Do we need cow for cow-milk anymore? Maybe not!

Plant-based milk produced from scratch using a deep learning model

Deepak Singh
5 min readDec 9, 2020
Source: NotCo

Do we need to farm cows if we can build milk from scratch now? That’s right, I said create milk without animals. Good news for vegan folks 😊

A product called NotMilk developed by NotCo, a Chilean startup backed by future positive (that backs twitter). But what is AI in here? Well, this product is a result of a deep learning model called Giuseppe that finds combinations of plant products that mimic the characteristics of animal-derived foods. NetCo has also developed plant-based mayonnaise, ice cream, and hamburgers in a similar fashion. This is not just a one-off case where the use of AI is doing great advancement in the food industry, whether it be a food processing, food delivery, food discovery, or the food hospitality industry. Let’s take a short tour of what different doors of the food industry has already allowed AI to enter and make big dents.

Sorting and segregating the produce

Proper sorting of food products & packages is one of the greatest challenges for the manufacturing facilities because the stocks received from farmers are not identical or uniform. The farm produce has different shapes, sizes, colors, and rotten items in it. To segregate them and route them to the appropriate direction is no simple task for a machine. The advent of computer vision systems has eased life and cut down costs from bringing the product to the market to taking them to the appropriate buyers. AI-powered automation has helped the food sorting industry in many ways from speed to yield improvement and from weather prediction to minimization of cost.

Source: YouTube

The model and systems are trained on faulty, multi-shapes, sizes, and colored products. This is then deployed to identify and segregate the items in appropriate bins in no extra time, space, or labor. One of the best examples is TOMRA, a leading sorting and collection provider. Check out this video to get an understanding of how well and efficient they have become using the power of computer vision in AI.

Coming up with new products

In the food and beverage sector, year over year, it has become very challenging to find out or predict what customers might like next and when they get bored by the current products. They have started using AI to figure out the solution. One of the examples can be Coca-Cola’s installation of self-service soft drink vending machines. Customers can customize the drinks as per their preference and the company can update the contents and test their so-called MVPs as to what combinations are sold in which area the most.

These fountains collect complete information to be analyzed without having any privacy concerns since no personal data is required. Using a basic ML algorithm of logistic regression on the collected data, the company came out with a new ‘Cherry + Sprite’ drink which was mass manufactured later instead of just having mixing options in the vending machines.

Source: Sprite website

Another example is the story we started with. The milk developed using deep learning models. NotCo scientists fed Giuseppe (the ML model) the molecular characteristics of cow’s milk. The model was fed a database for plant-based ingredients that combine to replicate the physical and chemical properties of milk. Some of the ingredients were completely out of the box e.g. pineapple juice, cabbage juice, and coconut.

Source: NotCo

The prototypes were tested by human testers and were rated for flavor, physical feel, and visual appeal, and these were fed back to the model to fine-tune it and create the final plant-based milk. Why this is important because creating animal-based food can take enormous energy and natural resources. These plant-based replacements can hugely scale and distributed at a lower cost with healthy versions.

Food safety and management

Cleanliness is a major concern in the food hospitality industry. Many restaurants claim to be clean by saying that all their processes are automated. However, just by saying that the process is completely automated does not mean it is clean. Moreover, fully automated might also mean pieces of equipment are not touched for a long time, and hence there is a higher chance of customer’s doubt about the cleanliness of the final product. The University of Nottingham claims that AI-powered cleaning systems could save €100 M a year. The ML model, which detects the small food particles on the uncleaned vessels, can easily sort out the uncleaned ones for re-cleaning. They claim that equipment cleaning accounts for almost 30% of energy and water supplies in a food processing plant and their AI systems can save up-to €100 M along with 50% saving in time and energy.

Preparation and delivery

Now let’s come to the making and delivering the food. AI is being used from a restaurant search to food finding options available in different apps and websites. Moreover, when it comes to delivery of the cooked food through kiosks, the computer vision algorithms detect the face of the person making the order and release the food upon verification. The Natural Language Processing using the latest transformer network algorithms are being used to sort the relevant review and keep it at the top. Also, a model trained on flagging foul words from the comments is used to keep the comments free of abusive words and expressions in online reviews. Moreover, AI-powered robotics arms have already started cooking in the kitchen.

My take on it …

Food is one of my favorite industry which I follow along with healthcare and I would love to see the impacts of AI in all facets of the food industry from using AI to predict the weather to improve the farm produce to using AI to deliver the food to the end customer. Given a chance, I would love to be a part of a food company and utilize the power of AI to enhance different elements of this industry. It is not far when we will have AI-powered chefs cooking AI build recipes in an AI-augmented kitchen using AI milk. Let me start with the AI developed milk since I found it in whole foods. Till then keep learning 😊.



Deepak Singh

Product Enthusiast — Utilizing the power of AI and Design to rethink possibilities and reframe the problem statement! Website: