For anyone wondering whether “agentic” is a term they’ll need to commit to understanding, or if it will go the way of the “metaverse,” Google made one thing very clear at NRF 2026 this week: “The future is absolutely agentic.”
That’s what Google Cloud vice president of solutions and industries, Carrie Tharp, said during a session on “The AI-Empowered Organization,” at the Big Show last week, adding that, “Now with the power of generative and agentic, we have this whole future in front of us…and that also means the future of work is changing in each of your organizations.”
The comment came on the heels of Google Cloud unveiling its Gemini Enterprise for Customer Experience, an agentic solution set to marry shopping and customer service in a single platform. It’s designed to deliver configurable agents businesses can manage across the full customer lifestyle, a “digital concierge,” it says will provide support from discovery through post purchase.
More and more, the types of experiences that have caused the friction brands and retailers are always working to remove, are going to be handed off to autonomous AI agents to act on a user’s behalf.
Think of the moment we’re in the midst of like this: When house phones rang and we answered them, there was likely little thought about whether the experience of communicating with someone could have been improved; it worked, it was fine. The same might now be true for e-commerce experiences, for the most part, they work and are fine. But that online exchange between brand and consumer hasn’t changed in 20 years and now it’s ripe for a transformation akin to what the iPhone did to the house phone. And with a change in the retail experience for the consumer, comes the corresponding change in the organization fueling that experience.
Across retailers, top agentic use cases are coming to life, with Google reporting that 47 percent of retail and consumer goods executives plan to allocate more than 50 percent of their future AI budget to agents. When it comes to the back end specifically, 38 percent of companies are already using agents for supply chain and logistics, and 66 percent are already seeing a return on their investments.
“In the digital era, we’ve all been leveraging predictive AI to do search a little bit better, to improve our forecast, but we were kind of stuck there, it wasn’t really changing the experience of the future. Then we shifted into the generative era, so brands like Puma and Swarovski and L’Oreal are using our generative models to change the way they interact with consumers, [but] still not removing all those friction points that we joyfully experience in each of our journeys.” Tharp said. “So here we are at agentic. We have lots of folks that are in the early days of agentic, maybe single agent interaction — this is where we’re using AI to think, reason and do consented actions. We’re rapidly going to be faced, with your teams and organizations, the need to move into multi-agent systems and actual agentic workflow. That is a fundamental change in the way we all think about how we do our jobs.”
In marketing that could look like writing briefs differently and changing what messaging goes to individual customers, Tharp explained. In service, it may be interacting with customers differently but tapping agents to support associates in facilitating those interactions more effectively.
“The sky’s the limit and we need to evolve both sides of the house at the same time,” Tharp said. “What this means is that each of these journeys, the consumer journey and employee journey, becomes more complex over time, and the tools that you now need and the mindset around your roadmap is rapidly evolving.”
As Google develops its Gemini AI model and its Nano Banana Pro for things like image generation and editing, and one AI development rolls out after another, companies now need to think about where and how employees will interact with agents, how to get their most AI-resistant team members on board, and what leadership should be doing to drive that change.
Making AI a company-wide shift
For Catalyst Brands, parent company of labels like Brooks Brothers and Nautica, its company-wide AI journey started with 200 employees it identified as change-drivers in the organization. They came from finance, marketing, merchandising, technology and then HR and their job was to play with AI, embrace it, understand that it wasn’t coming for their jobs but rather to help them succeed.
“We got a lot of great results and innovative things from people outside of technology,” said Philip Glebow, Catalyst’s chief enterprise architect & VP of data, AI and MarTech. “I think as we scale this up though, we need a little bit more structure.
“We’ve gotten gains in terms of personal productivity but those are soft financial gains; they’re not the hard financial gains that the CFO of the organization is looking for…AI today is still a software project so we resource it as such, we manage it as such,” he said. “For now, I think it’s really important to settle on a couple of use cases you think will flow through to the bottom line and manage those as software projects and really measure them. But when you roll it out to humans, it’s a change management exercise, and if you nickel and dime and measure every moment, you’re just going to kill that innovation. So, you really need a two-speed approach.”
As many companies are, Catalyst is working to improve its data foundation as it gets its business on board with AI, making sense of that information in order to put it toward delivering more informed and insightful agent-supported consumer experiences.
Initially, Glebow said, “We just threw data at [the AI system] and it didn’t give us quite the right answer. We had to take a step back and say, ‘Wait a minute, we have to predigest some of this data and think about this a bit differently.”
AI systems are not deterministic, he added. “People have the expectation that if I add 2 + 2 I get 4, so finance people were very reluctant [to use AI], and I think what that points to is that it needs a semantic layer so that LLMs and other services really understand the data. [For example] when I say ‘gross margin,’ what does it mean for me in my context?…Really understanding that content is a problem that we’ve yet to solve and I think that’s going to aid adoption once we do. If I can go into an LLM and ask a question and get a reasonably reasoned answer that’s explainable and that’s grounded in our data, that’s going to be very helpful.”
One thing that has proved a beneficial use case for AI at Catalyst — beyond transforming software development in terms of ease and speed of getting things done — is what it has meant for marketing. The team has had a host of unexpected changemakers embrace visual AI models to place a garment at a wedding versus a graduation, for example, and test out what that means for messaging around the product. Gone are the photoshoots and manual labor it would have taken to get that done in a pre-AI world.
“The speed at which this happens is phenomenal and we’ve had some great success within marketing and a lot of interest,” Glebow said.
Taking a multi-pronged approach
While some companies might have hoped to focus on internal adoption before rolling AI out to improve the consumer experience, the two things really need to happen in tandem today.
“You can’t just put all your eggs into making sure the consumer experience is all that. You need to make sure that actually everything else [internally] is going to surface that [improved experience],” Wayfair chief technology officer Fiona Tan, said. “It’s like you have a multiprong disruption.”
Tan tackles this layered disruption at Wayfair with a focus on three pillars: improving customer and supplier experiences, powering employees and powering the platform.
“We’ve been working really hard…to be able to have much better product data because that’s going to feed the customer experience. So that’s one very practical internal workflow operation that we have spent a lot of time building with predictive and now AI,” Tan said. On the employee side, it has been about empowering employees to use new tools, and making sure they are trained – and even having a little fun (at Wayfair this comes in the form of hackathons) doing it. “It’s not just a technology [function]. A lot of the impact is going to come from the commercial operations and supply chain teams and making sure that they also feel empowered with the tool set.”
When it comes to the platform, trying to hold your own corner and so-called proprietary customer data instead of playing nice with external AI programs won’t work. Consumers are going to end up shopping via AI, and brands would do best to make sure they’re supporting the experience rather than hindering it.
“We want to make sure that if the customer starts on Google Gemini, that they have the ability to go through the entire discovery to checkout experience,” Tan said. “That’s the power of the platform.”
Once Wayfair was able to dial back its hyper focus on getting developers on board with AI, and focus more on getting cross-functional teams to play together on improving processes, that’s when the company really started to feel an impact, Tan explained.
“What we’ve found to be helpful is…to have some domain experts, say merchants that know their workflows really, really well but they may not be as conversant in the tool set…then [bring them together with] the technologists that know the toolset but they don’t know the merchant workflow,” she said. “More so than ever before, the coming together of the different functions to go after these disruptive processes has been really key.”
What she’s looking at next is how to weigh the impact of these AI-driven technologies in areas of the company where it’s not as cut and dry as a coder spending one day on a project that would have previously taken 10.
“I think that’s maybe the next unlock: how do we start mapping those things that are happening throughout the organization and how do we get some efficiencies and disruptions around them?” Tan said.
Redesigning workflows for agents
This year will no doubt be a game changer where AI is concerned, and certainly the pace and adequate adoption of the technology will draw a dividing line between those brands that succeed and those that may fade from relevance. The time we’re upon is not unlike the one that saw some embrace e-commerce early and others who resisted — and that did not bode well for the latter.
But it can’t be about using AI just for the sake of it, Tan said.
“You don’t want to just agentify and automate an existing process just because. These agentic tools can understand so much more data, they can actually reason and do things you couldn’t quite expect a human process to do,” she said. It’s really about: “How do you come up with a new process that’s really optimized for agents and how do you know when not to do that? That’s been a learning for all of us.”
Catalyst’s biggest focus for 2026, according to Glebow, is making all of these advancements move faster for the organization.
“For us it’s [now about] focusing on speed and making sure that we can use the capabilities that we have to give us the insights we need to make better decisions,” he said. “And there are a lot of tools out there, there are a lot of techniques, and I think we are now at a point where we can begin to reasonably ask for an acceleration, and that’s what we’re going to do.”
