April 7, 2026 · 14 min read

Agentic Commerce: When AI Agents Do the Buying and Selling

Your customers' AI agents are already shopping. Walmart, Google, and Shopify have gone live. Here's what agentic commerce means for operators — and the 5-step playbook to get ready before your competitors do.

$1T
Agent-mediated retail by 2030 (McKinsey)
73%
Of consumers using AI in shopping (2026)
50%
Of shoppers will use AI agents by 2030 (Morgan Stanley)

The Shopping Cart Is Dead. Long Live the Agent.

Here's a scene that's already playing out millions of times a day: a consumer opens ChatGPT, Gemini, or Perplexity and says, "Find me noise-canceling headphones under $200 that work well for open offices."

No Google search. No website visits. No comparison tabs. No cart page. The AI agent researches, compares, and presents a curated shortlist. In some cases, it completes the purchase without the consumer ever seeing a checkout page.

This is agentic commerce — and it's not a 2030 prediction. It's happening right now.

In February 2026, Walmart announced a partnership enabling purchases through Google's Gemini app. Google launched "Buy for me" functionality in AI Mode search. Perplexity rolled out conversational product discovery with instant checkout via PayPal. OpenAI introduced shopping research in ChatGPT using reinforcement learning for comparative product guides.

The world's largest retailers aren't experimenting anymore. They're building new storefronts — inside AI agents.

What Exactly Is Agentic Commerce?

Agentic commerce is a model where software agents act on behalf of consumers to search for products, evaluate options, negotiate terms, and complete transactions — all within predefined parameters set by the human.

It's not a chatbot recommending products. It's not a search engine with AI summaries. It's a fundamental shift in who makes the purchasing decision.

❌ Traditional E-Commerce

  • Human searches, browses, compares
  • Visits multiple websites
  • Manually adds to cart
  • Enters payment details
  • Clicks "Buy Now"

✅ Agentic Commerce

  • Human sets goal + constraints
  • Agent researches across sources
  • Agent compares and filters
  • Agent negotiates or selects deal
  • Agent completes purchase autonomously

"AI effectively functions as a personal digital agent that can source and negotiate and complete purchases using pre-approved payment methods on behalf of the customer," said Mladen Vladic, Head of Product at FIS, in a PYMNTS interview.

The key word here is pre-approved. The human sets the guardrails — budget, brand preferences, quality requirements — and the agent executes within those boundaries.

The 7 Forces Driving Agentic Commerce in 2026

Force 1

GenAI Platforms Become Commerce Channels

ChatGPT, Gemini, and Perplexity have evolved from assistants into full retail channels. A recent study found 73% of consumers already use AI in their shopping journey — 45% for product ideas, 37% for summarizing reviews, 32% for comparing prices. And 70% are "at least somewhat comfortable" with an AI agent making purchases on their behalf.

Force 2

Zero-Click Commerce Disrupts Traditional Retail

When an AI agent handles the entire purchase flow, consumers never click, search, or visit a website. This means traditional SEO, PPC, and landing page optimization become less effective. The brands that win will be the ones whose product data is machine-readable and structured for agent consumption.

Force 3

New Protocols Enable Agent Interoperability

Google's Universal Commerce Protocol (UCP), launched January 2026 with 20+ partners including Shopify, Walmart, Target, Stripe, and Visa, creates the infrastructure for agents to transact across platforms. Google's Agent-to-Agent (A2A) protocol adds a transport layer for agent coordination. Combined with MCP for tool access, agents now have the plumbing to actually buy things.

Force 4

Payments Infrastructure Adapts to Agent Transactions

FIS, Stripe, Visa, and Mastercard are building agent-aware payment rails. Transactions still flow through existing authorization, authentication, and dispute frameworks — but the initiator is now software, not a human clicking "Buy." Payment providers are becoming decision engines, not just transaction processors.

Force 5

Discovery Dominates — Post-Purchase Follows

Right now, most agentic capabilities cluster at the top of the funnel: product discovery, comparison, and recommendation. Cart creation, checkout, and payment authorization are catching up fast. Post-purchase (returns, loyalty, support) is the next frontier — and the brands that solve it first will lock in agent loyalty.

Force 6

Purpose-Built Agents Beat All-in-One Solutions

Enterprises aren't deploying one massive commerce agent. They're building small, high-trust agents embedded into specific workflows — a procurement agent, a reorder agent, a price comparison agent. This modular approach delivers faster ROI and lower risk than trying to automate everything at once.

Force 7

B2B Presents the Biggest Opportunity

While consumer-facing agentic commerce gets the headlines, B2B is where the real money is. Agents automating order workflows, approval chains, and supplier negotiations can save enterprises millions. Forrester predicts 1 in 5 sellers will need to respond to AI-powered buyer agents with dynamic counteroffers by 2027.

Who's Already Live?

This isn't theoretical. Here's what's deployed right now:

Platform What They Launched Status
Google "Buy for me" in AI Mode search + Gemini, agentic checkout on retailer sites Live (US)
Walmart + Google Agent-mediated purchases through Gemini app Live
Perplexity Conversational product discovery + instant checkout via PayPal Live
OpenAI / ChatGPT Shopping research with RL-powered comparative product guides Live
Google + Etsy/Wayfair UCP-powered in-search purchasing without leaving Google Live
Shopify Agentic commerce APIs for merchants to be "agent-discoverable" Rolling out
Target + OpenAI Agent-assisted product discovery and checkout Live
💡 Operator Insight:

Notice the pattern? The largest retailers moved first. As FIS's Vladic noted: "We are seeing something different with agentic commerce where you have the largest brands announcing first." When Walmart and Target go live, their suppliers and partners must follow — or become invisible to AI agents.

Answer Engine Optimization (AEO): The New SEO

Here's the uncomfortable truth for marketers: if an AI agent can't parse your product data, your product doesn't exist in agentic commerce.

Traditional SEO optimized for humans reading search results. AEO optimizes for machines reading structured data. The difference is fundamental:

Traditional SEO

  • Keyword-rich page titles
  • Backlink building
  • Click-through rate optimization
  • Human-readable product descriptions
  • Review aggregation widgets

Answer Engine Optimization

  • Structured JSON-LD product data
  • Machine-readable specifications
  • Agent-accessible API endpoints
  • Standardized attribute schemas
  • Real-time inventory + pricing feeds

Your product catalog needs to speak the language of AI agents: structured, standardized, and always up-to-date. Enriched metadata, clean taxonomy, and complete product attributes aren't nice-to-haves anymore — they're the difference between being recommended and being invisible.

The $1 Trillion Question: What Changes for Operators?

McKinsey estimates agent-mediated purchasing could account for $1 trillion in US retail revenue by 2030. That's four years away.

For operators — the people who build and run AI systems for businesses — this creates three massive opportunities:

1. Agent-Readiness Consulting

Most businesses have no idea their product data isn't agent-readable. They don't have structured JSON-LD schemas. Their APIs aren't designed for agent consumption. Their inventory feeds aren't real-time. This is a consulting goldmine.

2. Commerce Agent Development

Businesses need agents on both sides: buyer agents that find the best deals for procurement teams, and seller agents that respond to AI-powered inquiries with dynamic pricing and offers. Building these purpose-built commerce agents is high-value work.

3. Agent Analytics and Attribution

Traditional analytics is blind to agent-mediated purchases. When an AI agent buys something, there's no session data, no click path, no UTM parameter. Businesses need new attribution models — and the operators who build them will be in high demand.

⚠️ The Analytics Gap:

Your analytics stack was built for a world where customers click links, visit pages, and generate session data. AI-mediated purchases break every one of those assumptions. If you can't measure it, you can't optimize it.

5-Step Playbook: Preparing for Agentic Commerce

Step 1 — Audit Your Product Data

Make Your Catalog Machine-Readable

This week: Run your top 100 SKUs through Google's Structured Data Testing Tool. Check for complete JSON-LD Product schema, including price, availability, brand, description, and image. Target: every product should have at minimum 15 structured attributes. Missing attributes = invisible to agents.

Step 2 — Implement Agent-Accessible APIs

Build the Agent Storefront

Week 2: Create or expose read-only APIs for product search, filtering, pricing, and availability. Ensure sub-200ms response times. Add standardized attribute schemas (color, size, material, compatibility). Support Google's Universal Commerce Protocol if you're in retail. Think of this as building a new storefront — except the customer is an AI agent.

Step 3 — Optimize for Answer Engines

AEO Before SEO

Week 3: Restructure product descriptions to answer specific questions directly (not just "features and benefits" marketing copy). Create comparison-ready content. Ensure your FAQ sections cover the questions AI agents ask: compatibility, warranties, shipping, return policies. Every product page should be a complete answer, not a sales pitch.

Step 4 — Build Agent Transaction Flows

Enable Autonomous Checkout

Week 4: Work with your payment provider (Stripe, Adyen, etc.) to support agent-initiated transactions. Implement server-to-server checkout that doesn't require a browser session. Add support for pre-authorized payment methods. Test the full flow: agent discovery → comparison → add to cart → checkout → confirmation, without any human browser interaction.

Step 5 — Measure What Agents Can't Tell You

Build Agent-Aware Analytics

Ongoing: Identify agent-mediated traffic patterns (unusual session lengths, zero page views before purchase, API-only interactions). Build attribution models that account for zero-click conversions. Track which AI platforms drive purchases. Create dashboards that separate human and agent conversion funnels. This is your competitive advantage.

The B2B Opportunity Nobody's Talking About

While everyone focuses on consumer shopping, B2B agentic commerce might be the bigger story.

Consider what AI agents can automate in B2B procurement:

Forrester predicts that by 2027, 1 in 5 B2B sellers will be compelled to respond to AI-powered buyer agents with dynamically delivered counteroffers via seller-controlled agents.

💡 For Operators:

B2B commerce agents have higher price tags, longer contracts, and stickier relationships than consumer tools. A procurement agent that saves an enterprise $500K/year in better pricing and reduced manual work commands serious consulting fees. This is where the real revenue is.

5 Mistakes That Will Kill Your Agentic Commerce Strategy

1. Treating It Like Another Sales Channel

Agentic commerce isn't just a new channel to push products through. It fundamentally changes who makes the buying decision. Your marketing needs to convince algorithms, not humans. Your pricing needs to be competitive in real-time, not "call for a quote." Your product data needs to be perfect, not "good enough for the website."

2. Ignoring Structured Data Until It's Urgent

Every day without proper structured data is a day your products are invisible to AI agents. This isn't a 2028 problem. Google's "Buy for me" is live right now. If your JSON-LD is incomplete, your products aren't in the consideration set.

3. Building One Giant Commerce Agent

Enterprises that try to build a single agent handling everything — from product discovery to checkout to returns — will fail. The winning approach is purpose-built agents: a pricing agent, a recommendation agent, a fulfillment agent. Small, focused, testable, composable.

4. Neglecting Agent Trust and Safety

When an AI agent completes a purchase worth $5,000 on behalf of a business, who's liable if it goes wrong? What happens when an agent exploits a pricing error? You need clear guardrails: spending limits, approval thresholds, human-in-the-loop for high-value transactions, and audit trails for every agent action.

5. Waiting for Standards to Mature

UCP, A2A, MCP — the protocols are still evolving. But waiting for "the standard" to settle is a trap. The companies implementing now are training their systems, building agent-readable catalogs, and establishing relationships with AI platforms. By the time standards crystallize, early movers will have years of data advantage.

What Happens When Agents Negotiate With Agents?

The next evolution is already visible: agent-to-agent commerce.

Imagine a buyer's procurement agent sending a structured RFQ to three sellers' agents. Each seller agent evaluates the request against inventory, margin targets, and relationship history, then responds with a tailored offer — all in under 30 seconds. The buyer agent evaluates the responses, negotiates on price and terms, and presents the human buyer with a recommended winner and rationale.

Google's A2A protocol, launched in 2025 and now adopted by companies like Nokia, Salesforce, and SAP, provides the transport layer for this kind of coordination. Combined with MCP for tool access and UCP for transaction execution, we're seeing the emergence of an "agent internet" for commerce.

"This is a transformational inflection point in the industry. Not only in this country, but globally." — Mladen Vladic, Head of Product, FIS

The implications are staggering. When agents negotiate with agents, the speed of commerce increases by orders of magnitude. Pricing becomes truly dynamic. Supplier relationships are optimized continuously. And the businesses without agents in this game simply don't get the best deals.

The Operator's Bottom Line

Agentic commerce is not a feature you bolt onto your existing business. It's a fundamental rewiring of how commerce works.

The pattern is clear: largest players move first (Walmart, Google, Target), standards emerge quickly (UCP, A2A), and the rest of the market must follow or become invisible. The window for early-mover advantage is measured in months, not years.

If you're an operator, this is the biggest greenfield opportunity since the rise of eCommerce itself. Businesses need agent-ready product data, commerce agents for buying and selling, and analytics that work in a zero-click world. The companies that solve these problems will capture a meaningful slice of $1 trillion in agent-mediated commerce.

If you're a business owner, the question isn't whether to prepare for agentic commerce — it's how fast you can make your products discoverable by AI agents. Start with structured data. Build agent-accessible APIs. And find an operator who understands this new world.

The shopping cart is dead. The agent is the new customer.

🛒 Ready to Build AI Agents That Drive Revenue?

The AI Employee Playbook covers agent architecture, deployment, and monetization — including the commerce patterns that matter in 2026.

Get the Playbook — €29

Sources