How Cox Automotive Leaders Used an AI Sprint to Accelerate Product Development

Why Cox Automotive Invested in an AI Sprint

Leaders from Cox Enterprises’ automotive division, Cox Automotive, recently made a bold business decision. They pulled a select group of employees from their day-to-day responsibilities for 30 days and challenged them to prove what was possible if teams used agentic AI to build agentic AI solutions.

“We are committed to leading the way with AI and not sitting back waiting for others to pave the way,” said Senior Lead AI/ML Engineer Daniel Brasuell. “We know AI is moving fast and so are our competitors, so we wanted to focus on making sure we laid the groundwork for not just one product, but for many products.”

Brasuell was one of several engineers, product leaders and technologists who participated in the focused AI sprint, which began in late summer 2025. The goal was to rapidly develop proof of concepts while establishing a scalable foundation for future innovation.

Several projects from the 30-day sprint advanced into production, ranging from consumer-facing AI experiences to internal workflow automation. Within months, the teams built and refined new AI capabilities and solutions that were showcased at the 2026 National Automobile Dealers Association Show in February. There, Cox Automotive leaders discussed market-shaping trends and demonstrated how the organization is driving innovation in its field.

“The ambition was to prove that Cox Automotive could lead with AI in a meaningful way, on a compressed timeline, using internal talent,” Director of Machine Learning Udit Luthur said.

 

What It Was Like Working on the AI Sprint at Cox Automotive

Brasuell, who focuses on researching, building and implementing AI into products and product development, said that the company leveraged industry-leading language models, making it easier to move quickly during the AI sprint. He and his teammates also relied on cloud-based tools to support development and experimentation throughout the process.

In addition to its cutting-edge technology, Cox Automotive’s AI sprint stood out for its distinct approach to product development. Luthur, who leads data science practices across Cox Automotive’s logistics, retail and consumer divisions and helps guide the evaluation of generative AI systems, said the process proved just as significant as the solutions that emerged from it.

“The mandate was to invest heavily upfront in writing thorough, detailed requirements, and then use AI tooling to accelerate the actual build,” Luthur said. “That sounds simple, but it was a real shift in how development work gets sequenced.”

Teams focused precisely on what they were building, what “good” looked like and where the edges were. From there, AI helped compress the distance between a solid spec and a working system. This gave teams a real-time understanding of what that working model could actually do, highlighting where human judgment was still essential and what needed more care than expected.

“It was as much an experiment in how to build as it was in what to build, and that intentional structure from leadership is a big part of what made the pace possible,” Luthur said.

Monica Le Blanc leads product for Cox Automotive’s Agentic Marketplace team, which owns the consumer conversational AI experience across Autotrader, Kelley Blue Book and Dealer.com. Her team is responsible for innovations like AI Mode, the Autotrader ChatGPT app and the embedded AI assistant shaping how consumers navigate the car-buying journey.

While she wasn’t part of the original proof-of-concept sprint, Le Blanc joined the initiative shortly after, stepping into a dual role as the Agentic Marketplace team came together. She helped keep cross-functional teams aligned while also bringing her buying knowledge to the group shaping the product and product design approach for the shopping journey. 

At the same time, early contributors from the original proof of concept shared their key insights, passing along the context and product instincts that allowed the broader team to move quickly and stay focused on what customers needed.

“Cox Automotive’s leadership made a bet on AI that changes everything about how we serve consumers and dealers, and they backed it with real investment, real autonomy and real cultural change,” Le Blanc said.

As the work progressed, the team also began rethinking how they built. They adopted spec-driven development, shifting the way the work had traditionally been approached. 

“Spec-driven development changes what a product person spends their time on — less typing of acceptance criteria and more thinking about what should actually be built and why,” Le Blanc said.

 

 

Cox Automotive’s Strategy for Building Responsible AI

Given the scale of Cox Automotive’s product portfolio and customer base, “move fast and break things” wasn’t an option during experimentation. Product and engineering teams had to approach every new capability with strategic consideration, ensuring that security, reliability, resiliency and customer experience standards were maintained.

“AI is moving fast enough that a six-month discovery cycle gets you a beautifully researched answer to a question that no longer matters,” Le Blanc explained. “Rather than spending months in discovery, we get features into production and let the real world tell us what’s working. We develop to discover, not the other way around.”

For Luthur, the biggest risk the teams took during the AI sprint was the operating model itself: a requirements-first, AI-assisted development approach across multiple live projects under a hard deadline that produced real products that would eventually reach customers.

“We were rethinking the way software gets built while actively building software,” Luthur said. “There’s no risk-free version of that.”

Given these risks, governance was key to Cox Automotive’s AI sprint strategy and was embedded into how every system was built.

“These weren’t tools with low stakes — they were systems that would interact with customers, process real documents and inform real decisions,” Luthur said. “That raised the bar on how carefully we worked.”

Before development began, specifications were thoroughly reviewed, code was closely examined and testing was used to validate that what was built matched what was intended. 

“The spec-first approach that defined the sprint wasn’t just about speed; it created a paper trail of intent that made it easier to verify correctness at every step,” Luthur said. 

Meanwhile, from an evaluation perspective, the team was always focused on answering the question, “Is this system doing what we think it’s doing, and how do we know?” This kept responsibility at the forefront throughout the entire sprint. 

“Building mechanisms to answer that question honestly — rather than just assuming the answer was ‘yes’ — is what responsible AI looks like in practice when you’re moving fast,” Luthur said.

 

How Leadership Supported Cox Automotive’s AI Sprint

Cox Automotive product and technology leadership support was critical during this AI sprint.

Le Blanc explained that leaders set clear priorities and trusted teams to figure out how to accomplish every element of the AI sprint.

“We get pushback when needed, but we don’t get micromanaged,” Le Blanc said. “It’s easy to talk about leading with AI, but it’s harder to actually get out of the team’s way and let them do it. Cox has done both.”

 

 

How Cox Automotive’s Culture Supported the AI Sprint

The AI sprint reflects Cox Automotive’s willingness to make a real bet on its people. Leadership framed AI as a business decision, not just a technology investment. The company’s leaders allowed internal teams to discover a new way of working — and trusted them to figure it out.

“That says something meaningful about the culture,” Luthur said. “It also says Cox is in a genuine learning posture around AI — not just adopting it as a talking point but actually wrestling with the hard questions about how to build it responsibly, how to evaluate it rigorously and how to scale it thoughtfully across a complex organization.”

That doesn’t mean the sprint was perfect. Teams discovered what didn’t work along the way, and their learnings informed decisions for how to approach projects in the future.

“The organization had the appetite to try, the humility to learn in real time and the ambition to show up at one of the industry’s biggest stages with something real to show,” Luthur said. “For someone who thinks deeply about how AI systems should be built and evaluated, that’s a compelling place to do this work.”

As AI disrupts the automotive space, Cox Enterprises is empowering employees to take on the most exciting challenge of their career while building the future of automotive technology. 

“Companies have a choice to make: You can sit back and watch it happen to you, or you can sit in the driver’s seat and decide what you’re going to disrupt, when and how,” Le Blanc said. “Cox Automotive chose the driver’s seat.”

 

Frequently Asked Questions