While hosting a hackathon last year, an employee at Levi Strauss & Co. presented a new generative artificial intelligence concept that envisioned stitching together product details about denim, operational procedures, and training materials to make work easier for store associates.
Working closely with Google Cloud and leveraging Gemini’s large language models, Levi’s Chief Digital and Technology Officer Jason Gowans moved quickly to run a limited pilot in just 10 stores in late 2025, allowing employees to ask various questions in natural language, including the difference between 501 and 505 jeans, how to process returns without a receipt, and explaining how to complete an online order that’s been fulfilled in the store.
Called “STITCH,” the AI assistant can be accessed through a tablet or smartphone and has had such a strong start that Levi’s has rolled it out to more than 70 U.S. Levi’s stores. Gowans intends to bring STITCH to even more locations, add more features, and make it available in additional languages beyond English. Levi’s reports that stores where employees have access to STITCH saw an eight-point improvement in consumer satisfaction versus locations that do not have the technology.
“That gives us some intuition that there’s real value here, and the stylists do seem to be showing up as more knowledgeable and more confident,” says Gowans.
In his role as Levi’s first-ever C-suite digital officer, Gowans is responsible for both for the underlying technology that supports the internal enterprise and the direct-to-consumer channels that include levi.com, the apparel brand’s mobile app, and a fleet of approximately 3,300 Levi’s branded stores and shop-in-shops. Growing the DTC business has been a strategic priority for CEO Michelle Gass, as margins from that part of the business are higher than those of the wholesale channel, which encompasses Levi’s sales at stores like Target and Macy’s.
The DTC business accounted for nearly half of Levi’s revenue for the fiscal year ending November 30 and has reported 15 consecutive quarters of positive comparable sales growth, according to the company.
Gowans is also keenly focused on infusing AI and other technologies into every point of the journey, beginning in product conceptualization and continuing through the moment it shows up on the retail shelf. He authorized the use of AI in the design process, demand forecasting, and price optimization to determine the right price for its goods. Gowans has also made Microsoft Copilot widely available and the company’s workers have already built more than 800 AI agents.
“We’ve seeded this by training our employees, both our corporate employees and our stylists,” says Gowans of his vision that AI implementation shouldn’t be a top-down mandate.
As a result of that broad encouragement, Gowans says that agentic AI use cases have emerged in pockets of the business that he hadn’t anticipated. One area of excitement is within SAP’s enterprise resource planning software system, which is used to integrate processes like finance, manufacturing, and supply chain management.
With half of Levi’s sales still coming from 50,000 points of sale in approximately 120 countries, some smaller retailers still submit their orders as PDF forms sent through email. With AI, Levi’s is now automating that step, rather than requiring a human to manually input all of the order details.
Another AI project that Gowans is working on this year is a so-called “super agent” that will be embedded within Microsoft Teams and act as a one-stop shop to retrieve information and take action from subagents that have already been deployed across AI, human resources, operations, and other parts of the business. Levi’s has also built an initial proof of concept and has piloted the tool with some employees, with some communication already taking place between the “super agent” and vendors like ServiceNow and Workday.
But the “super agent” doesn’t have access to all 800 AI agents at Levi’s, as Gowans says protocols around these systems—including Google’s Agent2Agent and Anthropic’s Model Context Protocol—aren’t yet settled.
“The architecture that we built at the time has since evolved, given what Microsoft and others have laid out as far as protocols,” says Gowans. “This idea of a super agent is something that we’re definitely going to keep building towards to make that employee experience as intuitive as possible.”
He is also keeping a close eye on how generative AI is changing how shoppers discover brands and make their purchases. One piloted application is an AI-enabled styling agent in the Levi’s mobile app, which is currently only available to the company’s U.S.-based employees, and which can give guidance on denim, styling, and make personalized recommendations. Gowans hopes to roll out this tool externally in 2026.
With hundreds of millions of users now on ChatGPT, Gemini, and other chatbots, a new practice has emerged called generative engine optimization, or GEO. Gowans says that Levi’s is working toward making its products available on those channels, but wants to think bigger than simply offering a website link in response to a chatbot prompt.
He also feels that agentic commerce—which is when an AI agent shops on behalf of a consumer—is further in the future for brands like Levi’s, which sells apparel that’s highly subjective when it comes to fit and style. Agentic commerce is likely to be more popular for commodities, like groceries, Gowans asserts.
“It’s a space that we’re paying a lot of attention to,” says Gowans. “I expect that we will be testing something at some point.”
John Kell
Send thoughts or suggestions to CIO Intelligence here.
This story was originally featured on Fortune.com
