The Search Bar Lasted 25 Years
The dominant interface for online commerce — type keywords into a search bar, browse results, click through to a product page, add to cart, check out — was invented in the late 1990s and has not fundamentally changed since. Its persistence through the smartphone era, the social commerce era, and the first wave of recommendation engines is a testament to how difficult it is to displace a habitual interface.
AI shopping agents are not an incremental improvement on search. They are an architectural replacement. Rather than responding to keywords, an AI agent understands intent: “I need running shoes for trail running, my budget is $150, I’ve had knee issues before, and I want to order by Thursday.” The agent queries product databases, applies constraint matching, reads review summaries, checks shipping timelines, compares prices across retailers, and returns a ranked recommendation — or, with user authorization, completes the purchase autonomously. The consumer’s interaction shifts from a series of clicks to a single conversation.
The behavioral data from 2026 confirms the transition is underway. According to commercetools research, 73% of consumers already use AI somewhere in their shopping journey: 45% for product ideas, 37% for review summaries, and 32% for price comparison. Only 13% have completed a purchase directly via an AI agent — but 70% report feeling comfortable with the concept of autonomous agent purchases. The intention gap between adoption and comfort is small. When the infrastructure catches up to the comfort level, the transition will accelerate.
What Google’s Universal Commerce Protocol Actually Does
The most significant structural development in agentic commerce in 2026 is not a consumer-facing product. It is the Universal Commerce Protocol (UCP), co-developed by Google with major retailers including Shopify, Etsy, Wayfair, Target, and Walmart, and endorsed by 20+ ecosystem partners including Adyen, American Express, Stripe, Visa, Mastercard, and Best Buy.
UCP is an open standard for agentic commerce — a protocol that allows AI agents to interact with retail databases, checkout systems, and payment providers in a standardized way. It is compatible with Agent2Agent (A2A), the Agent Payments Protocol (AP2), and Model Context Protocol (MCP), the emerging vocabulary of machine-to-machine commerce.
Without a protocol, an AI agent that wants to shop across multiple retailers would need a bespoke integration with each one — the same fragmentation problem that plagued early e-commerce payment integration. With UCP, a retailer that adopts the standard becomes accessible to every AI shopping agent built on the protocol, regardless of which AI platform the consumer uses.
The commercial logic is identical to early-internet logic: the merchants that adopt the protocol early get indexed by AI agents early, which means their products get surfaced to AI-assisted shoppers early. The merchants that delay are invisible to the agent layer until they adopt. The first-mover dynamic is real.
Google has also launched three concrete consumer-facing products built on this infrastructure. Business Agent enables shoppers to chat with brands directly on Google Search — live with Lowe’s, Michael’s, Poshmark, and Reebok. Direct Offers is a Google Ads pilot that presents exclusive discounts to purchase-ready shoppers in AI Mode — early partners include Petco, e.l.f. Cosmetics, Samsonite, and Rugs USA. Agentic Checkout enables direct checkout from eligible U.S. retailers on Google’s AI surfaces using Google Pay or PayPal.
These are not experimental features. They are the commercial deployment of the agentic commerce architecture.
Advertisement
The Platform Landscape: OpenAI, Perplexity, and the AI Commerce Stack
Google is not alone. The AI commerce layer is being built simultaneously across platforms with different strengths.
OpenAI introduced shopping research in ChatGPT, using GPT-5 mini with reinforcement learning to identify products, summarize reviews, and compare options. Its integration with major retailers and payment providers is ongoing, and the conversational shopping mode is already available to ChatGPT users in the United States.
Perplexity launched free shopping with conversational discovery and PayPal-powered checkout — positioning itself as a search-to-purchase platform that bypasses the traditional retailer website entirely. A consumer asking Perplexity for a product recommendation can complete the purchase without ever visiting the brand’s own site. This disintermediation model is the most aggressive form of the agentic commerce shift.
Anthropic’s Claude has been integrated into merchant-facing tools and customer service systems, though its primary commerce role in 2026 is B2B rather than direct consumer purchase.
The Gartner projection provides the B2B dimension: 90% of B2B buying will be AI agent-intermediated by 2028, driving over $15 trillion through AI agent exchanges. Forrester’s 2026 predictions add operational specificity: one-third of B2B payment workflows will use AI agents, 20% of B2B sellers will engage in agent-led quote negotiations, and 1 in 5 sellers will respond to AI-powered buyer agents with dynamic counteroffers.
What Retailers and Brand Leaders Should Do Now
The agentic commerce transition creates an urgent optimization challenge for any brand or retailer with digital commerce revenue. The playbook is not the same as SEO or social commerce optimization — it requires a different set of technical and commercial actions.
1. Audit Your Product Data for Agent Readability, Not Just Human Readability
AI shopping agents retrieve product information from structured data sources — APIs, product feeds, schema markup — rather than by scraping a visually rendered product page. A product page designed for human browsing (hero image, lifestyle copy, tabbed specifications) may be largely opaque to an AI agent querying for structured attributes. The immediate action is a product data audit: does every SKU have clean, machine-readable attributes for material, dimensions, compatibility, shipping time, return policy, and price? Does your product feed include question-answer pairs that an agent could use for conversational recommendation? Google’s Merchant Center updates for 2026 specifically added new data attributes for conversational commerce discovery — the first retailers to populate those attributes will be surfaced by Google’s Business Agent for relevant queries.
2. Evaluate Universal Commerce Protocol Adoption Before Competitors Complete It
UCP adoption is currently voluntary and in early rollout. The window between a protocol’s launch and its widespread adoption is historically the highest-value period for early adopters: they get indexed by AI agents before the protocol becomes table stakes, capturing the traffic and conversion advantage that latecomers cannot recover. Shopify merchants have UCP integration available through their platform; retailers on other stacks should check with their commerce platform vendors about UCP support timelines. The adoption decision is not primarily technical — it is commercial. The question to answer is not “can we implement UCP?” but “what is the revenue cost of not being surfaced to AI agents in the next 12 months?”
3. Redesign the Conversion Funnel for Agent-Mediated Discovery
The traditional e-commerce conversion funnel assumes a human user who browses, discovers, considers, and eventually converts — with retargeting and email sequences designed to recover abandonment at each stage. An AI agent that reaches a purchase decision before the consumer visits a product page eliminates most of that funnel. The implications are significant for advertising strategy, for the value of a branded website, and for customer relationship management. Brands that depend on product page content to do the selling — through photography, storytelling, user-generated content — need to think carefully about how that content translates into agent-readable signals. The conversion funnel redesign for agentic commerce is the strategic planning task that most retail organizations have not yet started.
What Comes Next: The Trust and Attribution Problem
The agentic commerce transition is not frictionless. Two structural challenges will define the medium-term trajectory.
The first is trust. The 70% consumer comfort rate with autonomous agent purchases conceals significant variance: consumers comfortable with an agent buying detergent may draw the line at electronics, travel, or anything requiring subjective preference judgment. The agent commerce platforms are investing heavily in purchase confirmation flows, preference profiles, and return policy integration to raise that comfort threshold. But trust is built slowly and broken quickly — a misaligned agent purchase, especially an irreversible one, will create the kind of negative anecdote that sets adoption back.
The second is attribution. If Perplexity’s agent recommends and purchases a product without the consumer visiting the brand’s website, who gets the attribution? What is the brand’s advertising relationship with Perplexity vs. Google vs. OpenAI? The emerging answer — sponsored placement within AI responses, analogous to paid search but native to conversational interfaces — is already visible in Google’s Direct Offers pilot. The cost structure, targeting capabilities, and measurement frameworks for AI-native advertising are being negotiated right now, and the norms that emerge from 2026’s early deployments will set the commercial terms for a $385 billion market.
Retailers and brands that engage with these questions now — before the transition matures — will be in the room where those terms are set.
Frequently Asked Questions
What is the Universal Commerce Protocol (UCP) and why does it matter?
UCP is an open standard co-developed by Google with Shopify, Walmart, Target, and 20+ other partners including Stripe, Visa, and Mastercard. It defines how AI shopping agents interact with retail databases, checkout systems, and payment providers in a standardized way. Without UCP, an AI agent needs a bespoke integration per retailer; with UCP, any UCP-compliant retailer is accessible to any UCP-compatible agent. Retailers that adopt UCP early get indexed by AI agents before competitors, capturing the same first-mover advantage that early Google indexing provided in the early 2000s.
How accurate are Morgan Stanley’s $385 billion agentic commerce projections?
Morgan Stanley’s AlphaWise survey found that 23% of Americans made at least one AI-assisted purchase in the past month, providing a current adoption baseline. The $385 billion bull-case projection for 2030 represents 10-20% of the total U.S. e-commerce market and assumes that the 70% consumer comfort rate with autonomous purchases converts to behavioral adoption as trust-building infrastructure matures. The base case is $190 billion. Both figures are U.S.-only; global projections from multiple research firms put the total addressable commerce volume through agentic systems at $3-5 trillion by decade’s end. These are projection ranges, not certainties — actual figures depend on trust resolution, platform adoption rates, and regulatory environment.
How should retailers think about the threat of disintermediation from platforms like Perplexity?
Perplexity’s conversational discovery-and-checkout model — where a consumer completes a purchase without visiting the brand’s website — represents the most disintermediation-intensive form of agentic commerce. The risk for brands is losing the direct consumer relationship, the retargeting data, and the brand storytelling that a product page provides. The mitigation strategies are: (1) ensuring your product data is high quality in every structured feed that AI agents consume, (2) building direct consumer relationships through post-purchase communication that survives agent-mediated acquisition, and (3) evaluating sponsored placement within AI shopping responses as the emerging substitute for direct-to-site advertising.
—
Sources & Further Reading
- Agentic Commerce Impact Could Reach $385 Billion by 2030 — Morgan Stanley
- 7 AI Trends Shaping Agentic Commerce in 2026 — commercetools
- Morgan Stanley: AI Agentic Shoppers Could Drive $385 Billion in Online Sales — Digital Commerce 360
- Agentic Checkout: Google’s New AI Commerce Tools — Google Blog
- Trends in Agentic Commerce for 2026 — Very Good Security
