# What Is Agentic Commerce? Definition, Examples, and How It Works

Agentic commerce is an autonomous AI agent discovering, evaluating and buying products on a shopper's behalf. The definition, live examples and the stack underneath.

By Rohin Aggarwal · 2026-04-12 · (updated 2026-11-11)

A shopper types "buy me a birthday gift for my brother, budget £200" into ChatGPT. An order lands in your store with no session and no funnel behind it. That is agentic commerce. Here is the working definition, three live examples and the stack underneath.

"Agentic commerce" has a marketing problem: everyone says it, almost no one defines it. Vendors stretch the term to mean "we shipped a chatbot" or "we have a recommendation engine". The change underneath goes unnoticed. This piece is two things only: a definition, and the stack it sits on.

**Quick answer**

- Agentic commerce: an autonomous agent buys on a human-issued goal
- Three live examples already running in 2026
- The six-layer stack underneath
- Protocols that matter (MCP, llms.txt, OASF)
- What it means for merchants today

## What is agentic commerce?

> **Working definition:** Agentic commerce is the class of transactions in which an autonomous software agent, operating under a human-issued goal, discovers candidate products, evaluates them against the goal, and either presents a ranked shortlist or completes the purchase with delegated payment.

Three terms in that definition do real work.

- Autonomous: the agent is allowed to make non-trivial choices between candidates without consulting the human at each step.
- Goal: the input is plain language. "A travel jacket I can wear over a suit" instead of "navy travel jacket size M".
- Delegated payment: a wallet, card-on-file, or buy-with-agent flow that lets the agent complete the purchase without re-prompting the user.

Drop any one of these and you are describing something else: a recommender (no agency), a search engine (no natural-language goal), or a chatbot (no delegated payment). The whole reason the term needs a tight definition is that all three keep getting sold under it.

## Three live examples in 2026

### Example 1. Shopper-side personal agent

A consumer opens ChatGPT and says "buy me a birthday gift for my brother, he is into vintage cameras, budget £200". With the user's permission and a connected payment method, ChatGPT queries several retail APIs, scores the options against the goal (vintage, photography, £180-220), picks one, and completes checkout. The user gets an order confirmation and nothing else to do.

The retailer in this flow never sees a session land in their analytics. The conversion shows up as a checkout API call from an OpenAI IP range with a "buy-via-agent" tag in the order metadata. No pageviews, no funnel, no cart-recovery email. Just an order.

### Example 2. Retailer-side concierge agent

A shopper lands on a retailer's homepage and a branded agent greets them. They say what they want ("something for a beach holiday, breathable, under £150"). The agent searches the retailer's own catalogue, returns three matched products with reasoning, and offers to drop the chosen one in the cart. Analytics still record a normal checkout, but the path to purchase is one conversation instead of seven pageviews.

This is where Swap, Klevu's RetailMedia, and Algolia's NeuralSearch sit today.

### Example 3. Marketplace-side agent

A buyer on a B2B marketplace says "source 200 units of biodegradable phone cases: RoHS-compliant, ships from the EU, lead time under three weeks". The marketplace agent fires quote requests at qualified suppliers, scores the responses against the constraints, and hands back a ranked vendor list. The buyer picks one and the marketplace handles escrow. The procurement legwork that used to take a week of emails happens in a single prompt.

Alibaba's "Ask AI" and Amazon Business's "Source-with-agent" are the early production examples.

## How does the stack underneath work?

Six layers, top to bottom.

**The six-layer agentic commerce stack**
1. **Goal layer** — The human input in natural language. Increasingly multimodal (voice + image).
2. **Reasoning layer** — The LLM that plans, decomposes the goal and evaluates candidates. GPT-4-class or better.
3. **Tool layer** — The function calls and API access the agent has. Catalogue search, payment, shipping APIs.
4. **Evidence layer** — Reviews, UGC, structured product data, Q&A. This is where merchants influence outcomes.
5. **Wallet layer** — Delegated payment with spending limits and per-merchant trust scores. ApplePay-with-agent, Visa AI checkout, Stripe Agent Pay.
6. **Audit layer** — Agent logs the user can read, to see why a SKU was chosen.
_Top to bottom. Merchants shape outcomes in the evidence layer._

The evidence layer is the one merchants control. It is also where the structured-data work pays off: the shapes an agent reads are a short, well-bounded list, which we cover in [the 12 JSON-LD shapes agents actually quote](/blog/twelve-json-ld-shapes-agents-quote).

## Protocols that matter

Three emerging protocols are worth tracking.

- MCP (Model Context Protocol), an Anthropic-led standard for how tools expose themselves to LLMs. Increasingly used for retailer APIs.
- llms.txt, a draft standard for telling AI crawlers which pages are canonical and answer-ready. Different from robots.txt; complementary.
- OASF (Open Agent Settlement Framework), emerging payment protocol for agent-to-merchant trust. Still early.

| Pattern | Who runs the agent | What the merchant sees |
| --- | --- | --- |
| Shopper-side personal agent | ChatGPT, Claude, a vertical app | A checkout API call, no session, no funnel |
| Retailer-side concierge | The retailer (on-site) | A normal checkout reached via one conversation |
| Marketplace-side agent | The marketplace platform | A ranked quote and an escrow-handled order |

_The three roles of agentic commerce, side by side._

## What does agentic commerce mean for merchants today?

Five concrete things to do this quarter.

1. Audit your robots.txt: make sure GPTBot, ClaudeBot, PerplexityBot, Google-Extended are explicitly allowed.
2. Ship an llms.txt at /llms.txt that lists your most citable pages with one-line descriptions.
3. Validate your JSON-LD on every PDP, category page, and resource page. Use the schema.org validator.
4. Tag verified-buyer reviews with a clear isVerifiedBuyer field. Agents weight this heavily.
5. Stand up a citation dashboard: at minimum, monitor referer headers from GPTBot, ClaudeBot, PerplexityBot.

We cover the full playbook in the [answer-engine optimisation playbook](/blog/answer-engine-optimisation-2026-playbook), and you can grade your own readiness against [the 42-signal AEO scorecard](/blog/the-42-signal-aeo-scorecard).

## Closing

Agentic commerce is not a 2030 problem. It is a 2026 distribution channel that already moves measurable revenue for the merchants who wired it up. The cost of acting now is small: a few config files, clean structured data, a handful of resource pages. The cost of waiting climbs steadily as more shoppers route through agents and your absence stops being a gap you can close in a sprint and starts being structural.

**In one line:** Agentic commerce is an autonomous agent buying on a human-issued goal: three live examples already running in 2026, and a walk through the protocol stack underneath. Written for operators who want the technical model.

- **41%** — of 18-34s use GenAI to search products (Bazaarvoice 2025 SEI)
- **$2.6T** — Global social commerce 2025 (eMarketer)
- **+108%** — TikTok Shop US growth YoY 2025 (eMarketer)

_AEO + AI shopping benchmarks._

### Sources & notes
- [Bazaarvoice, 2025 Shopper Experience Index](https://www.bazaarvoice.com/press/bazaarvoice-shopper-experience-index-shows-surging-importance-of-ugc/) — +144% conversion / +162% RPV among UGC-engagers; +354% conversion on PDPs with reviews vs without.
- [Model Context Protocol (Anthropic)](https://modelcontextprotocol.io/) — Open standard for exposing tools that AI agents can invoke. Transport layer of agentic commerce.
- [Schema.org](https://schema.org/) — Shared vocabulary for structured-data markup. Defines Product, Review, AggregateRating, FAQPage, DefinedTerm types.
- [Google web.dev, Core Web Vitals + conversion](https://web.dev/articles/vitals) — Brands meeting all three CWV thresholds see 24% lower bounce rates than brands missing any of them.

---
Canonical: https://idukki.io/blog/what-is-agentic-commerce-definition-examples
Tags: agentic-commerce, protocols, how-it-works
