Exclusive: New AI-powered drive-thru technology gains traction with backing from a16z and Arc

(SeaPRwire) –   For Mike MacLennan, CEO of a drive-thru AI firm, conducting market research involves spending time at Burger King. His company’s initial earnings came from a group of older women who mistook him for a panhandler at the restaurant.

His startup, Arc, develops voice AI for taking orders and enhancing business efficiency at drive-thrus. MacLennan and co-founder Ali Hussain—both former Square and Cash App employees who worked on payment systems—have secured $10.76 million in seed funding, according to an exclusive report. Andreessen Horowitz spearheaded the investment. Arc is currently collaborating with two major fast-food chains, each operating hundreds of outlets.

The potential market for Arc is vast. The United States is home to about 200,000 drive-thru locations, which generate around 70 to 75% of quick-service restaurant (QSR) sales. The AI-in-QSR sector is forecast to grow to $12 billion by 2034. However, the field has seen numerous failures. McDonald’s terminated its AI drive-thru test in 2024 due to scalability issues. Presto Automation faced an SEC fraud charge for misrepresenting its AI’s capabilities to investors. Meanwhile, Taco Bell is reconsidering a rollout across more than 500 sites after viral customer complaints, which included a joke order for 18,000 cups of water.

Arc contends that the prior generation of AI drive-thrus failed for a particular, solvable reason: companies deployed “80% good models trying to take out labor,” MacLennan explained. “Chapter two—which I think is just starting now—is 99% good models focused on wins for customers, wins for employees, and wins for businesses.”

Arc develops customized models for each brand by training them on thousands of actual drive-thru exchanges, which include various accents and dialects. It then conducts A/B testing, allowing operators to evaluate different model versions in real time, tracking accuracy, average order value, and speed concurrently. When one client wanted to trial a spontaneous milkshake promotion at the order’s conclusion, Arc implemented it live. The conversion rate was just 5% to 6%, whereas effective upsells usually achieve 20% to 40%. “We let the data do the talking,” MacLennan stated. “He had 20 years of experience and was going the wrong direction on that one.”

Olivia Moore, the a16z partner who led the funding round and has extensive experience in voice AI, says the team’s expertise is the key factor. “This is by far the hardest kind of environmental setup,” she noted. “The car has to drive over a sensor, trigger the voice agent, then the voice agent needs to run with screaming kids in the background, with people changing their mind in the middle of the order.” During early due diligence, she visited MacLennan’s garage, where he had constructed a simulated drive-thru for testing. “It really gave me an appreciation for the difficulty here,” she said.

Arc asserts its system correctly processes orders over 95% of the time autonomously, representing a 5% to 10% gain in accuracy. It also reports that partner restaurants experience a 4% to 5% increase in the average check size due to more effective upselling. These figures are validated through two checks: whether the AI completed the order without human intervention, and whether the prepared food matched the customer’s request.

MacLennan recognizes that the voice technology itself is nearly secondary. “Voice AI is a cool technology that’s an enabler of the data and the observability, the measurability, the testing,” he said. “It lets you take this huge piece of the economy that was operating in a bit of a black box and bring the same precision you get with highly-tuned digital channels.”

In an industry where net margins are often under 5%, such insights, along with each additional milkshake sold, quickly accumulate.

See you tomorrow,

Lily Mae Lazarus
X: @LilyMaeLazarus
Email: lily.lazarus@.com
Submit a deal for the Term Sheet newsletter here.

Joey Abrams curated the deals section of today’s newsletter. Subscribe here.