Here's how artificial intelligence is revolutionising the food tech biz

Swiggy is already using machine learning-based models to predict the time it would take them to reach the restaurants

Fancy a robot cooking your meal for you? That day may be nearer than you think. In fact, the next time you order a chicken biryani from internet restaurant company Rebel Foods, chances are that it would be made by a robot. The Mumbai-based firm, which is known for its brands like Faasos and Behrouz Biryani, is looking to use artificial intelligence (AI) and robotics to automate the processes at its network of cloud kitchens. The aim is not just to standardise food items and maintain the quality and taste of food across all its kitchens in the country, but also to deploy less manpower and reduce human error.

“We are building robots that can prepare food items like dosas in minutes. These are AI and Internet of Things (IoT) enabled devices where we feed in the recipe to make the dishes,” says Soumyadeep Barman, chief technology officer at Rebel Foods.

Rebel Foods is among a growing number of food tech companies in India that are betting big on AI, machine learning (ML) and data analytics to automate their processes. The applications range from predicting demand, understanding customer preferences, preparing dishes and delivering the orders in the fastest possible time.

According to RedSeer Consulting, a research and advisory firm, the online food ordering and the delivery market in India is expected to grow at a compounded annual growth rate of more than 45 per cent to reach $11 billion in gross merchandise value (GMV) by 2023. Little wonder that all the major players are pushing to drive efficiency by employing cutting-edge technologies so as to be able to corner a bigger bite of the market. 

Rebel Foods, which was founded by INSEAD alumni Jaydeep Barman and Kallol Banerjee, operates about 1,100 online restaurants and 176 delivery cloud kitchens across 16 Indian cities. A ‘cloud kitchen’ is one that does not offer dine-in facility and accepts food orders only through an online ordering system. 

Many of the processes at Rebel Foods’ kitchens are already automated. For example, a dish like masala paneer tikka has 30 machine learning tags which help classify customer preferences such as the level of oil, calories and the number of herbs in it. Moreover, using machine learning, its kitchens are able to forecast the number of orders they can expect to get at any given time and location, and the kind of inventory they need to maintain to fulfil them. The foodtech firm has already seen a 60-70 per cent reduction in food wastage through the use of ML and data sciences. Moreover, by being able to predict demand, the delivery time has reduced by 20-30 per cent. 

Food delivery firm Swiggy is being equally aggressive in its effort to automate. It is building an AI-driven platform for hyperlocal discovery and on-demand delivery. One of its innovations involves ‘cataloguing’, where machine learning is used to automatically classify the images of food items as vegetarian, non-vegetarian and ‘eggetarian’. Swiggy, which commanded a valuation of $3.3 billion after its $1 billion funding round in December 2018, is also using technology to give recommendations to customers. 

“It’s similar to the way Netflix recommends films based on the movies you have watched earlier. For example, if you have been ordering a lot of paneer, I would like to show you more paneer dishes,” says Dale Vaz, head of engineering and AI at Swiggy. Vaz, who joined Swiggy last year and was previously the head of engineering at Amazon India, is helping Swiggy become an “AI-first company”. 

Unlike shopping on e-commerce sites such as Amazon and Flipkart where the products can be delivered in a few days, fulfilling food orders is extremely time sensitive. For instance, Swiggy promises to deliver food within 32 minutes. The Bengaluru-based company has close to 1.20 lakh active delivery partners. The mobile devices of these partners beam back their location to Swiggy’s servers every few seconds. And Swiggy uses machine learning-based models to predict the time it would take them to reach the restaurants, the time to prepare the order and then to deliver the food. Based on these, the system finds the best route and the rider who can deliver the product in the fastest possible time. 

The technology also enables Swiggy to factor in the impact of weather conditions, traffic patterns and road construction on the time taken for delivery. “We gather data at the zone level so we can predict the impact on the delivery time, say, the next time it rains,” says Vaz, adding that use of ML has helped the company see a 23 per cent reduction in driver wait time and a 50 per cent improvement in ‘on-time delivery’ of orders. 

Swiggy is also setting up an AI-research lab in Bengaluru which will focus on voice and computer vision technologies. Earlier this week, it acqui-hired Kint.io, a Bengaluru-based AI start-up which applies deep learning and computer vision to recognise objects in videos. Vaz says that in future, voice and computer vision would play a key role in how customers interact with Swiggy and order food using multiple Indian languages. “So when you wake up in the morning and say ‘hey Swiggy get me breakfast,’ we will be able to deliver it to you,” he explains. 

Swiggy’s New Delhi-based rival Zomato is also investing heavily in AI, machine learning and data analytics. The online restaurant guide and food ordering firm, which provides information on over 1.4 million restaurants across 24 countries, is busy developing the road to drone-based food delivery system. The idea is to create a hub-to-hub delivery network powered by hybrid multi-rotor drones. With this in mind, last December Zomato acquired Lucknow-based drone start-up TechEagle Innovations. In the first phase, Zomato is designing drones that can lift a payload of under 5 kg. “It is a big investment and as  the technology evolves, it will have a huge impact in the future,” says Gunjan Patidar, chief technocrat at Zomato. 

Patidar claims that in India, early investments into building an in-house data platform, AI, ML and data science have helped refine several aspects of the business such as reviews, search, order personalisation, demand prediction and route optimisation, among others. “User personalisation has helped decrease the average time taken by a user to decide the order by about 10 per cent,” he says.  

With AI and robotics revolutionising the food tech business, consumers will want to know when they can get a robot in their kitchens to take the pain out of daily cooking. All we can say is, watch this space.

Delivering the goods

• ML and data science have helped in 60-70 per cent reduction in food wastage

• Tech automatically classifies food images as vegetarian, non-vegetarian and eggetarian

• It recommends dishes the way Netflix suggest films based on movies customers watch

• Delivery time has reduced by 20-30 per cent as technology is able to predict demand

• In future drones might deliver food items to customers