It shouldn’t be that difficult. But deciding what to have for dinner can sometimes take from lunchtime to about 6pm – often including a slightly aimless internet browse as we try to find new recipes, keep things healthy and search for places that can offer the same quality for less.
What you might not realise however, is that there’s a company that’s helping you make your decision. That company is called Whisk.
Whisk was founded by Nick Holzerr, an entrepreneur and former Apprentice finalist, And while the company is just six years old, it’s powering the technology that convinces you to have steak for dinner when you visit the website of Tesco, Walmart, Amazon and even the BBC.
It’s used by over 100 million people every single month, yet Holzerr points out “it’s often not obvious you’re actually using Whisk.”
So if you can’t see it, and you don’t even know you’re one of its customers, what does it do? In its simplest form, Whisk is the tool that lets you turn your favourite recipe into a Tesco shopping list, or tailor a meal to your dietary requirements on Amazon Fresh. It’s a shopping planner that’s helping you create meal plans through your favourite nutrition site.
Effectively, the Birmingham-based company is turning the words you see on the screen into the food that then turns up at your door. And to do that it has to employ some pretty remarkable technology.
Powering the entire system is something called Whisk’s Food Genome. Think of it like a food brain that contains every ingredient in the known world, along with every nutritional and dietary fact about each of those ingredients. It then hypothetically applies that brain to the recipes on offer by, say, the Food Network and creates a shopping list that takes into account the fact that you can’t eat dairy, or that you don’t like pork.
What’s really quite remarkable though is how that brain is used when you log into your favourite food site and start hunting for recipes. When you see a recipe or food recommendation, the site is actually choosing that based on three distinct factors.
“The first is an individual’s explicit preferences: do they hate an ingredient, do they have a health condition or do they love a certain cuisine etc.” explains Holzerr.
″The second group is behaviour, so looking at what are they are looking at, what are they clicking on, what are they saving to their shopping basket etc.
“The final group is the context, so what’s around them. Location helps a lot with that, so what’s the weather like where you are right now, eg. if it’s sunny you might want a BBQ (to be super, super general here).”
It’s not just the weather either, Holzerr points out that many big chains offer hyper-localised savings. So Whisk actually takes into account if the shop near you is currently offering a sale on beef.
Whisk was created long before either AI or machine learning had matured, Holzerr explains – which created a challenge. “AI back then was a bunch of scientists at some universities in lab coats and it has drastically changed over the last two to three years especially.”
"At that point we were still going over to the US and buying chips and graphics cards and bringing them back in our suitcases."
Without readily-available machine learning technology the team were having to go to some fairly drastic extremes.“We were still going over to the US and buying chips and graphics cards and bringing them back in our suitcases but of course now it has completely changed,” he remembers.
One of the biggest new markets is in maintaining our health, and it’s here that Whisk has the potential to help consumers. The company has launched Foodient, an AI-powered dietician that helps people with Type-2 diabetes choose foods from their favourite shops based on their health condition.
“What we do want to do is go and work with all the other diabetes management services and organisations that are tackling obesity and help them deliver smart food recommendations,” explains Holzerr.
The last great frontier for the company however is the smart home. Earlier this year Samsung unveiled a ‘Smart Fridge’ with a vast touchscreen display and interior cameras that help you see what you have at home when you’re out and about.
Whisk’s technology then lets you create a shopping list based on what the fridge detects and buy the replacements straight from the screen. It’s an idea the company is already planning to take even further.
“What we’re working on with Samsung at the moment is being able to take a recipe based on your requirements and then turn the smart device on at the right temperature with a click,” says Holzerr.
While the allure of Silicon Valley is still there for Holzerr, he acknowledges that had the company started anywhere other than Birmingham, it wouldn’t be helping you decide your dinner.
“If you’re a startup - especially in the first few years when money was an issue and we didn’t have huge budgets - being able to stay here, pay ourselves a lower salary than one would need in London, was actually a really really big advantage.”
From carrying graphics cards in suitcases, to becoming a remote-working team powered by AI, Whisk’s growth has in many ways been a mirror for the growth of AI itself. Transforming from a Hollywood buzzword to an everyday technology that’s helping us take better pictures, use less energy and of course, decide what to have for dinner.
HuffPostListens – Birmingham
HuffPost wants to get out of the media bubble and tell the real story of the UK. For one week in July we relocated our newsroom to the heart of Birmingham and invited people to tell us what they care about - we listened, followed their tips, and went out and reported on what we heard. We’re also hiring more reporters out of London, starting in Birmingham. We don’t think the media has listened to people enough, so that’s what we’re doing. Listening to the stories of Birmingham, opening up our newsroom to its people and telling the real story of Britain from the heart of one of its biggest and best cities. You decide the news. We’ll tell your story. Birmingham, be heard. #HuffPostListens
Read more about the project here