I’m more interested in what people bring to an AI shopping assistant than what the assistant puts in their cart. A preference, a trusted brand, or a product they have been meaning to try can give that conversation a very different starting point. My instinct is that making shopping easier could make the work of earning that preference more valuable, not less.

Meta’s Muse makes the opportunity tangible: a personal AI agent that can navigate websites, research options, and complete purchases with the user’s approval (Meta’s launch announcement). What I find encouraging about the broader agentic commerce landscape is the possibility of helping people act on what they want with less effort. For brands and agencies, the opportunity is to connect the work that creates that interest with the experience that fulfills it.

That is where the conversation becomes particularly relevant to MCG. My view is that a more intelligent shopping experience deserves an equally thoughtful media operation: one that connects the brand’s ambition to the right environments, makes good use of the budget, and adapts as we learn. We do not need to predict which assistant will win to start making those connections stronger.

Muse is part of a broader shift

Muse is one expression of a larger development across commerce. Google has introduced the Universal Commerce Protocol with partners including Shopify, while OpenAI is using shared merchant product data to support discovery and recommendations in ChatGPT (Google’s commerce announcement, OpenAI’s merchant overview).

These experiences are not interchangeable: Meta describes Muse acting on a user’s behalf, while OpenAI’s current merchant guidance emphasizes discovery in ChatGPT followed by checkout on merchant-owned websites or apps (Meta’s launch announcement, OpenAI’s merchant overview). I would be cautious about treating “agentic commerce” as a single channel with one buying journey or operating model.

There is also an important distinction between commerce assistance and advertising access. Meta says Muse conversations and private virtual-machine data are not shared with its advertising systems (Meta’s launch announcement). For now, I see the most useful planning question as how these experiences could change customer behavior, rather than what new targeting signals we might gain.

Brand preference still has a place in the conversation

Imagine asking an assistant, “Find me a good pair of running shoes under $150.” Now imagine asking, “Find me this brand’s running shoes in my size.” The task is similar, but the starting point is quite different.

That is why I would not separate the agentic commerce conversation from the brand-building conversation. If consumers delegate more of the shopping process, the preferences they bring into that process may become especially valuable. Advertising can help establish those preferences, even when the eventual purchase takes place somewhere else.

For brands, I see two complementary priorities: give people a reason to want you, and make it easy for a shopping assistant to understand what you offer. Clear product details, current availability, and reliable fulfillment information belong alongside a distinctive proposition and relevant creative.

For agencies, this creates an opportunity to connect the media plan more closely to the commerce experience. Premium video, CTV, and other programmatic environments should be evaluated for the consideration and incremental demand they can create, not simply their proximity to a final click. That is a hypothesis worth testing, rather than a reason to move budgets automatically.

A different journey may need a different view of performance

If an assistant handles some of the comparison and navigation, a purchase may involve fewer of the familiar human browsing steps. That would not necessarily mean advertising had less influence; it might mean that influence is less visible in the usual sequence of clicks.

I would want brands and agencies to prepare for that possibility together. Where data access and consent allow, verified purchases, new-customer acquisition, and incremental revenue should help explain performance alongside platform reporting. Teams should also distinguish automated browsing from human engagement without discounting legitimate agent-assisted purchases.

There is a practical programmatic requirement underneath that measurement work. We need to understand what inventory ran, through which paths, at what cost, and with what delivery quality. A clearer view of the media inputs gives measurement teams a better foundation for evaluating outcomes.

Where I see MCG contributing

At MCG, we focus on the supply feeding the buyer’s existing programmatic platforms, using visibility across major SSPs to inform supply quality, pricing, and path selection (MCG’s supply-layer perspective). I see that capability becoming particularly useful as brands and agencies test how their media should support new shopping behaviors.

The connection is practical. Consumer agents may change how demand becomes a purchase; our role is to help improve the media execution supporting that demand.

• Better alignment between the brief and the inventory: If the goal is to build consideration before a consumer begins shopping, the supply should reflect that objective. Our work organizing premium CTV and OTT inventory around buyer goals before the bid is one way we support that alignment (MCG and Marketing Architects partnership).

• More value from the available budget: As teams explore new commerce experiences, I would also look for opportunities to reduce redundant supply paths and unnecessary costs in existing activity. Our cross-SSP perspective helps identify inefficient routes and opportunities for better media value (MCG’s supply-layer perspective). The aim is to create room for more effective investment, while preserving quality.

• An operation that can adapt with the learning: A test may suggest a different environment, supply mix, or investment priority. Our active deal management and cross-platform activation capabilities are designed to support adjustments within the buyer’s existing technology setup (MCG’s operational capabilities, MCG’s activation model). That gives agencies a practical way to act on what they learn.

I see this as a complementary role. Commerce partners help make products accessible, agencies guide strategy and investment, and measurement partners evaluate business impact. MCG helps strengthen the supply and execution connecting the media brief to delivery; that is distinct from influencing an agent’s recommendations or accessing its private user data.

A useful place to begin

My instinct would be to start with a focused working session, not a wholesale redesign. Bring the commerce, media, analytics, and operations teams together around one customer journey and one business objective.

Ask where brand preference needs to be established, whether product information supports assisted discovery, and how a resulting purchase would be measured. Then review whether the current programmatic supply is well suited to that job, and choose one improvement to test.

For an agency, that might mean comparing an existing supply approach with a shaped-supply approach while holding other important variables as consistent as possible. The evaluation should consider delivery quality and all-in costs alongside business outcomes, with the appropriate measurement partners involved.

Muse is a useful prompt for this discussion, but I would not make the plan dependent on any one assistant becoming the dominant shopping interface. The more durable opportunity is to build a media operation that can learn and adapt as customer behavior changes.

That is where I believe MCG can be helpful: bringing greater visibility, better supply choices, and responsive execution to the investments brands and agencies are already making. If agentic commerce is on your planning agenda, we would welcome a conversation about what it means for your media operation and where a practical first step could create value.