top of page
  • StartU HQ

Halla I/O

Halla Transforms How Grocery Retailers Interact With Customers

Halla I/O is an AI, SaaS company that operates at the intersection of personalization and food; its B2B software suite allows grocery retailers to offer their customers an individualized shopping experience through customized item recommendation, smart shopping cart, and home page. I spoke with CEO Spencer Price to learn more about the future of Halla I/O and its potential to change the way customers shop for groceries online.

The Challenges with Online Grocery Shopping

When Amazon bought Whole Foods in mid-late 2017, every major grocery retailers received a wake up call. Incumbents took a look at how Amazon doomed the bookstore industry and made aggressive preparations to defend its $800 billion grocery industry. Walmart, Kroger, Costco, and Target expanded online delivery and in-store pickup, kept prices low in spite of higher costs, and most importantly for Halla, invested money into supply chains and technology improvements.

The retail giants recognized that part of Amazon’s competitive edge comes from their personalized recommendations. In fact, 35% of Amazon’s revenue come from its personalized recommendation system. According to Halla CEO Spencer Price, while a quick Amazon search for cell phone cases may result in recommendations for screen protectors and chargers, a search for bagels on a grocery retailer’s site may yield results for a Fitbit, a kid’s toy, and a drawer. The retailers were hunting for ways to compete, and that’s where Halla I/O comes in — to offer grocers their own personal recommendation technology.

Intelligent Ordering

Halla I/O (for intelligent ordering) started out as app and personalized recommendation engine for restaurants and dishes in 2016. It differentiated itself from Yelp, Foursquare, and TripAdvisor by operating to accommodate the varying preferences of a group.

Price and his co-founders Henry Michaelson (CTO) and Gabriel Nipote (COO) went to high school together and later to different colleges — Nipote studying economics at UCLA, Michaelson at Berkeley for math and CS, and Price in business at USC. Each had a unique relationship with food.

From left to right: Henry Michaelson, Spencer Price, Gabriel Nipote

“I have a ton of dietary restrictions. Lactose does not agree with me. Gabriel has a ‘food is a fuel’ mentality — if it has the right macronutrient content, he’s going to put it in his mouth. Henry is the most adventurous eater I’ve ever met. If you can’t pronounce it or you’ve never heard of the ingredients in your life before and it has 25 eyeballs, he’s going to love it.”

– Spencer Price, CEO and Co-Founder of Halla I/O

The mobile app was the intended solution to their preferences and cravings until the founders realized hey might be better off if they just “gut the core elements of the app” and license their core IP and technology.

About a year ago, Halla went full force into B2B, angling the technology to serve grocers and groceries as the most fundamental food shopping experience. What started over Skype from dorm room to dorm room lead to 800K in total pre-seed funding to date and a spot in the FoodX Accelerator (500 companies apply to 8 spots) in NYC from September to December 2018.

Personalized Item Recommendations, Smart Cart, and a “Fearless Discovery” Homepage

Price explains the current technology: if you’re looking at a brand new bagel the store is selling, and you have previously bought bounty paper towels the past five times you purchased bagels, then you have a 100% correlation between paper towels and bagels.

“When you look past the bar code, it becomes clear that bagels and paper towels don’t go together, but the best retailers can do is look at item #1 and item #2 and how often they overlap.”

– Spencer Price, CEO and Co-Founder of Halla I/O

Halla maps the product back to an abstract layer. The technology understands that the new bagel is in the category of bagel in the category of bread, and can recommend related food items such as cream cheese, peanut butter, and jam. Halla takes the customization a step further by taking the shopper’s previous browser and purchase history into account. This allows the technology to distinguish whether the customer prefers butter and jam with their bagel or lettuce and tomatoes. “If you’ve only ordered lactate milk and you click on a bagel, I don’t need to show you every kind of cream cheese — I’m going to recommend you one that is lactate cream cheese,” said Price.

Additionally, Halla offers layers of intelligence in their smart shopping cart that helps customers complete their shopping list. By recognizing that the ground beef and tomato sauce in a shopping cart means the buyer can make Bolognese, Halla will recommend at the cart level “your favorite pasta, your most ordered seasoning, minus another ingredient you got yesterday.” Just a few clicks can complete a cart. Price refers to a survey from over 20,000 US online grocery shoppers that reveals the most demanded feature as the smart shopping cart.

The last component of Halla is the personalized homepage. Leading online grocers currently display the same home page to every user. In contrast, Netflix and Amazon feature customized homepages that make recommendations to products and TV shows the user has not tried but are most likely to try.

Price reveals a simple statistic: the average grocery retailer sells over 30,000 different products, but the average consumer buys around only 260 of them on a yearly basis. To encourage exploration of new products, Halla’s homepage creates a “fearless food discovery experience,” in which shoppers won’t see products they’ve purchased before but those they are most likely to buy.

Piloting and Onboarding Clients

Halla has strategic data partners but receives most of their data from retailers they’ve worked with. “We’ve got transaction histories of tens and millions of orders, everything that’s in the basket at those retailers plus 18 million restaurant menus, a recipe partner and nutrition partner,” said Price. The startup has over 9 billion data points over these food to food relationships and customer relationships.

Halla’s API is ready to go and its team is there to help. Pilots are usually run before integration — the team receives a sample set of data to train their model before moving onto integration.

For one leading technology client that has raised nine figures, Halla’s I/O increased the accuracy of their engine by over 600%, which they measured as the “reduction in misclassified product recommendations.” With Halla’s technology, the client saw an increase in the items added to cart by 31%.

A Collaborative Lens in the Food Recommendation Space

Other AI food recommendation startups include Dishq (rebranded as Spoonshot), Plant Jammer, and Foodpairing, but Price sees more collaborative than competitive potential in the companies.

“At the end of the day, we’re working to enable grocery retailers to power personalized recommendations for their customers,”

-Spencer Price, CEO and Co-Founder of Halla I/O

From Price’s perspective, Dishq’s core offering is to work on developing the next best product with their client — “sitting down with the culinary team from food brands, using AI to inform the best product development strategy, and asking ‘what’s the best new mac n cheese flavor?'”

The others are geared directly towards customers rather than businesses; their technology provides personal recommendations for the consumer. “Halla has danced in the consumer section before and has no plans of going back,” said Price.

The key differentiator, Price believes, is their dataset and their multi-pronged approach and models. Additionally, he believes each startup is doing different things — they’ve considered one another’s solution before but have stuck to their own methods, which shows how much opportunity exists in the AI food recommendation space.

Keeping Talent on Deck

Halla’s technical advisor, who was previously the head of cloud computing and data science at Ebay, recently joined Halla full-time after leaving his workload as the leading computer scientist at Harvard Medical School. “Moving forward our biggest challenge is rockstars like him, how do we keep the current team as incentivized and excited as possible, but at the moment I don’t see that a problem,” said Price. “Our challenge is going to be communicating that what we’re doing is special and incredible to the people we want to work with most.”

By the end of the calendar year, Price hopes to have commercialized one of their major pilots into a client relationship. There is one, in particular, he is excited to announce as it would allow Halla to power their personal recommendation functions in-store and online.

Halla is in the midst of a closing their seed round. Their current investors include E&A Venture Capital (Edward Bernstein & Adam Haber) and SOSV.

bottom of page