The Gap Between Activation and Transaction
While millions of people are beginning to interact with agentic AI tools, the transition from 'chatting' to 'buying' remains a significant bottleneck. Meta reports that over 10 million people in the United States and Canada have activated its Muse agentic AI tool. However, despite this high adoption rate, the actual volume of transactions processed by these agents across the broader economy remains remarkably low.
For many major platforms, the activity is currently negligible. Companies like DoorDash, Hilton, and eBay have reported very small to insignificant volumes of orders or bookings originating from AI agents. Even in sectors like insurance and storage, reports of agent-led cancellations are virtually non-existent. This discrepancy suggests that while the technology is being turned on, it is not yet the primary driver of consumer behavior.
Why AI Agents Aren't Scaling Yet
If the technology exists, why isn't it dominating the market? The data suggests a divide between specific retail integrations and general commerce. Meta has successfully integrated Muse into workflows for 10 major retailers, including Walmart and Best Buy, where over 25 percent of users are already utilizing the tool for shopping. In these controlled environments, users are spending an average of more than $200 per purchase.
The lack of scale elsewhere points to a lack of universal integration. For an AI agent to move from a niche tool to a primary commerce driver, it requires deep, standardized access to inventory, real-time pricing, and fulfillment APIs across all industries. Currently, many businesses have not yet established the infrastructure to allow an agent to act on a user's behalf autonomously. Without this widespread 'plug-and-play' capability, agents remain confined to the specific partners who have opted into the ecosystem.
Furthermore, there is a significant difference between a user asking an AI to find a product and an AI executing a multi-step purchase. The latter requires the agent to navigate diverse web architectures, varying checkout flows, and real-time availability updates. Until these interactions are standardized, agents will continue to struggle with the friction of non-integrated platforms.
The Shift Toward a Brokerage Model
The goal for platforms like Meta is to move beyond being mere hosts for advertisements. By implementing transaction fees for businesses for orders facilitated by Muse, the platform aims to create a revenue stream that could eventually exceed traditional advertising margins. This represents a fundamental shift in the business model: moving from a 'search-based' economy to a 'brokerage' economy.
In this new model, the AI agent acts as the gatekeeper. Instead of a user browsing dozens of ads to find a product, the agent identifies the best option and facilitates the purchase. For the reader, this could mean a more streamlined shopping experience, but it also introduces a new layer of mediation where the platform's fee structure determines the visibility and accessibility of products.
However, this transition is not without debate. Executives are currently weighing how these agents might encourage comparison shopping in ways that could impact corporate profit margins. While agents are designed to find the 'best' fit, the mechanics of how they prioritize those results—and how those results affect the bottom line of the retailers involved—remains a primary point of uncertainty.
What Remains to be Seen
The current landscape is one of early adoption and fragmented success. While we know that Muse users are spending significant amounts at specific partner retailers, the total volume of spending across all partners is not yet fully transparent. We do not yet know the long-term impact these agents will have on brand loyalty or whether they will lead to a permanent shift in how consumers perceive value.
For the average user, the immediate change is the availability of more 'do-it-for-me' tools in specific stores. For businesses, the challenge is deciding whether to integrate with these agentic workflows to capture new revenue or to guard against a model that might prioritize price and utility over brand identity. The path from a few million users to a mainstream commerce standard depends entirely on how quickly these technical and structural hurdles can be smoothed over.