A Heritage Retailer Bets on Agents
Gap Inc. is undertaking a substantial reengineering of how it markets its portfolio of Old Navy, Gap, Banana Republic, and Athleta, and the effort is a useful case study in what AI-led transformation looks like at a large, established retailer. Rather than bolting a chatbot onto an existing stack, Gap is rebuilding its marketing organization around AI from the data layer up, beginning with its owned channels. The ambition, in the company's words, is to use AI to personalize customer experiences, enhance interactions, and accelerate campaign delivery, which is the kind of broad mandate that signals an operating-model change rather than a tooling upgrade.
Senior Vice President for Marketing Shared Services Damon Berger framed it as bringing together the creativity of our teams with the power of AI, removing silos, unlocking better data, and building a marketing model that learns, adapts and improves with every customer interaction. We pay attention to that phrasing because removing silos and unlocking data are the unglamorous prerequisites that determine whether any AI marketing initiative actually works. The flashy agent demos are easy. The data foundation underneath is where these programs usually succeed or fail.
The Three-Partner Stack
Gap's architecture leans on three partners, each occupying a distinct layer. Google Cloud provides the foundation, both the unified data platform and a suite of agentic and generative tools including Agent Studio, Agent Engine, the Gemini models, and the Nano and Veo tools for content creation. Zeta Global supplies an intelligence layer through its Athena system, and Publicis Sapient connects marketing functions with consumer behavior analysis. This is not a single-vendor bet. It is a deliberately composed stack that spans infrastructure, intelligence, and integration.
The multi-partner approach is itself instructive. Large enterprises increasingly assemble AI capabilities from several specialists rather than committing wholesale to one platform, balancing best-of-breed capability against the integration complexity of stitching vendors together. Gap is wagering that a unified Google Cloud data foundation can be the connective tissue that keeps the other pieces coherent. That is a reasonable bet, but it also concentrates dependence on Google as the substrate, which is the kind of strategic lock-in that technology leaders should weigh carefully even as they chase the near-term capability.
Why Owned Channels First
Starting with owned marketing channels is a smart sequencing decision. Owned channels, the brand's own email, app, and site, are where a retailer has the most data, the most control, and the least regulatory and platform risk. Cutting teeth on AI personalization there lets Gap learn and iterate without the added complexity of third-party advertising platforms or the privacy minefield of paid media. It is the lower-risk proving ground before extending AI-driven approaches into channels the company does not own.
This phasing reflects a maturing understanding of how to deploy AI in marketing responsibly. The retailers that have stumbled often did so by aiming AI at the most complex, externally dependent channels first, where data is messy and outcomes are hard to attribute. Gap is doing the opposite, building competence and a clean data foundation in the environment it controls, then expanding. We read that as a sign of discipline, and it is the pattern we would advise any large enterprise to follow when operationalizing AI across a sprawling function.
Brand-Led and Intelligence Powered
Chief Executive Richard Dickson has summarized the strategy on earnings calls with a phrase that doubles as a thesis, describing Gap as a fashion company that is brand-led and intelligence powered. That formulation is worth taking seriously because it puts brand and creativity first and positions AI as an amplifier rather than a replacement. For a fashion retailer, that ordering matters. The product and the brand are the reasons customers show up, and AI that erodes brand distinctiveness in pursuit of efficiency would be self-defeating.
The tension every retailer now navigates is between AI-driven efficiency and the human creativity that defines a brand. Berger's language about combining the creativity of our teams with the power of AI acknowledges that tension directly, framing the technology as a tool for human marketers rather than a substitute for them. Whether Gap can hold that balance in practice, using AI to move faster and personalize deeper without flattening what makes each of its four brands distinct, is the real test. The strategy is sound on paper. Execution across four brands and a global organization is the hard part.
The Broader Retail Lesson
Gap's overhaul lands amid an industry-wide scramble to operationalize AI in commerce, and it offers a more grounded template than the agentic-checkout hype that dominates the conversation. The headline-grabbing stories are about AI agents buying products on consumers' behalf, but the less glamorous and more immediately valuable work is exactly what Gap is doing, rebuilding internal marketing operations on a unified data foundation with AI woven through. That is where established retailers can realistically capture value now, rather than waiting for a speculative agentic shopping future to arrive.
For technology leaders in retail and beyond, the takeaways are concrete. Treat AI marketing as an operating-model change, not a feature purchase. Invest in the data foundation before the flashy applications. Sequence deployment to start where you have control and data. And keep humans and brand at the center so that efficiency does not hollow out distinctiveness. Gap is a heritage retailer trying to become intelligence powered without ceasing to be brand-led, and how well it threads that needle will be watched closely by every incumbent facing the same imperative.
Measuring Whether It Actually Works
The hardest discipline in any AI marketing transformation is honest measurement, and it is where these initiatives most often disappoint. It is easy to ship more personalized campaigns faster, and far harder to prove that the additional speed and personalization actually drive incremental revenue rather than simply rearranging spend. Gap will need rigorous experimentation, clean attribution, and the willingness to kill tactics that do not pay, or the program risks becoming an expensive exercise in activity that looks modern without moving the business.
We would watch for a few concrete signals over the coming quarters. Does the unified Google Cloud data foundation actually reduce the silos Berger described, or does it become another platform layered on top of the old fragmentation? Do the four brands maintain distinct voices, or does shared AI tooling homogenize them? And does customer engagement improve in ways that show up in results, not just in case studies? Gap has assembled a credible stack and articulated a sensible strategy. The verdict will come from the numbers, and from whether intelligence powered turns out to mean more than a slogan.
