Even though Frank Pica has spent much of his career in tech-enabled marketing and advertising, he realized how difficult it is to understand consumer behavior online after looking at himself in the mirror.
"I took a look at myself, and I was like, 'Man, if I was a freaking company trying to understand my preferences from one day to the next, I'd f------ quit,' " Pica said.
But Pica didn't quit. Instead, he launched a company that uses artificial intelligence to try to gain a deeper understanding of why people do what they do online when looking at food and beverage CPG products and produce. Pica serves as CEO of Native.AI, the company he founded with Sarah Sanders, who is the startup's chief operating officer.
Native.AI uses technology known as natural language processing to try to parse the sentiment and meaning of things that consumers write about products online — on reviews, surveys and online comments. This information comes directly from consumers, meaning it's not filtered through retailers or online services, and can be extremely valuable. But by definition, it doesn't necessarily have a logical structure and requires attention to detail or sophisticated technology to make it useful.

Native.Ai recently announced a $1.75 million pre-seed funding round with investments from current and former leaders at top companies and government agencies including Blue Apron, Kellogg and the U.S. Department of Agriculture. It also hired veteran data, solutions architect and cloud computing programmer Mike Jackson as chief technology officer.
Native.AI's goal is to create a system that both automatically turns consumer-created data into something useful and provides more opportunities for shoppers to provide direct product feedback. Pica said that getting this information from consumers to manufacturers can be a challenge. But directly processing consumer feedback and giving it straight to the manufacturers can help companies create better products, packaging and formulations.
"Our job is to simply bring those two endpoints closer together," Pica said.
Industry needs
Pica and Sanders met in 2017 as they were doing consulting work with agriculture companies. Sanders had realized that there are several intermediaries between the companies that grow the produce and the consumers who buy it. Growers work with distributors, who bring that produce to retailers, who sell it to end consumers. Sanders said all of these layers make it difficult for produce companies to get fast and accurate consumer feedback on their products — the type of thing that is vital for an industry that requires quick delivery and turnaround.
Taking a wider look at different types of products — ranging from CPGs to wellness products to up-and-coming entrants — showed that this issue of slow consumer feedback filtered through many sources was not unique to fresh produce, Sanders said.
"All these brands seem to have a similar problem," she said. "We actually ... started to build different tools to help them at the source of production, but realized consistently the large opportunity with it was to create tools that better help who's producing our goods and the end consumers of them. Most of them [existing tools] have such blind spots and rely on tools that help them find [quantitative] data, such as category growth, or just their sales within specific retailers. But we really lacked information on the qualitative side of data."

Sanders, who worked with cloud-based waste and recycling solutions company Rubicon and for cloud-based restaurant management system Toast, has deep experience in building and scaling a tech solution for industry.
Pica has similar experience with tech-based solutions companies. He previously was director at AI-enabled digital advertising platform LockerDome, and led the U.S. expansion of European AI-powered ad tech platform Adyoulike.
Pica and Sanders are bringing their experience leveraging technology to solve business problems to a customized solution for food companies. They said their technology can bring insights to any size of company, including smaller startups that may not have a large digital team but instead rely on consumer reaction to shape the trajectory of their business.
In terms of early clients, Native.AI is starting small. Current users include greenhouse operator AppHarvest; dessert specialist Milk Bar; and gluten-free snack maker Every Body Eat, Pica said. There are others with confidentiality agreements, he said.

Native.AI is getting a lot of support from big players behind the scenes. Beyond the funding round in June, the company is announcing its membership in Mista — a platform with a curated ecosystem of companies to provide ingredient, scientific, data and sustainability solutions to CPG companies. Large businesses including Danone, Givaudan, Conagra Brands, Mars and Ingredion are Mista members. Pica said that being a part of the Mista ecosystem will help Native.AI gain exposure and work with companies of all sizes.
"We went through a pretty rigorous selection process, and they've identified as our technology is being highly useful within those organizations," Pica said.
Mista says it welcomes technology solutions like Native's.
"Today’s consumers are rapidly changing the way they eat, shop and live," Scott May, vice president of innovation at Givaudan and head of Mista, said in a written statement. "Native is helping brands stay ahead of trends by transforming how they listen to and act on feedback. The insights are being used to ensure the future of food is customer centric, innovative, eco-friendly and sustainable. We’re looking forward to supporting Native's growth and providing the team with the connections and resources needed to thrive."
Creating (and using) better data
While there are many places where consumers can leave written comments about products, Native.AI is also creating more direct opportunities for them to provide feedback.
The company is using scannable QR codes on packages to lead to surveys or comment forms. While QR codes flopped in the United States when they were first introduced about a decade ago, the tech is making a comeback as a way for consumers to get in-depth product information. They are deployed as part of the Consumer Brands Association's SmartLabel initiative, which can provide nutritional, sourcing and allergen information with a quick scan. They are also a way a brand can disclose bioengineered ingredients under the GMO labeling law.
Pica said QR codes seemed to take off again in the early days of the pandemic, when consumers were leery to touch products on shelves. Through QR codes, they could get a wealth of information through a quick scan.
"Most of them [existing tools] have such blind spots and rely on tools that help them find [quantitative] data, such as category growth, or just their sales within specific retailers. But we really lacked information on the qualitative side of data."

Sarah Sanders
Co-founder and COO, Native.AI
Adding a survey to a QR code on a label creates another way to get data directly to a manufacturer, Pica said. The survey can speak directly to product attributes as well as demographic information, which companies can use to better target their offerings. But QR codes have another use, Pica said: connecting consumers to a subscription form so they can get products delivered through a direct-to-consumer channel.
"That's, I think, an enormously powerful tool for these guys, especially in a world right now where they're constantly ... dealing with the middleman from a variety of tech and/or retailers," Pica said.
Going forward, Pica and Sanders would like to add point-of-sale data to their platform — which also includes data scraped from publicly available places like retailers' websites. Pica said the company is in talks with data providers.
In the future, Native.AI may want to expand to other areas with active online consumer bases, like fashion or nutraceuticals, Pica said. However, the company wants to take it slow to ensure it doesn't expand beyond its capacity or ability to provide useful insights, he noted.
"We have infinite ability to to scale, but you also have to have the capital and the resources to do so properly," he said.
Correction: This article has been updated with a clarification from Sarah Sanders about the data existing tools provide. They could provide quantitative data, she said.