CHOPCHOP – Smart Cooking Backed by Artificial Intelligence
The story of my startup aimed at disrupting the $2bn cooking assistance and $10bn connected kitchen markets.
In the beginning…
Around 2 years ago, my girlfriend and I were driving along Australia’s Great Ocean Road in Victoria. During our drive we placed our trust in the car’s GPS to guide us to our destination, which allowed us to better enjoy the experience and take in the stunning coastal views. This got us thinking about other life events where we wish we could outsource the thinking to maximize our enjoyment in the moment. Being food lovers, our thoughts drifted to cooking and dinner parties. What if we could be guided through the process of cooking multiple dishes, without the stress of potentially making mistakes on sequencing and timing, so we could better enjoy cooking and eat perfectly cooked meals? That was the spark. Since then, we’ve developed CHOPCHOP – a dynamic task management app for cooking – essentially, a GPS for cooking[1].
Cooking up value
The problem with traditional recipes is that they’re static and specific to only one dish with a fixed serving size. As we’ve all experienced when cooking, timings often change, we usually prepare more than one dish, and group sizes vary. CHOPCHOP’s value comes from providing users with a customized “gameplan” (step-by-step guide), optimized for their chosen dishes and serving quantities, that dynamically adjusts based on the user’s individual cooking pace. Additional services include a shopping list aggregator linked to supermarkets for home delivery, and RSVP management tools to communicate with guests.
At the core of CHOPCHOP, is a cooking algorithm – developed by two team members who are PhD quantum physicists from Imperial College in London – that replicates how trained chefs determine the optimal sequence of steps to cook a meal. This is done by drawing on a database of “time curves” for hundreds of ingredient-action combinations and employing ideas from particle physics and biological optimization. “Time curves” describe how the time required to perform a specific action (e.g., chopping) scale as the number of ingredient items (e.g., onions) increase. These curves are continually refined each time users cook with CHOPCHOP to help improve the algorithm’s accuracy for all users[2].
The ability to combine discrete recipes into one optimized sequence is what sets CHOPCHOP apart from competitors. Current competitors fall into one of two buckets:
- Recipe discovery: offline and online recipe books that help you find individual dishes but leave you to cook them on your own (e.g., Yummly and BBCgoodfood);
- Recipe following: products that guide you through a single pre-programmed recipe for a fixed number of diners (e.g., BlueApron, and SideChef).
Capturing value
During the launch phase, CHOPCHOP captures value in three main main ways:
- Grocery delivery revenue share: 6-10% commission, depending on retailer, on every order through the app;
- Subtle product placement and advertisements: Include branded products or logos in visual assets (e.g., step-by-step photos) on a targeted basis using user activity data;
- B2B licensing: Annual fee to license CHOPCHOP’s “recipe parser” (currently in development) to enable online recipes websites to provide a bespoke white-label “gameplan”. The “recipe parser” uses natural language processing to automatically codify recipes into a format that can be read by CHOPCHOP’s algorithm[3].
Looking forward
CHOPCHOP is first addressing the cooking assistance market but its ultimate vision is to become the central platform, or “brain”, for the connected kitchen, that seamlessly integrates all smart cooking gadgets. Because CHOPCHOP knows exactly where users are in the cooking process, it can ping each device when they’re needed.
In the near term, CHOPCHOP needs to focus operationally on three areas:
- User acquisition and competition: CHOPCHOP is the only dynamic cooking guidance product and, as the first mover, it needs to educate consumers and develop buy-in. From consumer testing, users typically need to try CHOPCHOP 2-3 times before they trust the product. This is a significant barrier to adoption. Furthermore, there is the risk that a new, perhaps more well-funded competitor could enter the market at any time;
- Content development: To acquire and retain users, it’s critical to grow and refine its recipe library based on market trends and user feedback. To do this, it needs to navigate legal copyrights, construct robust feedback loops, and deeply understand market trends;
- Quality control: As the number of recipes grow, so too does the number of recipe combination permutations the algorithm must compute. This requires the algorithm to be not only accurate, but also much faster.
CHOPCHOP has launched (download here: https://itunes.apple.com/us/app/chef-chopchop-smart-cooking/id1058876494?mt=8) but will it ultimately be under-done or well-done?
Word count: 771
[1] CHOPCHOP website, 2015, [http://www.mychopchop.com/ ], accessed November 2016
[2] CHOPCHOP Investor Presentation, 2015, MYCHOPCHOP LIMITED
[3] CHOPCHOP Investor Presentation, 2015, MYCHOPCHOP LIMITED
I very much enjoyed reading the idea of applying AI in cooking and could actually imagine cooking maybe the first best implementation of AI even before driverless cars. CHOPCHOP’s ability of Recipe Discovery is especially impressive! Yet I cannot help wondering the level of competition in this field , because just a while ago, I read that the Watson computer (what we discussed in TOM class this Friday) has actually already been applied to cooking along its most interesting feature: Recipe Discovery . Chef James Briscione, who has co-authored a cook-book with the Watson computer, gives an example of the sort of thing which Watson’s targeted customers can discovery with the help of Watson: Apples are usually cooked in butter. In American cooking it is most associated with Northern European styles, where cooking with dairy products is more common. In Mediterranean cooking, which typically uses olive oil in the place of butter for frying, and rarely involves apples, which can struggle in arid climates. Yet, Watson doesn’t care about any of that. In fact, Watson found that apples share more flavour profile with olive oil than butter. So Chef Watson readily recommends mixing the two ingredients, leading Briscione to discover the joy of poaching apple sous-vide in olive oil with sage as a condiment for roast duck.
I can imagine the competition level in the AI cooking industry to be high in the near future, and I would be curious to hear how CHOPCHOP would prepare itself for this market with great potential as well as challenges.
I love this idea and am downloading it! I am an absolutely terrible cook and I love the planning aspect. I constantly pick out a recipe and then go to the grocery store and forget half of the ingredients that I need so I especially appreciate the link with the shopping list and grocery delivery. My biggest concern is that it is such a crowded space and how you are going to break down the barriers to entry. I wonder if you could pair with a famous chef and have his/her recipe pairings to attract people to the platform.
The concept of being the “brain” of the kitchen is also awesome. If the app could pre-heat my oven for me at the right time and set timers that would be amazing! I think this has a lot of potential.
I love this, and like Maria I definitely need it. I wish you guys all the best!
You mention issues with content acquisition – is the idea that you would sign certain chefs up to provide exclusive content? I think you’re right that this is a really big challenge, especially because so many of the ‘blockbuster’ chefs have their own brands beyond recipes-on-the-page and must be reluctant to share their info at anything other than a very high price! I guess I might go even further and suggest that one or several of these chefs might be well placed to introduce a competitive product, that allows users to specify numbers, other recipes and timings and so on- and that, given their resources and following, is surely a potential threat. I can see, for example, an Ottolenghi, Heston Blumenthal, Ina Garten or Jamie Oliver version of this being super popular. (That just shows it’s a good idea!)
To guard against it: have you guys thought about easing customer adoption, one of your other operational focuses, by offering a cooking tutorial, from the basics onwards? Something akin to Coursera? I would think that marketing it as a teaching tool rather than a recipe aggregator and cooking algorithm might capture share where chefs catering to (ha,ha) experienced cooks and passionate foodies might not be able to.
This is a rich and delicious example of how new digital technologies can transform industries and create new opportunities. I love how CHOPCHOP actually augments the demand for recipes, groceries, and cooking appliances by empowering more people to cook at home. I also love how the audience could easily be novice chefs who need more guidance, to competent chefs looking to take on a much more complex task of cooking several courses for many people.
It’s extremely clever how CHOPCHOP handles the entire process, from recipe discover, to ingredient aggregation, to ingredient delivery, to sequencing all the cooking instructions.
Next steps:
I would argue some low hanging fruit CHOPCHOP should bake into its app:
1. Convert units to each user’s preference. (User craigers66 reviewed CHOPCHOP on the app store on Jan 3, 2016: “One thing I might add is a conversion chart for us non-metric users.”)
2. Add videos that help users learn different prep and cooking techniques. These can be sourced from YouTube for free.
Future:
Uber has used its technology to become more than a ride hailing app, and is now a logistics powerhouse. CHOPCHOP could become the ultimate planning and sequencing company for much more than cooking. The outlook is looking pretty sweet.
Awesome post, thanks a lot!! I’m sure that many things can be done to improve the app and keep it growing, in spite of the increasing competition as raised above. But I want to point out a different dimension from your post – we all are in a position to capture a lot of the opportunities created in the digital world. Thanks for showing that it can be done!
I’m curious to hear about the main challenges you faced to set up the app, especially regarding it’s AI. How hard is it to develop the necessary algorithms, in terms of skills required, time, and cost? How do you manage to enable the self-learning with a strong feedback loop? And on a whole different aspect – how do you keep these algorithms safe, and for your own use only?
Thanks!
I love this idea as well, thanks for sharing! ChopChop demonstrates how digital technologies can transform an industry, creating huge opportunities for both consumers and the company – but also many of the challenges that companies face when they are first movers in a space, and introducing a “disruptive” new technology. I’m excited to see where you go! It seems like you have huge potential for growth, and the main challenges ahead (apart from continuously improving the technology and algorithms used) is racing the clock.
This post also got me thinking about other ways in which AI can be used to enhance the cooking experience. I think ChopChop has already hit on a lot of the challenges that people face in the kitchen, particularly with sequencing. I would also love a tool that helps me adjust recipes for the number of people I want to cook for – i.e. the recipe says it makes enough for 8 people, but I only want to cook for 4. I would also love to be able to input my own recipes. For example, I may be cooking for a group of people and I’m happy to source most of the recipes from ChopChop, but maybe I find a recipe somewhere else that I really want to try or have a family recipe I want to add to the menu too. The ability to enter my own recipes, and let ChopChop help with the sequencing would be amazing. I wonder if there might also be a crowdsourcing benefit there…many other companies have benefited from crowdsourcing as a way to gain new users but also to gain new content at minimum cost.
Excited to see where this goes!