Why Muse, Dots, and the rise of the agentic consumer make visibility a holiday priority
Last winter, HUMAN asked whether consumers and brands were ready for agentic commerce, but the answer was mixed: consumers were moving quickly toward AI-assisted shopping, while many marketers were still unsure how agentic traffic would appear in their data.
With the recent debuts of Meta’s Muse and OpenAI’s Dots, that question has become a good deal more urgent.
These two new agents are the first to really reach the mainstream. They can research products, compare options, browse websites, and take actions on a person’s behalf, just as earlier agents could, but with the added benefit of Meta’s and OpenAI’s reach (and marketing budget). And as the holiday shopping season begins in earnest, more of the actions that can be offloaded to agents will take place inside the same digital journeys that marketing teams measure and security teams protect.
The best holiday customer may, in fact, be a bot. So the practical question for businesses is simple: can they tell which automated interactions represent valuable demand, ordinary research, or unacceptable risk?
The visibility question was already urgent before Muse
HUMAN’s new Iris Report was fielded well before Muse launched, which makes the findings a useful baseline for understanding the challenge that businesses now face, with a lens of the “before times.”
Both the CMO and CISO cohorts polled believed agents were already reaching their websites or applications. Ninety-four percent of marketing leaders and 98% of CISOs said agents were possibly or definitely present. Yet only 16% of marketing leaders and 4% of CISOs said they could quantify that activity.
That gap existed even before a broadly distributed consumer agent made agentic AI easier to access. HUMAN’s early observations after Muse launched showed how quickly the composition of agentic traffic can change. In the first month of data including Muse, the overall volume of agentic AI sessions tripled, and Muse made up 40%—by far the largest single share—of the total…despite not being available for the entire month.
Muse didn’t invent agentic AI. Its significance comes from its distribution, low-friction access, and consumer positioning. A new agent can become a meaningful traffic source in days. Organizations need the visibility to recognize that shift while it is happening.
The agentic consumer changes what a conversion means
Last year’s Iris Report found that 64% of consumers planned to use AI (not necessarily agentic) for holiday shopping in 2025, compared with only 11% who had used AI for holiday shopping the year before. Nearly three-quarters said they were at least willing to consider an AI-powered browser that could make purchases on their behalf.
Those findings described a consumer shift before today’s newest agents had reached the market. They also pointed to a change in the customer journey. An agent may research several retailers, compare prices, check inventory, or initiate a purchase without behaving like a conventional human visitor.
That’s a measurement problem for marketing. An agent acting on behalf of a real shopper can be high intent, but it may not fit neatly into the assumptions built into standard analytics. If the session is counted as a human visit, attribution and conversion reporting can become harder to interpret, especially since the session may not behave in a similar fashion as a human carrying out the exact same tasks. But if it’s treated as unwanted automation, a brand may block a potential customer before the transaction begins.
“Automated traffic isn’t always bad traffic. In many cases, it’s high-intent, high-value traffic that brands should want to support.”
Aaheli Guhathakurta, HUMAN Product Marketing Manager
“Good bot or bad bot?” is no longer enough
The new Iris Report shows how little confidence organizations have in their existing tools. Forty-six percent of marketing and security leaders say they are not very or not at all confident that their tools can distinguish malicious bots from beneficial AI agents.
That uncertainty creates a difficult choice. Blocking all automated activity may suppress useful discovery and shopping behavior. Allowing every agent—or anything that claims to be an agent—to move through a site without additional controls may expose accounts, data, and transactions to abuse.
The useful decision is more specific:
Answering those questions requires more than detection. It requires a shared model for identity, intent, behavior, and trust. That is the role of agentic trust.
Marketing and security need a shared view of the gaps
The survey shows that two executive functions report different consequences while neither group has strong confidence in its ability to classify the traffic producing them.
For marketing leaders, the reported consequences are concentrated in measurement. Fifty-eight percent saw broken attribution tracking in the previous 18 months. Fifty-two percent reported distorted KPIs, and the same proportion reported inaccurate CAC and LTV calculations.
For CISOs, the reported consequences are concentrated in threat activity. Forty-four percent saw growth in automated scraping against applications or APIs. Forty-two percent reported account takeover attacks, and 36% reported difficulty defining or operationalizing incident-response triggers for fraud and abuse.
These findings describe two views of an incomplete picture. The business question is therefore not which function owns agentic traffic in isolation. The more useful question is where marketing and security have visibility, where they do not, and who is responsible for closing each gap.
The conversation CMOs and CISOs need before holiday traffic peaks
A practical conversation can begin with five questions:
The answers may not produce a perfect picture immediately. They can establish a shared one. That is an important distinction as holiday traffic accelerates. Marketing and security do not need identical priorities, but they do need a common understanding of the actors moving through the customer journey.
The holiday season will put that understanding under pressure. Agents will influence what shoppers discover, compare, and buy. They may also create new opportunities for scraping, account abuse, and fraud. Teams that cannot distinguish those behaviors will have difficulty deciding what to optimize, what to protect, and what to govern.
Visibility comes before trust
The agentic consumer is already becoming part of the digital customer journey. Muse has made that shift visible at consumer scale, and Dots adds to the momentum behind a market that is moving quickly toward delegated browsing and action.
HUMAN’s Iris Report found that both marketing leaders and CISOs believe agents are present, while very few can quantify them. It also found that both groups are trying to make decisions with limited confidence in their traffic classifications and tools.
The first step is to make the conversation explicit. CMOs and CISOs should compare what they know, what they do not know, and who owns the gaps between those two states.
Read the full HUMAN Iris Report: CISO and CMO Perspectives on Agentic Traffic Visibility for the complete findings and role-by-role breakdowns, or explore an AgenticTrust demo if you’re ready to see agentic traffic on your own properties.