Let’s say you’ve bought a chicken sandwich. Based just on that purchase, what do you think is a better recommendation for another meal you might enjoy: chicken piccata, or a hamburger?

This is a typical type of problem that LA-based startup, Halla, is trying to solve with its new artificial intelligence (AI) powered recommendation platform. The new Halla I/O (which stands for “Intelligent Ordering”) software integrates with a restaurant or grocer’s existing website or app, so it’s invisible to the end user. People buying groceries or meals through those apps would continue to do so, but would receive recommendations through Halla’s algorithm.

Halla I/O does this, according to Co-Founder and CEO Spencer Price, by combining data and psychographics around how people think and interact with food. Let’s go back to the chicken sandwich example. According to Price, a traditional data science approach would recommend chicken piccata, because it’s, well, chicken.

But, “The human experience of eating food is different,” Price said. He went on to describe how the chicken sandwich is also meat-centered, covered on both sides with bread and eaten by hand. And when you think of it that way, a hamburger recommendation makes sense.

Price didn’t say that the chicken piccata was wrong, per se, but just provided it as an example to illustrate how Halla I/O’s AI works differently. Using a proprietary dataset from more than 10,000 grocery items, 20,000 ingredients, 175,000 recipes and 20 million restaurant dishes, Price said that his company has created an entirely new model and a new taxonomy that doesn’t just look at what a food item is, but also the molecules that make it up, a map of attributes linked to other food as well as how people interact with that food.

With that information, Price said Halla I/O can provide more contextual and more meaningful recommendations. On a surface level, this means Halla I/O can make very basic recommendations. If you are buying salt, it might recommend pepper. But if you put salt and ground beef and onions into your cart, Halla I/O might infer that you’re making a bolognese. Not only will it recommend more items needed to make a bolognese, Halla I/O will show you recommended recipes and automatically add all the ingredients for that recipe to your cart. If you’ve shopped with that app before, Halla I/O will know if you purchased tomato sauce recently, and automatically remove that from the recommendation.

Though Halla I/O works for both restaurants and grocers, right now Halla I/O is only in pilot programs with four grocers. Interestingly, one of the pilots was with Green Zebra in Portland OR. Instead of online cart recommendations, Halla provides shelf recommendations. For example, they may see that lots of people are buying bananas and peanut butter, and set up a physical display to take advantage of that.

Halla is at the nexus of two trends we’re following at The Spoon: personalization and shoppable recipes. If its software works as promised, not only will it deliver better recommendations, but the ability to automatically recommend and adjust cart items based on recipes could be another step towards customized meal kits.

Halla is just seven employees, and has raised $500,000 in angel funding. It’s going to need more than a scrappy startup spirit, however, as the AI-powered recommendation space has plenty of competition. Startups like Dishq, Foodpairing and Plantjammer are all running their own algorithms to deliver more relevant recommendations.

Pretty soon, grocers and restaurants will need an AI-based recommendation engine to recommend the best AI-based recommendation engine.

An earlier version of this article stated that Green Zebra does not have online ordering capabilities. The grocer does through Instacart, and we have updated the post.

Subscribe to The Spoon

Food tech news served fresh to your inbox. 

Invalid email address

Leave a Reply