AI drive-thru ordering is reaching more fast-food lanes, though the rollout looks more like supervised automation than a clean handoff from workers to software. Taco Bell said in July that it had installed the technology in 890 drive-thru lanes, more than 10% of its US locations. Dairy Queen is deploying it in 25 states, while White Castle has put its system, Julia, in 12% of locations, according to WIRED.
The customer-facing mechanism is straightforward: a voice system takes an order at the speaker and can pitch add-ons. Restaurants say that frees crew members to prepare food, process payment or spend more time with customers at the window. Those are operational claims, not proof that an unattended drive-thru has arrived.
For chains, the appeal is immediate response when a car reaches the speaker, plus a consistent upsell script. SoundHound restaurant-sales executive Ben Bellettini told Restaurant Dive that its technology can engage customers immediately. He also said the company’s systems needed employee help on roughly 5% to 10% of orders, a vendor-reported figure.
Can AI drive-thru ordering work without an employee?
Sometimes, based on the chains’ accounts. Bojangles said its Hi Auto system, branded Bo-Linda, was operating at about 200 of its more than 800 restaurants. CIO Richard Del Valle told Restaurant Business that the system completed more than 95% of orders correctly without human intervention and reduced a drive-thru worker’s workload by a third. Those metrics come from Bojangles and are not directly comparable with other chains’ results.
At Bojangles, managers and drive-thru workers listen to each transaction and can take over if the system gets stuck, Del Valle said. The company reported that about 3% of orders were diverted to an employee. During the pilot, Bojangles added roughly 300 new price lookups after encountering customer requests the system had not been prepared to handle.
That is the awkward engineering detail behind the friendly synthetic voice. A drive-thru order is a noisy conversation full of local phrasing and unexpected requests. Restaurant Dive reported that Bite CEO Brandon Barton cited passing trucks and regional accents as situations that can impede recognition. WIRED also reported that systems can struggle with accents and dialects, and that chains commonly keep a headset-wearing employee available to intervene.
Why did some fast-food chains pull back?
Early tests produced enough visible mistakes to puncture the sales pitch. WIRED documented social-media clips of erroneous orders at McDonald’s and Taco Bell, but those incidents are illustrations rather than a measure of ordinary performance. More materially, Del Taco ended its AI drive-thru test in 2024, according to WIRED. McDonald’s sold McD Tech Labs to IBM and later ended its own effort.
The evidence points to a mixed rollout, not an industry conquered by software. Restaurant crews still matter when a voice bot mishears a customer or encounters something outside its prepared order flow. Some operators are expanding the systems anyway, betting that a bot handling routine exchanges can move workers toward the parts of service the bot still cannot reliably do.
This story draws on original reporting from WIRED.
