
Food for robots needs to be designed around them. Traditional aspects of a recipe, as well as how to check that food has been cooked properly, need to be broken down into science and data, so we created a whole new methodology to turn what comes naturally to a human into formulas, algorithms, and digital recipes.
We’ve developed all sorts of recipes: from Asparagus to Za’atar focaccia, passing through xia long bao, truffle mac & cheese, gulab jamun and hundreds more dishes. We’re planning to develop even more, creating more variety, focusing on improving our quality even further- with more precision from robots and great creativity from our chefs.