The Definition

Agentic commerce is the use of AI agents to autonomously discover, evaluate, negotiate, and complete purchases on behalf of a human user. The keyword is autonomously. Not a chatbot that fetches a product page. Not a voice assistant that places a reorder. An agent that takes a goal ("find me a lightweight laptop under $1,200 with at least 16 GB RAM, prioritize battery life, buy the best option") and executes the entire chain: search, compare, read reviews, check inventory, apply coupons, and check out, all without human hand-holding at each step.

This is distinct from conversational commerce, where a human chats with a bot to narrow choices but still clicks "buy." It is distinct from voice commerce, where you tell a smart speaker to reorder detergent. Both of those are human-in-the-loop. Agentic commerce removes the loop. The human sets intent and constraints. The agent handles execution.

Why does the distinction matter? Because the removal of the human from each micro-decision changes everything about how products get discovered, compared, and sold. When a person browses, they are swayed by layout, branding, emotional copy, and impulse. When an agent shops, it reads structured data: specifications, price, availability, shipping time, return policy. The storefront becomes an API endpoint.

The Autonomy Spectrum

Not all AI shopping is created equal. The market is sliding along a spectrum, and most of what gets labeled "agentic commerce" today sits somewhere in the middle.

Traditional ecommerce. Human browses, human clicks, human pays. The website is the entire experience. AI-assisted shopping. A chatbot or search bar powered by an LLM helps narrow options. Amazon Rufus sits here: it answers questions, compares products, but you still click the button. Shopify data suggests roughly 20% of engaged Rufus shoppers ask for brand-specific information, which means the AI is influencing discovery but not closing the deal. Delegated agentic commerce. The human specifies intent and constraints. The agent handles discovery, evaluation, and purchase with minimal check-ins. This is what Amazon Alexa for Shopping and Google Gemini Intelligence are building toward. The human might approve the final cart or set a spending ceiling, but the heavy lifting is autonomous. Fully autonomous commerce. The agent has ongoing authority to purchase within defined parameters. Think: a household AI that restocks groceries weekly based on consumption patterns, or a procurement agent that negotiates bulk pricing with suppliers. Mastercard and PhotonPay demonstrated the first live agentic payment on May 14, 2026, a proof of concept for exactly this tier.

Most of the industry is somewhere between AI-assisted and delegated right now. Fully autonomous is the destination, but the infrastructure, trust, and regulation are still catching up.

How Agentic Commerce Actually Works

Strip away the marketing slides and the process looks like this:

1. Intent capture. The user expresses a goal, either explicitly ("I need running shoes for a marathon in October") or implicitly through behavior and context. The agent parses this into a structured request: product category, constraints (budget, timeline, preferences), and priorities.

2. Discovery. The agent queries multiple sources: retailer APIs, product databases, review aggregators, price comparison engines. Unlike a human, it can scan thousands of SKUs in seconds. It does not care about hero banners or "recommended for you" carousels.

3. Evaluation. The agent scores products against the user's criteria. This is where structured data becomes currency. Reviews, specifications, return policies, shipping windows, and total cost (including tax and shipping) all get weighted. An agent can factor in things humans struggle with, like cross-referencing 400 reviews for sentiment patterns.

4. Negotiation. In more advanced setups, the agent can negotiate price, apply coupon codes, or select bundled offers. This is early-stage but moving fast. Alibaba's Qwen-powered Taobao integration already handles dynamic pricing queries.

5. Payment. The agent executes the transaction using stored payment methods or embedded payment rails. The Mastercard-PhotonPay agentic payment was a milestone here: the first time an AI agent completed a live commercial transaction through purpose-built payment infrastructure.

6. Fulfillment and feedback. The agent tracks the order, handles returns or issues, and learns from the outcome to improve future purchases.

The entire loop can take seconds. A human doing the same research might spend 45 minutes across six tabs.

May 2026: The Breakout Month

If you had to pick a single month where agentic commerce stopped being a conference-panel buzzword and started being a product category, May 2026 is it. The density of launches is remarkable.

Amazon Alexa for Shopping went live on May 13. This is not the Alexa that tells you the weather. It is a purpose-built shopping agent that can browse Amazon's catalog, compare products across criteria you specify, and complete purchases with voice or app approval. It is the first time Amazon has let an AI agent act as the primary shopping interface rather than a support tool. Google Gemini Intelligence landed on May 12, embedded into Search and Shopping. Google is approaching agentic commerce from the discovery side: Gemini can now hold multi-turn shopping conversations, compare products across retailers, and surface structured data in a way that is clearly designed for agent-to-agent interactions, not just human reading. As Google's own official AI search optimization guide acknowledges, the shift toward AI-mediated discovery is reshaping how content needs to be structured. OpenAI Personal Finance launched May 15. While positioned as a financial management tool, its shopping module can evaluate purchases against budget constraints, find better deals, and in limited cases execute transactions. It is the clearest signal yet that OpenAI sees commerce as a core use case for agents. Walmart Sparky is Walmart's answer: an in-app AI shopping assistant that handles product discovery and comparison. The data is already interesting: Walmart reports that shoppers who engage with Sparky spend 35% more per order. That is not a typo. Whether it is sustained or a novelty bump remains to be seen, but the early signal is strong. Alibaba Taobao/Qwen went live with a deeply integrated agentic shopping experience in its Chinese marketplace. The Qwen-powered Taobao integration is arguably the most advanced implementation globally, partly because Chinese consumers have higher comfort with AI-driven shopping and partly because Alibaba controls the entire stack from LLM to marketplace to payments.

And then there is Shopify, which reported that orders originating from AI-powered searches are up 13x year-over-year. Not 13%. Thirteen times. This is not a retailer launching a feature; it is an entire commerce platform seeing a structural shift in how products get discovered and purchased.

The Evidence It Is Real

The data points are stacking up fast enough that dismissing agentic commerce as hype requires willful ignorance.

Adobe reported that AI-referred traffic to retailers is up 393% year-over-year. That is not a trickle. It is a flood. When AI agents discover and recommend products autonomously, the traffic pattern changes fundamentally. It is not a human clicking a Google link. It is an agent querying a product database.

NIQ found that 42% of consumers are willing to let AI shop for them. Willingness does not equal behavior, and survey data should always be treated with skepticism, but 42% is a number that would have been single digits two years ago. Consumer attitudes are shifting faster than the technology.

ResearchAndMarkets adds nuance: 62% of AI adoption in commerce is concentrated at the comparison and research stage, versus 23% at checkout and 19% post-purchase. This maps to the autonomy spectrum. We are in the early phases where AI handles discovery and evaluation but humans still want to approve the final purchase. The checkout and post-purchase phases will automate as trust builds.

Market projections vary wildly, as they always do with emerging categories. Estimates range from $190 billion to $385 billion by 2030. The wide range tells you the analysts are guessing. But the direction is unanimous: this is a major market forming in real time.

The Gaps: Why It Is Not Overnight

Here is the counterweight. Criteo, one of the largest commerce advertising platforms globally, stated in its most recent earnings call that its "guidance does not assume any material revenue contribution from agentic AI." Translation: the ad-tech infrastructure that powers modern ecommerce does not yet see meaningful dollars flowing through agentic channels. When a company whose entire business depends on commerce traffic says "not yet," it is worth listening.

The accuracy problem is real. AI agents are only as good as the data they consume, and product data across the internet is messy. Inconsistent specifications, outdated pricing, missing inventory information, and conflicting reviews all introduce errors. An agent that buys the wrong product because it misread a spec sheet is a liability, not a convenience.

Payment infrastructure is still catching up. The Mastercard-PhotonPay transaction was a proof of concept, not a scaled system. Existing payment rails were built for human-initiated transactions with fraud detection keyed to human behavior patterns. Agentic payments need their own fraud models, authentication mechanisms, and dispute resolution frameworks.

Trust is the deepest gap. Consumers may say they are willing to let AI shop for them, but actual delegated purchasing authority is a different psychological threshold. People who happily let Netflix auto-play the next episode are nervous about an AI spending $200 on their behalf. The trust curve will be gradual, not a hockey stick.

What Brands Need to Do Now

The brands that win in agentic commerce will not be the ones with the prettiest websites. They will be the ones with the most machine-readable product data. This is not speculative. It is already happening.

Structured product data is the new SEO. When an AI agent evaluates your product, it reads your schema markup, your API responses, and your structured feeds. If your product specifications are buried in a paragraph of marketing copy, the agent will miss them. If your pricing is dynamically generated and not available in a structured format, the agent will skip you for a competitor whose pricing is clean. Pricing transparency matters more than ever. Agents compare total cost, not sticker price. If you hide shipping fees until checkout, the agent will flag you as opaque and deprioritize you. Flat-rate shipping, clear return policies, and upfront pricing are no longer nice-to-haves. They are discoverability requirements. Agent-friendly APIs are becoming a competitive advantage. Amazon's Alexa for Shopping launch works partly because Amazon has the most comprehensive product API in the world. Smaller retailers who expose clean, structured APIs will be discoverable by agents. Those who do not will be invisible. Review data is agent food. Agents parse review sentiment at scale. A product with 500 reviews averaging 4.2 stars but with a recent cluster of complaints about durability will get flagged. Monitoring and responding to review sentiment is no longer just reputation management. It is agent optimization. The creative layer does not disappear, but it shifts. Brand storytelling, emotional connection, and visual identity still matter for the human who receives the product and decides whether to trust the agent's next recommendation. But the top-of-funnel discovery and evaluation process is increasingly agent-mediated. Brands need both: structured data for the agents, and compelling brand experiences for the humans.

---

Is your commerce setup ready for AI agents? Find out with Searchless.ai's agentic commerce audit and see how your product data performs in an agent-driven world.

Sources

  • Shopify Q1 2026 merchant data: AI-powered search orders up 13x YoY (Shopify)
  • Adobe Digital Insights: AI-referred retail traffic up 393% YoY (Adobe)
  • Walmart Q1 2026: Sparky AI shoppers spend 35% more per order (Walmart Corporate)
  • Amazon Rufus engagement data: ~20% of engaged shoppers request brand info (Amazon)
  • NIQ Consumer Survey: 42% willing to let AI shop for them (NIQ)
  • ResearchAndMarkets: 62% AI adoption at comparison stage, 23% checkout, 19% post-purchase (ResearchAndMarkets)
  • Criteo Q1 2026 earnings: no material agentic AI revenue in guidance (Criteo Investor Relations)
  • Mastercard + PhotonPay: first live agentic payment, May 14, 2026 (Mastercard)
  • Market projections: $190-385B by 2030 (various analyst estimates)

---

Ready to make your brand discoverable by AI agents? Explore Searchless.ai pricing and start optimizing for the agentic commerce era.