The Next Disruption in AI: Machines Becoming Customers, Not Just Employees
For the past two years, much of the discussion surrounding artificial intelligence has centered on productivity enhancements. From automating repetitive tasks and generating code to answering customer queries and summarizing documents, AI has predominantly been viewed as an employee substitute within businesses. However, a more profound shift is emerging—one that redefines AI’s role from an internal workforce tool to an external decision-maker in commerce.
AI systems are increasingly capable of searching, comparing, selecting, and ultimately purchasing products and services on behalf of humans. This development transforms AI from an employee role to that of a customer, fundamentally altering the flow of commerce.
The difference is crucial: while employees operate inside companies’ existing frameworks, customers dictate where money is spent. Should AI agents begin to make buying decisions, the entities controlling these agents could influence which retailers receive orders, which hotels get bookings, and which products or services gain or lose visibility in the market.
Early indicators of this transformation are already apparent. Amazon reported that its AI shopping assistant, Rufus, was used by over 300 million customers in 2025, generating nearly $12 billion in incremental annualized sales. OpenAI introduced Instant Checkout in September 2025, initially linking ChatGPT users to Etsy sellers when ChatGPT had over 700 million weekly active users. Shopify also noted that AI-driven traffic to its stores increased eightfold year-over-year in Q1 2026, with orders from AI-powered searches rising nearly 13 times over the same period.
This shift signals that the AI story is no longer just about productivity—it is becoming a story of distribution and influence over the entire commerce ecosystem.
The Customer Is Moving From The Browser To The Agent
Traditional online shopping follows a human-controlled path: search, browse, compare, click, checkout, and pay. Search engines like Google help locate websites, Amazon enables product comparisons, marketplaces aggregate sellers, and payment processors complete transactions.
AI agents, however, threaten to compress this entire journey into a simple conversation. A consumer might soon say, “Find me a laptop under $1,000 with 16GB RAM, good battery life, and delivery by Friday,” and the AI agent will perform all the necessary research, comparison, and purchase steps without the consumer visiting multiple websites or marketplaces.
This evolution is economically significant because it changes the interface through which demand enters the market. OpenAI’s Agentic Commerce Protocol exemplifies this direction by enabling AI agents, consumers, and businesses to communicate seamlessly during transactions while merchants retain fulfillment and support control. OpenAI’s Instant Checkout, initially available for Etsy sellers, is expected to expand to over a million Shopify merchants.
For businesses, this means that simply having a user-friendly website is no longer sufficient. Products need to be machine-readable—accurate, structured, and continuously updated to be “understandable” to AI agents.
Amazon Shows What Happens When The Retailer Owns The Agent
Amazon is a pivotal case in this emerging landscape because it controls both the marketplace and the AI shopping assistant interface. In 2025, Amazon reported net sales of $716.9 billion, a 12% increase from the prior year, with AWS contributing $128.7 billion and operating income reaching $80 billion.
More notably, Amazon claimed that Rufus contributed nearly $12 billion in incremental annualized sales after usage by over 300 million customers. Rufus not only assists with product research and comparison but can also purchase items from other online stores on behalf of customers. In May 2026, Amazon rebranded Rufus as Alexa for Shopping, integrating its shopping capabilities with Alexa’s personalized context.
This strategy highlights Amazon’s ambition not just to facilitate navigation within its platform but to become the starting point of the purchasing decision, controlling not only inventory and fulfillment but also what consumers decide to buy.
However, tensions are emerging. Amazon’s recent decision to block Meta’s Muse AI shopping agent from its platform demonstrates that retailers may resist relinquishing control over customer relationships and transaction data to third-party AI agents. The future battle may revolve around which AI agents gain permission to shop on which platforms.
Shopify Is Betting That AI Agents Become A New Distribution Channel
While Amazon aims to own the entire customer interface, Shopify pursues a different approach: enabling millions of merchants to be accessible wherever customers interact with AI agents.
Shopify reported a 34% year-over-year revenue growth in Q2 2026 with an 18% free cash flow margin. The company positions itself as the infrastructure backbone behind AI-driven commerce. In March 2026, Shopify announced that millions of merchants could sell via AI channels such as ChatGPT, Microsoft Copilot, Google Search’s AI Mode, and the Gemini app. Additionally, Shopify introduced Agentic Storefronts and helped develop the Universal Commerce Protocol with Google.
Shopify’s commerce data from Q1 2026 reveals that AI-driven traffic to its stores increased eightfold year-over-year, and orders from AI-powered searches jumped nearly 13 times. New buyers arriving via AI channels also ordered at nearly double the rate of other channels. Moreover, AI-referred visitors landing directly on product pages converted at rates nearly 50% higher than organic search visitors, and AI-powered search orders had 14% higher average order values.
Though still early, these trends suggest that AI agents could send fewer casual browsers but more qualified buyers, potentially transforming the economics of customer acquisition.
Advertising Has A Problem: AI Does Not Shop Like A Human
For decades, businesses have invested heavily in influencing human attention through advertising. Consumers would see ads, remember brands, research, read reviews, and then purchase.
AI agents disrupt this process by removing many steps. An AI does not value celebrity endorsements but evaluates specifications, prices, reviews, delivery times, return policies, and availability instantly and objectively.
Branding remains relevant insofar as trust, reputation, and quality influence the data AI agents use. However, the method of persuasion shifts toward making product information machine-readable, accurate, and up to date.
Retailers with inaccurate inventory data risk losing favor to competitors. Suppliers without structured data may become invisible to AI agents despite the quality of their products.
In this new world, search engine optimization (SEO) evolves into what might be called “agent optimization.” Companies will need to ask not just, “How do we rank on Google?” but also, “How does an AI agent decide our product is the best answer?”
Payments Are Building The Infrastructure For Machine Buyers
For AI agents to become effective economic customers, they must be able to transact autonomously. Payment networks are moving swiftly to enable this capability.
Visa launched Visa Intelligent Commerce, facilitating AI agents to conduct transactions with proper credentials, controls, and authentication. Mastercard introduced Agent Pay in 2025, based on agentic tokens and controls to enable AI systems to make purchases securely. In June 2026, Mastercard expanded this toward machine-to-machine payments, envisioning a future of rapid, small-value transactions between AI agents.
This concept transcends consumer shopping, envisioning software agents purchasing cloud computing resources, API calls, advertising inventory, or data services from other agents automatically.
Thus, the “customer” may no longer be a human, but software buying from software.
While this automation promises efficiency and scale, it also raises risks. Financial institutions warn of potential fraud, privacy breaches, security weaknesses, and unclear liability if autonomous systems mishandle transactions. According to Reuters, major banks are concerned about the safe management of financial information and consumer protections in AI shopping contexts.
Consequently, the technology requires robust identity verification, authorization, spending limits, audit trails, and clear liability frameworks—not just faster checkouts.
The New Middleman Could Be The AI Interface
One of the most critical questions in the AI-driven economy is: Who owns the customer relationship when AI makes the purchase?
Consider a hotel booking scenario. Today, travelers search Google, visit platforms like Booking.com or Expedia, compare options, and reserve rooms. Tomorrow, an AI agent could handle the entire process based on simple commands such as, “Book me a four-star hotel in Midtown New York for under $300 with free cancellation.”
The hotel provides the room, payment networks process the transaction, but the AI interface controls the decision.
This dynamic can redistribute economic power. Hotels may lose direct customer access, marketplaces might see reduced traffic, and AI platforms could gain significant influence over which hotels are selected.
Market reactions already reflect this potential shift. Following Meta’s launch of Muse AI assistant, travel stocks such as Expedia, Booking Holdings, TripAdvisor, and Airbnb faced pressure. Bloomberg Intelligence analysts, cited by Barron’s, estimated that a 5%-10% business shift toward AI agents could expose over $5 billion in revenue risk across travel, ride-sharing, and delivery sectors. While speculative, this underscores where bargaining power might move.
The Biggest Businesses May Become The Ones Behind The Agent
The opportunity extends beyond retail. AI agents require a complex infrastructure spanning multiple domains:
- AI models to reason and make decisions
- Cloud infrastructure to run these models
- Data platforms to provide relevant context
- Commerce systems to expose product inventories
- Payment networks to settle transactions
- Fraud detection systems to authenticate purchases
- APIs to connect AI agents with businesses
- Logistics networks to fulfill orders
Salesforce exemplifies this enterprise-focused approach. The company reported $41.5 billion in revenue for fiscal 2026, with its Agentforce platform reaching approximately $1.2 billion in annual recurring revenue by Q1 fiscal 2027. Unlike Amazon’s consumer-centric AI overlay, Salesforce embeds agents inside enterprise workflows.
This distinction highlights the dual nature of the emerging AI economy: consumer agents purchasing goods and enterprise agents procuring software, services, data, and computing resources.
What Happens To The Economics Of Customer Acquisition?
This evolution is especially pertinent to investors. Today, companies spend heavily on advertising, search marketing, affiliates, marketplaces, and sales teams to acquire customers.
As AI agents assume more purchasing decisions, some traditional customer acquisition costs could diminish. AI agents can directly identify suitable products, reducing the need for expensive clicks and broad advertising.
However, a new dynamic may emerge: AI platforms themselves could become the gatekeepers. Rather than paying Google for search prominence, businesses might need to compete to become the product selected by an AI agent.
This could foster a new form of platform dependence. Today’s digital economy already shows the consequences of concentrated distribution power, and AI agents—who don’t just list options but potentially make the choice—could amplify this trend.
For investors, this means revenue growth alone does not tell the full story. Critical questions become: Who owns demand? Who controls customer data? Who manages recommendations? Who processes transactions? And who earns the economic take rate?
The First Battle May Be Between AI Agents Themselves
Another disruptive possibility is that both buyers and sellers become software agents. The traditional internet model assumes humans buy and businesses sell, but agentic commerce could replace this with machine-to-machine transactions.
Procurement agents could request quotes, supplier agents respond with pricing, negotiation agents compare contracts, purchasing agents select vendors, payment agents settle invoices, and logistics agents arrange delivery—all autonomously.
Humans might become increasingly distant from individual transactions. Mastercard’s 2026 Agent Pay announcement explicitly envisions a future where businesses offer services for AI agents to purchase and use, enabling continuous, high-speed transactions.
This would establish a new economic category where machine-to-machine commerce becomes the fundamental unit, with potentially vast scale—though still difficult to quantify today.
The Risks Are As Large As The Opportunity
While autonomous shopping could make commerce more efficient, substantial challenges remain:
- Trust: Why should consumers rely on AI agents for high-value purchases?
- Liability: If an agent buys the wrong product, who is responsible—the consumer, AI provider, retailer, or payment processor?
- Manipulation: Could companies pay to bias AI agents’ decisions?
- Data Privacy: AI agents hold vast commercial information including budgets, locations, preferences, purchase histories, and financial credentials.
- Platform Concentration: A small number of AI assistant providers could dominate as intermediaries between millions of businesses and customers.
These concerns are not theoretical. Financial institutions already call for greater transparency in AI-initiated transactions, stronger data protections, and clearer accountability when errors occur.
The Investor Question Is Not “Who Will Replace Workers?”
While replacing human workers dominated the first AI boom phase, the next phase centers on control over economic decisions made by machines.
Companies positioned at the decision-making layer—those who own customer relationships, data, and transaction processing—may enjoy fundamentally different economics than firms focused purely on AI software.
Amazon boasts enormous transaction volumes supported by sophisticated shopping AI. Shopify is making millions of merchants accessible through AI interfaces. OpenAI is developing protocols to facilitate AI-merchant transactions. Visa and Mastercard are building the payment infrastructure. Salesforce is embedding agents within enterprise workflows. Meanwhile, companies across cloud computing, cybersecurity, data infrastructure, and logistics prepare for a commerce world increasingly initiated by automated software.
While agentic commerce may not fully replace traditional shopping—consumers might still prefer human browsing, physical stores, or direct brand relationships, and regulators may impose restrictions—the investment trend is clear: the infrastructure enabling machines to participate directly in commerce is rapidly being built.
Conclusion
The internet was designed around human interaction: searching, clicking, reading, comparing, entering payment details, and pressing “Buy.” AI agents have the potential to compress much of that process into a few simple instructions.
This shift fundamentally alters internet economics. Future businesses will not only compete for human attention but also to become the answer selected by AI agents.
As a result, product data, APIs, payment systems, trust mechanisms, distribution channels, and agent access become strategically critical, reshaping the hierarchy of economic power.
The product seller still earns revenue, logistics companies fulfill orders, and payment networks process transactions. But increasingly, the entity controlling the AI that decides what gets bought sits between all of these players and the customer.
The AI revolution may not end with machines becoming employees—it may culminate with machines becoming customers, presenting businesses the unprecedented challenge of winning over customers that can evaluate every option in milliseconds.
Source: Here
