Are AI agents strengthening your brand or quietly replacing it?

Earlier in the year, I returned from NRF in New York – the world’s largest retail conference – where one theme cut across conversations and keynotes: AI agents are steadily moving into the core of how consumers search, compare, and buy.

The shift is already visible. Technology platforms are integrating transactional capabilities directly into conversational interfaces, and major retailers are making product data, pricing, and availability accessible to these agents. The ambition is to enable end-to-end purchase journeys without ever leaving the interface.

From a consumer perspective, the value is obvious. A single conversation can replace hours of browsing – you define your preferences, set a budget, and let the agent do all the hard work. In Deloitte’s most bullish case scenario modelling, agentic and GenAI-initiated commerce within beauty – a category that typically leads digital adoption – is expected to comprise up to 30% of US beauty sales by 2035. Within that, agentic commerce alone is projected to grow at more than 35% annually.

At the moment, most of the conversation is centred on how fast brands can move and how much they can gain. Very few are asking what it means when AI agents begin to own the interaction between your brand and your customers. For some, leaning into this shift may make sense. But for most, I would argue, it introduces a real risk of losing control of the experience and how your brand is ultimately perceived and chosen.

It also raises more fundamental questions. Does the shift to agentic commerce begin to challenge the role of your own channels? Why continue to invest in platforms, product teams, and experience layers if the transaction increasingly happens elsewhere? And what about implications for retail economics: Early signals point to lower conversion rates and smaller basket sizes in agent-driven interfaces – even if volume may increase. That dynamic needs to be understood and clearly articulated in terms of what it means for your P&L.
 

A playbook for agentic commerce

I believe brands need to take a more defensive stance towards agentic commerce on third party platforms. Not in the sense of opting out – you clearly need to play in this arena – but in being deliberate about where you engage and where you hold back.

Once a customer interaction moves into an AI agent, the mechanics of how consumer decisions are made begin to change. The journey becomes shorter, with fewer touchpoints and less exposure to the environments brands have spent years building. What used to be shaped through design, content, storytelling, and context risks being reduced to a set of signals the agent can work with.

In Deloitte, we recently published a playbook for how brands should approach agentic commerce. Its purpose is to bring structure to a space currently driven by momentum and hype. The first step is to understand how and where AI agents will interact with your business: what data is made available, how your products are surfaced, and which parts of the journey can be handled externally.

From there, the playbook focuses on defining boundaries. Not everything should be optimised for agent-driven flows. In practice, that means deciding what to expose and what to keep within your own ecosystem. The third element is about strengthening what you own. If more of the transactional layer moves elsewhere, the value of your own channels increases. That includes experience, content, loyalty, and the ability to engage customers on your own terms. The final element is integration – but on your own terms. You still need to show up in agentic environments, but this should be a conscious strategic choice, not a default reaction to market pressure.
 

Choose the battleground

What agentic commerce ultimately comes down to is where you choose to compete. If agentic commerce takes over parts of the transactional journey, the remaining battleground becomes even more important. And that is where many brands still have an advantage… if they choose to use it.

Customer experience is one of those areas. AI agents are highly effective at optimising for efficiency. They can search, compare, and execute faster than any human. But they cannot replicate the full experience of engaging with a brand across touchpoints. The inspiration, the context, and the sense of discovery in a physical setting – the smell, the lighting, the music. Think of walking into a flagship store, met by the smell of coffee, or into a building store where you notice the scent of sawdust. These are simple sensory elements that still sit firmly within the brand’s control.

The same applies to how you think about your assortment. Being more deliberate about how your assortment is exposed – what is available where, and under which conditions – becomes a way to maintain a degree of control and create a reason for customers to seek you out directly through exclusive SKUs reserved for your own channels. At the same time, you can use lower-priced “recruitment SKUs” in agent-driven channels to drive volume and visibility.

The brands that get this right will not be those that fully optimise for agentic commerce, but those that balance it with strong, owned physical, digital, and omnichannel experiences. This is why I call it a defensive approach to agentic commerce. Because if you get it wrong, you risk becoming just another supplier inside someone else’s interface. And where is the brand value in that?