The Challenge
Walmart's private label fashion and apparel design process was slow, opaque, and reliant on guesswork — taking nearly a year from trend identification to shelf. Designers made educated guesses about consumer preferences six to twelve months ahead of time, with long manufacturing lead times and limited feedback loops from the market. The result was a process prone to waste, misaligned assortments, and missed trend windows in one of the company's largest product categories.
What They Built
Human Machines rebuilt Walmart's fashion design process from the ground up using a custom multi-agent system (the Trentor engine), integrating earlier trend analysis, assortment recommendations, and market feedback loops into the design cycle, with designers actively involved in co-designing and testing the system.
Geoff Gibbins and Human Machines began by refusing the default approach of layering AI onto Walmart's existing fashion design workflow. Instead, the team decomposed the process into its core outcomes — trend identification, assortment efficiency, and speed to market — and rebuilt the entire workflow around those outcomes using a custom multi-agent system called the Trentor engine.
Designers were deeply embedded throughout the design, testing, and iteration process rather than consulted at the end, ensuring the system reflected how creative decisions were actually made and where human judgment was irreplaceable. The Trentor engine integrated AI-generated trend analysis and consumer signal processing far earlier in the design cycle than the previous model allowed, along with pricing signals and market feedback loops that had previously arrived too late to influence product decisions. Within six weeks of development, a working prototype was tested with real consumers. The solution now ships clothes for one of Walmart's biggest private label brands and is actively expanding across other product categories, having cut the design-to-shelf timeline by 18 weeks.
AI Role
What began as a six-week proof-of-concept has grown into a full organizational capability Walmart calls "Trentor." The company built an entire team and system around this engine, which now manages product design across fashion and is expanding to other private label categories — demonstrating full internal ownership rather than dependency on an external vendor.
Infrastructure
• Custom multi-agent system (Trentor / Trend to Product engine) • Trend analysis and consumer signal processing layer • Assortment recommendation model • Market feedback and pricing signal integration layer
Integration Points
• Trentor engine agents connected to trend data sources and consumer signal feeds • Assortment recommendation outputs integrated into Walmart's design and buying workflows • Market feedback loops and pricing signals feeding back into the design cycle at earlier stages • Designer review and override interfaces embedded throughout the multi-agent workflow
Fortune 500 and large enterprise organizations in CPG, retail, financial services, and healthcare/nonprofit — specifically C-suite executives, heads of innovation, and transformation leaders who have run pilots that showed promise but haven't scaled. Ideal when the client has complex multi-stakeholder processes where reinventing the workflow (not just adding automation) would unlock meaningful time, cost, or quality gains.
