How Levi Strauss & Co. is refashioning its brand with AI

Multiply that complexity across 50,000 retailers in 120 countries and the challenge of navigating data grows exponentially. Levi's Red Tab Loyalty Rewards Program alone consists of 40 million members, each with distinct size preferences, purchase histories, and style affinities. Layer on a century's worth of shifting tastes, and what emerges is a vast, largely untapped well of brand wisdom. But this treasure needed a radical refresh. Since 1853, Levi's has built its empire on understanding fabric and fit-not algorithms.

"What our products meant to miners a century and a half ago isn't the same as what it means to our culture today," says Jason Gowans, Levi's chief digital and technology officer. That's why Levi's partnered two years ago with Microsoft and Copilot to build agents that unify data across more than 400 technology systems, from the sales floor to the C-suite. In HR, finance, UX, and beyond, teams are building agents to modernize workflows and unearth insights that were previously out of reach. "From designing a product to the moment it shows up on shelves, AI and agents will touch every step of our digital transformation," Gowans says. "It unlocks a pace of innovation that wasn't previously possible."

Seeing an opportunity, Shirk set out to try something entirely new. One Friday afternoon last May, he built a custom agent in Copilot Studio. He named it Minerva, for the Roman goddess of wisdom, and trained it first on a handful of customer satisfaction, or CSAT, reports. Shirk gave Minerva guidelines in Levi's own language: Call customers "fans" and store employees "stylists." Ground every answer in Levi's data. Never guess-always cite the source with a reliable link. What emerged was like a living archive fluent in Levi's history, systems and language, ready to surface answers at a moment's notice.

The payoff was immediate. How have CSAT scores evolved over the last two years? Minerva returned trends with trusted sources attached. What checkout improvements have we tested before that fell short? It found forgotten experiments, like a loyalty sign-up screen that proved unsuccessful. How does the in-store experience differ between Levi's in the U.S. and Levi's Europe? Minerva compared regional studies on peak hours, staffing, and sizing preferences.

Before Minerva, answering those questions meant downloading individual PDFs, scanning through presentation decks and hoping no one had missed a crucial insight in slide 47. Now Shirk could query months of data and get cited responses in seconds. "As someone newly onboarded, this was an instant efficiency booster and a way to gain knowledge faster," he says. "It was a real wow moment."

This is what a Frontier Firm looks like in its early days. It's not a polished five-year rollout unveiled from the executive suite so much as a living laboratory where experimentation emerges organically across the organization. Levi's leadership deployed Copilot company-wide and stepped back to observe what employees would create when empowered to solve their own challenges. Each initiative begins modestly, demonstrates its value and expands through word of mouth. While fully autonomous agents handling entire workflows remain on the horizon, there's palpable enthusiasm today about how these tools can infuse a heritage brand with the agility of an AI upstart.

"Over 170 years, our products have evolved," Gowans says. "How we work has to evolve, too."

A $10 billion goal supported by agentic AI

Walk around Levi's global headquarters at Levi's Plaza in San Francisco and the transformation is hard to miss. Across departments, agents are analyzing which products Levi's fans buy together, and even fine-tuning the employee handbook-HR is buzzing about a new onboarding assistant called Ask Ben (as in benefits).

Each of these small breakthroughs ladders up to a larger ambition: pushing Levi's from $6.5 billion to $10 billion in annual revenue. But hitting that mark will take more than selling more jeans. It means rewiring a company born during the Gold Rush-one that's dressed everyone from Albert Einstein to Beyoncé-so every decision, from automating invoices to predicting demand, happens with the same precision and swagger that defined its style in the first place.

And innovation is bubbling up in unexpected places. As Levi's VP of Finance, US and Canada, Lisa Stirling never imagined she'd be designing her own digital workforce. When she joined the company a decade ago as assistant controller for the Americas region, finance operated the way it always had: controlled, audited, methodical. Change was hard because accuracy was sacred. "You didn't experiment with processes that auditors would scrutinize," Stirling says.

This past September, after hearing colleagues share their AI wins and seeing leadership encourage experimentation, Stirling decided to test whether AI could identify automation opportunities hidden in the department's sprawling collection of standard operating procedures-the step-by-step manuals that govern everything from invoice matching to month-end close. She picked five SOPs herself, and spent two days reading through them, around 15 hours total, looking for tasks that agents could handle. It was tedious work: analyzing screenshots, tracking decision trees, determining whether steps required human judgment or just adhering to rules.

Then she asked an agent to do the same analysis across all 1,100 SOPs in the finance department's library. That's when something stunning happened. In one day, the agent tackled what would have previously taken multiple people nearly a year to complete, Stirling says. It cataloged 18,000 individual tasks, classified them by complexity, identified which could be automated, and generated a dashboard with pie charts breaking down manual work versus approvals versus decisions. Rather than fearing that this type of automation was coming for her job, Stirling saw the results and thought: Now I have a lot of work to do. "The agent had just sized an opportunity I'd assumed was impossible to quantify," she says.

Since then, Stirling has been rethinking how finance divisions should approach their work. She's convening an internal finance summit where she'll bring teams together with all their reports so they can demonstrate what AI can do live. The point is to pose a fundamental question: How can agents streamline the work that keeps finance professionals from doing what they were hired to do in the first place?

"What I love about agents is, you're not exhausted by data by the time you get to the 'what went wrong here?' or 'where's the opportunity?'" says Stirling, who is now creating her own agents. She recently built one overnight after hearing a colleague mention theirs, training it to digest weekly store-level sales reports and spot patterns. "Having an agent's clean perspective keeps you super agile. Instead of doing manual work, you're getting stuff done, you're not bogged down, you're focused on supporting the business."

Many agents, one search bar

At Levi's, success is breeding more success. Agents are moving beyond retrieval into real decision-making. One monitors inventory and automatically triggers reorders. Another gauges regional demand and adjusts pricing in real time. Another detects quality issues in supplier shipments before products ever reach distribution.

But as the number of agents multiplies, another challenge emerges: remembering which agent does what. Levi's is now developing a super agent-one interface that handles the routing behind the scenes, and serves as a first stop for any AI-assisted task. "Employees ask their question in one place, and it connects with the right agents and returns the answer," explains Michael Womack, Levi's longtime head of global infrastructure and end user services. "Our objective is to drive all the complexity out of IT so we can have data at our fingertips and move faster than we ever have before."

There's no playbook for AI today. It's a new frontier. But patterns are emerging. Frontier Firms don't layer AI on top of existing work; they rethink how work gets done from the foundation up. They uncover hidden bottlenecks, track what matters, and build systems that learn from evidence, not assumptions. This is how AI stops being a flashy add-on and becomes an engine of agility and sustainable growth.

Call it legacy on demand. What began with jeans built to last has become a business built for the future. Derek Shirk-the designer who created Minerva-is already sketching what comes next: An agent that reads thousands of customer interviews and surfaces the most common frustrations. An agent that compares survey data across continents and spots confusing error messages before they reach shoppers.

It's a ground-up reinvention of how work gets done, all accomplished with intelligent keystrokes. "When you hear terms like agentic interface, it sounds complex," Shirk says. "But getting an agent up and running was easier than I expected. The hard part isn't the tech-it's being open to where it can take you." For Levi's, that's a practical expression of how a 19th-century brand adapts to the 21st.