The restaurant voice AI provider Kea is named after a species of parrot from New Zealand.
It’s a clever name for an AI voice company: Kea’s job is to listen to customer telephone orders and “parrot” them back to a restaurant’s POS system.
But in Kea’s first seven years in business, it did not really play up its avian namesake.
Like many other restaurant technology suppliers, it adopted a safe color palette of blues and soft purples—a sort of signal to prospective customers that it was high-tech.
The cookie-cutter approach worked well for Kea, at least for a while. It landed some big chain clients, including the massive Domino’s. But big clients can be risky and demanding to serve. Kea had to turn some away over the years because of the cost. And so recently, it has shifted its focus from the very biggest fish to smaller businesses—even single-unit operators who may not have had access to this type of technology before.
But in doing so, it realized it needed to look and talk in a way that would connect with that audience.
“The reality is, restaurant people are not tech people,” said Kea founder and CEO Adam Ahmad. “They’re in the people business. They want to have a friendly, easygoing demeanor.”
So Kea rebranded with that in mind. It replaced the corporate-approved purples and blues with bright green and orange. It changed its logo to a cartoon parrot holding a phone to its ear, and it put that parrot all over its website, which now bursts with animations, an interactive demo and even a live kea cam from a zoo in upstate New York.
Kea has made its parrot mascot a big part of its brand. | Courtesy of Kea
The result is that Kea feels much more like a consumer brand than a B2B software service. But Ahmad believes that will help the company stand out in what is becoming a crowded market for restaurant voice AI. As the technology gets easier to deploy and the cost goes down, almost anyone can get into the game now.
“You could have feasibly two really smart college kids try to go and build this and get to some level of accuracy that might be acceptable,” Ahmad said. “And so then, where is the uniqueness? It then comes down to the brand.”
Kea’s rebrand is more than feather deep: It coincided with a significant change to the company’s business model.
Kea started as a call center business. In late 2017, Ahmad was working with a restaurant labor technology supplier, and, while attending a Wingstop conference, he learned that the chain was missing out on significant revenue because 30% to 40% of its phone calls were going unanswered.
“It kind of hit me like a ton of bricks,” Ahmad said.
He went home to San Jose and asked his local Wingstop if they’d let him work the phones. He had calls forwarded to his cellphone and started taking orders from home. He brought on a couple of friends and they set up a makeshift call center at his dining room table. After two weekends of manning the phone lines, he had delivered $2,000 in weekly incremental sales at five locations.
The Wingstop operator called him. “He’s like, ‘You gotta do something about this, even if it’s manual. We’ll be your first customer. In fact, we’ll even be your first investor in the company,’” Ahmad said.
He quit his other job and started Kea. The startup soon developed machine learning to automate calls, with human agents as a backup—a common setup in AI known as human-in-the-loop.
In 2022, the year ChatGPT burst onto the scene, Kea began exploring more advanced large language models to supercharge its product. But the technology was far too expensive then: The cost for Kea to handle one order would have been about $16.
But the cost came down quickly as AI advanced, and today, the price of an AI-powered phone call is just pennies. So about a year and a half ago, Kea switched from an AI-human hybrid model to a fully AI-driven system. It lost some customers along the way, including Domino’s, which wanted to maintain the call center piece.
“They will still rather do that than AI because of everything that’s associated with it, the stigma, and also for the accuracy component,” Ahmad said. “That helped inform the strategy of what’s now live today, which is, let’s not just build for these large enterprises. … Let’s build a real product that will apply to many, many, many logos that actually want this technology, and they’ll embrace it. They need it.”
Domino’s accounted for 90% of Kea’s revenue, and the loss could have been devastating. But Ahmad says the gamble paid off: Going all in on AI has made the product much more scalable, and Kea’s business has quintupled since it made the switch. Today, it works with more than 40 brands, including Via 313, &pizza and Hopdoddy Burger Bar.
Kea's AI can be customized with different voices. | Courtesy of Kea
With human backup out of the picture, Kea has had to hone in even more on accuracy. It took two and a half years of heavy engineering to get to an accuracy rate of 99.3%, from 98%, which Ahmad says is enough for midmarket brands, especially on the phone, where there is less pressure to complete the order quickly.
But giant chains like Domino’s, Taco Bell or McDonald’s expect perfection, which might explain why voice AI has been slower to catch on at that level.
“I just don’t think they’re ready,” he said. “They have different sorts of priorities in mind. They’ve got to protect their brand.”
Kea has not ruled out working with bigger fish again in the future. But for now, it’s focusing on smaller restaurants. And its bird-forward rebrand has proven to be a good conversation starter for those potential clients, who tend to light up when they see the new website.
“Even today, we had a couple of demos, and I show the main page, and there’s this big bird with four phones in its hand. It’s like, ‘Wow, that’s kind of cool,’” Ahmad said. “They’re talking to other companies too, but we always get the kind of reaction, which I think is necessary for the future, right?
“If there’s 50 players out there, three, four years from now, which one are you going to choose?”