Leo Garcia is serious about getting his drivers home on time.
As a regional load manager for Walmart, Garcia is responsible for hundreds of semi-trailers hauling merchandise through the Chicago area.
He also knows what it’s like to be in the driver’s seat, having spent more than five years there before landing in his desk job.
“Being on the road for five days is difficult,” he told Business Insider. “You miss your wife, you miss your kids. We have to do our utmost best to get our drivers home.”
He’s using artificial intelligence to make that journey a little smoother for his team.
Garcia said he started vibe coding after taking a Google AI certification course through Walmart’s online education portal for employees. Walmart has a similar credential program with OpenAI.
“I’ve always wanted to build, I’ve always wanted to create things, but I was lacking the fundamentals to do it,” he said. “Doing the AI program gave me those fundamentals, it gave me that knowledge.”
Taking the AI course, as well as others on data analysis, changed the way Garcia looks at problems, he said.
Soon enough, he was tackling one of the most frustrating problems he faced every day: getting drivers home at the end of their routes without leaving the trucks empty.
Garcia could send a driver straight home, but that comes with costs, most notably a metric called empty miles. It’s more efficient for Walmart to send the trucks back after a delivery loaded with other merchandise than to send them empty.
Garcia said he used Walmart’s in-house coding agent, Code Puppy, to design a tool that analyzes hundreds of available truckloads in the region that need to be picked up and flags the best five or so for him to choose from based on location, timing, and other factors.
“If I’m in the San Gabriel Valley, how are you going to get me home to Oregon?” he said as an example. “It’s a difficult question to answer without knowing all the geographies, all our vendors, all our stores, and every option. What this does is it does that for you. It quantifies everything in seconds.”
Earlier this week, Garcia said a driver was scheduled to pick up a trailer on his way home to Wisconsin earlier this week. When the driver arrived, he learned the load wouldn’t be ready for three more hours.
“He could accommodate those three hours, but he would get home three hours later,” Garcia said. “And that’s rough.”
Instead, the program found a vendor five miles down the road with a load ready to go to the same town, keeping the driver on schedule. Later, the system assigned another driver to grab the initial load when it was ready.
Like many projects built with Walmart’s coding tool, the company can evaluate and distribute employee-made ideas across the organization. Garcia said that hasn’t happened with his tool yet, but it is being tested for wider use.
Garcia said he couldn’t have imagined himself designing software back when he first walked into a Walmart warehouse at age 18.
“If you had told me that now, over 15 years later, I would be sitting here running an area, learning how to implement tools that are going to help people learn and help the company grow, I wouldn’t have believed it,” he said.
