TOP NEWS

How Halla Is Using AI To Personalize Food, With Spencer Price

For this morning's interview, we spoke with Los Angeles-based Halla, a venture backed startup, focused on personalized recommendations for the food ordering industry. The company has raised $1.9M in seed funding, from E&A Venture Capital and SOSV. We spoke with CEO and co-founder Spencer Price to learn more about the company.

What is Halla?

Spencer Price: Halla is the only software company to dynamically profile human tastes so we can help people make better choices. Halla exists to help people understand a more curious world. We're an AI company that works at the intersection of personalization and food.

When you say you help people make better choices, what does that look like to a user?

Spencer Price: That's a fair question. We believe that there's nothing more disappointing than a bad recommendation. The online, retail space is full of it. Most recommendation systems in use today are primitive, and struggle to deliver recommendations that are relevant to shoppers, and require tech teams with deep specialties. They're not doing a very good job. The fatal flaw in those systems, is that they are based on human purchasing behavior, rather than taste. So, if lots of people are buying oven cleaner and apples, they will recommend that, even though shoppers are not looking for oven cleaner with apples. What makes better sense, is to recommend the purchase of pie crusts when people are buying apples. It's a much smarter way of understanding shopping behavior, using data about human tastes to deliver recommendations to grocery consumers. At the end of the day, we help retailers make smarter recommendations to their customers, which leads to better business.

What was your decision to focus on food?

Spencer Price: The customer experience with e-commerce has become less and less human, as algorithms on digital platform replace human interaction. The customer is really struggling to regain their lost humanity. Food is the perfect example. For every single person who eats, our relationship with food is one of the most fundamental and ubiquitous human behaviors. To represent that, requires irrational complexity. The challenge of any AI platform, is to understand that. Right now, the industry has relied on what people buy, not why people buy. What if AI could capture that complexity and ubiquity of the human experience? Food is a perfect data point to understand why people buy.

What were you doing before you started Halla?

Spencer Price: It's a bit of a circuitous route. I'm just 22, and both of my cofounders are now 23. We were friends from high school. We all went to Oakwood High School, in North Hollywood in Los Angeles. We ended up all going to three different colleges in the Fall of 2015. I went to Berkeley. We started Halla at that time, out of a classroom project, when we started thinking about businesses over Skype. We were really trying to solve a problem that would matter to us. Whenever we got together over breaks, or during in-town visits, we realized there was a recurring problem we would have on figuring out what to eat, and where to eat. We discovered there was nothing out there that could solve for that, there were no personalized restaurant or dish recommendation engines. There was no food recommendation engine which could accommodate for a group of people, with different dietary restrictions. As you can probable relate to, we would have these conversations where we'd discuss where to eat, and say “no, that won't work” for 25 minutes and you end up just going to the same three places every time. What we set out to do was develop something that we could use to solve that problem for ourselves, using personalized recommendation technology for food. We first packaged that as a consumer app, to recommend dishes to people and groups of people, and even If you might have a craving. What we learned, is there is so much more demand from existing retailers and platforms that sell directly to consumers, and discovery platforms like Foursquare, who wanted to user our IP as a layer, versus a standalone product. So, we stopped our B-to-C product about a year ago, and since then have exclusively been licensing our technology. Essentially, the core of our technology is being used to power recommendations, even at a more granular layer of recommending things that we buy at the grocery store.

Why use AI—it seems there are already a lot of personalized recommendation engines out there that don't use AI technologies?

Spencer Price: I'll start by saying, it's hard to see if anything requires AI specifically. In our case, it's just something that has never been done before. It's not a recommendation engine, it's a personalization engine, so it's all about tastes. I suppose there could be a team of hundreds, or tends of thousands who might individually map out potential relationships between food product popularity and individual customer affinity and diets, but that would be a daunting task. To be able to have a machine brain, for lack of a better term, train itself for that means you don't require manual work. That allows us, at the end of day, to look across an extremely large set of data. We have a database of 18 million restaurant dishes, menus, and hundreds of thousands of recipes, and even have aromatic molecules that come into play with respect to flavor profiles. There are all these different layers about learning about food, and piecing it together with people would be a daunting task. One of our key differentiators is we don't have to manually map things to each other. We're able to leverage lots of very specific data sets. It would be much harder to efficient scale and grow this if the system weren't teaching itself.

Talk a bit about your funding, and how you got to it?

Spencer Price: We started with a deck, in what was the summer after my sophomore year in college. The only thing we knew, is you sent around this set of slides which was supposed to convey what could better be conveyed in conversation. We tried that, and it opened the doors to enough conversations that we were able to prove our value to potential investors, some angel investors across the country in New York. One day, after they made their first investment, a six-figure sum, we figured out we needed to take this money seriously and not be in school. Three years later, there's no looking back, and we've now raised around $2M in total funding. We're venture backed, after going through the SOSV Food-X accelerator. It's one of the leading food accelerators in New York City. We finally completed that in September, and we were shortly back in LA, where we were founded and are based.

Where are you now in terms of deploying your software?

Spencer Price: We're currently running pilots with half o the top 10 grocers by revenue, and hope to commercialize a number of those relationships by the end of the year. Some of the pilots have made it far enough to have some metrics, to the point we're having conversations about integration. In every pilot or evaluation so far, our tests have placed first among other recommendation vendors. The beauty of the AI solution, is with every additional partner we interact with, the solution should get that much smarter and better for everyone.

Finally, what's been your biggest lesson as an entrepreneur so far?

Spencer Price: I think we've already touched on one, which is we're young guys, which might make it difficult in sales and business development. Youth and inexperience are not necessarily the most helpful tools when trying to convince some of the biggest players in one of the oldest industries in the world to work with you. Obviously, aside from bringing in another face, we realized that we would benefit from surrounding ourselves with people with a lot more experience and who know a lot more about this. We've on boarded one of our very first advisors, Rick Wedgeworth, or Chief Science Officer, who ran cloud computing at eBay, was a leading computer scientist at Harvard, and in between has done everything from automation to the Human Genome project. He has a wealth of experience with a wide variety of data type, and making brilliant recommendations, and really takes our experience and wisdom to a new level. That need was identified in the SOSV program, and we've been luck to have rick on us on our advisory board for the last couple of years, and finally had a great conversation and several months ago added him to our team.

Thanks!