Hi Auto, the Israel-based Voice AI company, already garnered some attention a couple years ago for their work using a microphone and a camera to read drivers’ lips—allowing for decent speech recognition no matter how much noise was coming from the engine, the radio, or other passengers. Despite a global speech and voice recognition market that’s expected to reach $27 billion by 2026, the company has since pivoted somewhat, though not completely illogically, from strictly in-car technology (for auto makers like Porsche) to, wait for it: fast-food drive-through ordering.
Having already perfected a speech enhancement technology that isolates speech while simultaneously eliminating all background noise, the firm met with representatives from Lee’s Famous Recipe, a fast-food restaurant in Englewood, Ohio, just outside of Dayton, along with a team from Far Hills Development, owner of 12 Lee’s restaurants. Lee’s and Far Hills wanted help taking orders, in part to improve service time, but also due to the hiring challenges posed by the COVID-19 pandemic.
Since late February, customers are now greeted by Hi Auto’s AI order taker, which takes their orders and puts them directly into the restaurant’s register system. (It’s this last ability that has some freaking out: Hi Auto’s AI workers not only remember orders more efficiently, but the firm’s technology can be scaled up to include video, thereby enabling the AI server to recognize license plates, greet drivers by name, and recall their favorite food and drinks. The fear is that this type of recognition will lead to various and sundry Big Brother scenarios.)
Hi Auto and Roy Baharav, Hi Auto’s co-founder and CEO (and who served in the elite intelligence unit of Israel’s Israel Defense Forces), were more focused on the accuracy of their speech recognition, and that their AI order taker’s dialogue come off as natural. “Drive-throughs are frequently located next to a highway, a fire station, etc.,” explains Baharav. “In addition, people speak in a variety of accents. General speech recognition in noisy environments is inaccurate and as a result makes it hard to understand the intent of the customer.
“Dialogue with the customer has to be short and natural,” adds Baharav, “as if the customer spoke to a well-trained employee.” (As a spokesperson for the restaurant told TV station WHIO: “Employees in the restaurant will be able to listen to every transaction through existing headsets and intervene in case an issue arises during the order process.”)
Natural dialogue at restaurants, as almost anywhere, is also unique. “Customers pause, get advice from others in the car, change their mind mid-sentence, order multiple items, and aren’t always aware of the menu,” says Baharav. “The only way to address those challenges is by building a customized voice stack that is optimized for the noise environment and the unstructured dialogue.”
So far, no complaints. “Customers have no problem taking orders from the AI,” observes Baharav. “Many of them thank the courteous digital restaurant assistant.”
As more people adapt to Voice AI, and to the adoption of conversational AI in phone ordering, and with the incorporation of Siri, Alexa, and Google assistants into people’s everyday lives, it’s Baharav’s opinion that AI order taking will become standard in the next few years.
“I believe that three years from now, we will have many Quick Service Restaurant chains looking for complete solutions,” says Baharav. “Using AI in the order-taking process can help customers order their usual, get relevant recommendations, enjoy a seamless loyalty program, and pay from an already stored card.”