The rapid rise of autonomous artificial intelligence agents is forcing major global retailers and financial institutions to redefine how they interact with non-human digital traffic. As consumers increasingly deploy software programs to automate daily digital tasks—such as grocery shopping, flight bookings, and account management—businesses find themselves at a strategic crossroads. An AI agent fundamentally alters the traditional digital consumer journey: it operates on a user’s behalf, yet exhibits browsing behaviors, processing speeds, and operational scopes that a human user would rarely replicate.
This operational shift has triggered an urgent policy debate across the digital economy. With automated agents routinely parsing entire website architectures, executing background transactions, and interacting directly with enterprise databases, corporations must rapidly establish frameworks to govern this new paradigm. In October, three major global enterprises provided starkly different answers to the challenge of managing AI agents, illustrating a fractured early landscape for agentic commerce.
The Diverging Strategies: Blocks, Liability Shifts, and Dedicated Profiles
The responses from major corporations to the influx of AI agents span a wide spectrum of approaches, ranging from outright exclusion to sophisticated infrastructure adaptation.
At one end of the spectrum, financial institutions like Chase have opted for a defensive posture, blocking AI agents entirely from accessing their systems. By imposing strict prohibitions on automated navigation and transaction execution, these institutions aim to eliminate security vulnerabilities, prevent potential fraud, and mitigate the unpredictable load that autonomous scripts can place on proprietary banking portals. While secure, this approach effectively locks out consumers who wish to leverage automation to manage their financial portfolios.
In contrast, other financial organizations, such as Bank of America, have adopted a permissive yet risk-shifting model. Under this framework, consumers are permitted to utilize third-party AI agents to interact with their accounts, but the financial liability and operational risk remain squarely on the customer. If an agent executes an erroneous transaction, misinterprets data, or triggers security protocols, the onus is placed back onto the human account holder to remediate the issue. This strategy accommodates the growing demand for automation without requiring the institution to build bespoke infrastructure to differentiate between human and synthetic inputs.
Taking a more forward-thinking approach, retail giant Walmart has charted a third path, preparing to introduce dedicated profiles for AI agents within individual customer accounts. This innovation recognizes a fundamental truth of agentic commerce: an AI agent does things its human customer would not, and yet it acts strictly on that customer’s behalf.
Walmart’s forthcoming Agent Profiles are designed to resolve this dichotomy. Historically, when a consumer deployed an AI agent, the software would sign in directly using the customer’s primary credentials. This blended human and synthetic activity into a single operational stream, making it impossible for retailers to distinguish between a human shopper browsing at a leisurely pace and an autonomous agent systematically scanning entire inventories within milliseconds. By establishing a separate profile linked directly to the primary account, Walmart enables a transparent ecosystem where the enterprise can simultaneously recognize the human customer and the specific agent authorized to act for them.
Understanding the Behavioral Gap Between Humans and AI Agents
The necessity for specialized frameworks like Walmart’s Agent Profiles stems from the unique operational characteristics of artificial intelligence. Unlike human shoppers, who typically view a limited number of product pages, read targeted reviews, and make purchasing decisions based on personal preference and emotional impulse, an AI agent operates through algorithmic logic and programmatic directives.
An agent might browse every single page of a retail catalog, download massive product data sets, or execute complex multi-variable price comparisons in seconds. Furthermore, an agent operates within predefined constraints—such as a strict spending limit or specific dietary restrictions—established by the user.
Using a real-world analogy, sending an AI agent to shop on a retail website is akin to sending a designated representative to a local market with a specific amount of cash and strict instructions. The representative acts exclusively on your behalf, but the transaction only maintains integrity if the merchant knows precisely who sent the representative and what boundaries govern their authority. Without a distinct profile or authorization token, merchants are left blind, unable to discern whether incoming traffic represents a loyal human customer, an authorized software assistant, or a malicious scraper.
The Scale of Agentic Commerce: Early Adoption Metrics
Despite significant media attention surrounding the concept of agentic commerce, empirical data indicates that autonomous transactions remain in their absolute infancy.
During a recent third-quarter earnings call, DoorDash Chief Executive Officer Tony Xu revealed that external AI agents had placed "maybe several hundred orders in total" across the platform. To contextualize this figure, DoorDash processed a staggering 993 million orders during the same three-month period, translating to an average of more than 10 million orders daily.
The juxtaposition of hundreds of agent-driven orders against nearly a billion human-initiated transactions underscores that agentic commerce is not yet a dominant commercial channel. However, the minuscule current volume provides businesses with a critical window of opportunity—time to design, test, and implement robust governance protocols before automated traffic scales exponentially.
Emerging Standards and Technical Protocols
As businesses grapple with how to identify and authorize incoming AI traffic, the technical community is actively working to establish standardized communication protocols.
The Personal Agent Protocol, spearheaded by AI customer-service firm Sierra and tech conglomerate Meta, is currently being drafted to solve the identity crisis of agentic commerce. More than 40 companies are collaborating to shape the initiative, which aims to create a universal framework allowing AI agents to properly introduce themselves to enterprise websites.
In late October, the Personal Agent Protocol draft incorporated a critical security recommendation: agents should cryptographically sign their requests using Web Bot Auth. This proposed standard enables an autonomous agent to conclusively prove which parent company or organization operates it. According to insights from bot-protection firm DataDome, Meta’s proprietary "Muse" agent has quietly begun signing a portion of its web requests using this authorization standard since mid-October, signaling a quiet industry shift toward verified machine-to-machine communication.
Strategic Implications for Enterprises
The fragmented responses from Chase, Bank of America, and Walmart highlight a broader strategic imperative for digital businesses across all sectors. As agentic commerce matures, organizations can no longer afford to treat all incoming web traffic through a binary lens of "human" versus "bot."
Enterprises must prepare for the mainstream arrival of consumer AI agents by focusing on three foundational actions:
- Infrastructure Audit: Review existing web application firewalls and access management systems to determine whether they can successfully identify and categorize authenticated AI agents distinct from malicious web scrapers.
- Account Architecture: Evaluate whether current user account structures can support separate agent profiles or delegated authorization models, ensuring transparency without compromising consumer security.
- Policy Alignment: Establish clear terms of service regarding automated interactions, defining explicit boundaries for liability, spending limits, and data access before regulatory mandates force compliance.
Ultimately, businesses that successfully bridge the gap between human consumers and their authorized digital assistants will capture early market share in the next evolution of digital commerce. Those that default to blanket blocks or push all operational risk onto the consumer risk alienating an increasingly automated customer base.
