Restaurant ordering in the age of voice AI - SmartBrief

All Articles Food Restaurant and Foodservice Restaurant ordering in the age of voice AI

Restaurant ordering in the age of voice AI

ConverseNow's announcement of a partnership with Jet's Pizza is just the latest news in the fast-growing deployment of voice AI solutions at restaurants.

7 min read

FoodRestaurant and FoodserviceTechnology

Jeffrey Greenberg/Universal Images Group via Getty Images

Pick up a phone to order a Domino’s pizza or pull up to the White Castle drive-thru these days and the voice taking your order is increasingly likely to be the product of AI. Voice AI at the drive-thru and on phone orders is so new and evolving that industry players like ConverseNow, Presto Automation and SoundHound have either expanded on their original missions or shifted gears completely since their startup days.

ConverseNow, which was founded in 2018 and counted Domino’s among its first restaurant partners, today announced a new partnership with Michigan-based franchisor Jet’s Pizza which will make the technology available to Jet’s individual operators.

“Voice AI is a game-changer for Jet’s Pizza as it allows us to address a big concern of ours: making sure our franchisees are making good use of their human capital,” Jet’s Pizza Chief Information Officer Aaron Nilsson said in a news release. “We aim to make technology as good as our pizza.”

A solution to labor challenges that also drives sales

Voice AI technology takes orders on the phone and at the drive-thru, freeing up human employees for other tasks and giving operators an ever-improving tool to manage labor costs and increase order speed and accuracy, companies say. Vendors also tout the potential for driving sales growth through upselling, which can be hit-or-miss when left up to human employees working long shifts. 

ConverseNow’s original mission had been to “voicify” print documents like the product manuals families across the country have stuffed into a kitchen drawer, co-founder and CEO Vinay Shukla said.

The then-startup began working with some big companies but about six months into the process it became clear that the original task might be both too complex and unnecessary in the age of YouTube DIY videos and Google. In addition, it dawned on the founders that the effort could be limiting.

“We realized we were building a voice assistant for a big company, we were not building a product that could be used by everyone, that could scale to everybody,” Shukla said. 

That’s when the focus shifted to “voicifying” restaurant menus.

Speaking of sales growth 

AI can upsell customers based on what they’re ordering today and what they’ve bought in the past, Shukla said, and it also provides a rich source of data for analyzing which menu items are proving popular and worth offering as add-ons. 

And, as the technology continues to improve, it also gets better at recognizing when a customer might be amenable to suggestions for adding items and when they’re in a rush or just not in the mood.

“There’s a fine line between upselling and annoying,” Shukla said. 

If a customer has rejected a suggestion for a specific item in the past, that item won’t be suggested to that person a second time. Additionally, if the customer’s tone points to them being in a rush that day, there won’t be any upselling at all. 

Emotional intelligence is another area where voice AI technology is learning and growing, and companies are continuing to work on creating technology that can recognize human emotions and respond appropriately, Presto Automation Chairman Krishna Gupta said. 

Human employees still play a role

Human employees play varying roles in the process, voice AI companies say, from Presto’s model that uses call center employees to ensure the accuracy of every order, to ConverseNow’s backup system, to SoundHound’s method of 100% automation unless the customer specifically requests to talk to a person.

When ConverseNow first introduced its voice AI in 2019, many people weren’t comfortable with talking to an automated voice instead of a person, Shukla said.

That’s changed significantly in recent years – the rapid rise of ChatGPT and other AI tools from companies including Google have built buzz and raised awareness for the technology, increasing consumers’ comfort level for talking to bots.

That said, not everyone’s on board. For customers who still want to talk to a person, or who have questions or concerns that the AI isn’t equipped to deal with yet, humans are available.

For ConverseNow, questions that require the human touch are referred first to employees in the local restaurant, then passed along to a call-center worker with access to the full order if necessary to resolve an issue.

Other companies also make use of call centers, including Presto Automation whose voice AI technology takes orders at the drive-thru. 

The seeds for Presto were sown over a meal between two MIT friends when difficulty splitting the bill led to the idea of tabletop tablets patrons could use to order and pay and also play games during the meal, said Gupta, who was Presto’s first investor. 

The company launched amid the 2008 recession and hit its stride a decade later in the two years that preceded the pandemic,  which slammed the brakes on in-person dining and led Presto to head in a new direction – the drive-thru.

Currently, Presto works with its own agents who review most orders and interact with the restaurant partner’s point-of-sale system to ensure accuracy as the technology continues to evolve, Gupta said.

The company clarified the set-up further in a statement:

“Presto has developed a unique machine-human interface whereby all orders are handled by AI, but Presto’s own human agents oversee this process to ensure accuracy and step in for more complex cases. While some competitors may cite 100% automation, the hybrid model is materially more accurate and leads to far less intervention from the actual store,” the company said.

“Over time the AI will become more and more autonomous,” Gupta said.

And that autonomy will be key at the drive-thru where speed is an essential feature.

“It’s a dance between the AI and the humans, and you have to figure out how to do that dance,” he said.

Missteps in the dance can be costly – unlike phone orders where an additional 10 seconds or so won’t make a difference, adding seconds at the drive-thru can add up to lost revenue, Gupta said. 

Pay structures also vary, and companies offer restaurant partners the option of paying a flat fee or a per-transaction fee. 

“Big brands tend to like a flat fee,” ConverseNow’s Shukla said. “Smaller ones are used to per-transaction fees” like the ones they pay to third-party delivery companies. 

SoundHound charges a flat software-as-a-service, or SaaS, fee, said Ben Bellettini, the company’s senior vice president of sales for restaurants, adding that the merchants he works with are becoming tired of per-transaction fees, which can make it difficult to budget and forecast costs.

From the dashboard to the drive-thru

SoundHound is one of the oldest players in the voice AI field, started nearly two decades ago by two Stanford graduates who were big fans of Star Trek and believed someday people would be speaking with robots, Bellettini said.

The company built its early tech stack from scratch in the areas of text-to-speech and natural language, and focused first on applications for the automotive industry. Hyundai Motor Group was an early partner for that technology, which allows drivers to use their voices to access information and ask questions hands-free while driving.

Today, SoundHound provides restaurant partners with voice AI ordering technology for both the drive-thru, with plans to be deployed at 100 or more White Castle drive-thrus by year’s end, and phone orders which it rolled out at 50 Jersey Mike’s units this year. 

Customers have the option to have a human take their orders, Bellettini said, and a very small number still choose that, but the majority are increasingly comfortable talking to the technology. 

Bellettini is quick to point out that, whether customers opt to speak with a human or interact with the technology, voice AI isn’t there to unseat employees but to make their jobs easier. 

Partners like White Castle “do this to help their employees,” Bellettini said. “It’s not a labor replacement, it’s a tool for alleviating their stress.”

Recent related stories:


If you liked this article, sign up to receive one of SmartBrief’s Food & Beverage newsletters. They are among SmartBrief’s more than 250 industry-focused newsletters.