10 Agentic Commerce Examples Defining 2026

TL;DR: Agentic commerce is commerce where an AI agent discovers, carts, and checks out on your behalf. In 2026 the clearest agentic commerce examples are ChatGPT Instant Checkout, Perplexity Buy with Pro, Amazon Rufus, Walmart Sparky, Google's AP2 protocol, the Visa and Mastercard rails, Instacart and DoorDash in chat, agentic travel, B2B procurement bots, and the creation-side outlier — Custyle's AI MerchAgent, which makes net-new products instead of buying existing ones. Salesforce says AI and agents already influenced $67 billion during 2025 Cyber Week.

Table of Contents

• What Counts as Agentic Commerce
• The 10 Examples Defining 2026
• Agentic Commerce Examples Compared
• What These Examples Reveal
• The Creation-Side Frontier
• FAQ

What Counts as Agentic Commerce

Agentic commerce is simple to define and easy to overstate. An AI agent — not a human clicking buttons — discovers a product, builds the cart, and completes the purchase for you, inside limits you set. The human delegates. The agent acts.

It runs on three layers. Discovery agents recommend and compare. Checkout agents transact inside the chat. Autonomous agents act without per-step approval — think auto-buy at a target price, or a procurement bot cutting purchase orders. The examples below climb that ladder, from gentle nudge to full delegation. → New here? Start with what agentic commerce actually is.

One honest caveat first. Market-size estimates for 2025 range from $1.9 billion to $5.7 billion, because firms count differently. Some count only in-agent checkouts. Others count any purchase an agent merely influenced. Hold that gap in mind as the numbers fly.

The signal is real, though. Salesforce reports that during 2025 Cyber Week, AI and agents influenced $67 billion in global sales — roughly 20% of all orders worldwide. Third-party agent traffic from ChatGPT and Perplexity tripled year over year and converted at 8x the rate of social media. Shopify, via Kore.ai, says orders from AI-powered searches grew 15x year over year through 2025. The demos became revenue.

The 10 Examples Defining 2026

1. ChatGPT Instant Checkout

OpenAI's "Buy it in ChatGPT" is the catalyst. It runs on the Agentic Commerce Protocol (ACP), an open standard co-built with Stripe, and it lets you go from chat to checkout in a few taps while the merchant keeps the payment and the customer relationship.

It launched with Etsy in September 2025, added Shopify brands like Glossier, SKIMS, and Vuori, and opened to all U.S. ChatGPT users — free tier included — on February 16, 2026. ChatGPT now fields roughly 50 million shopping queries a day, per Opascope.

The plot twist: by March 2026, CNBC reported OpenAI pivoting away from in-chat checkout toward retailer apps inside ChatGPT that route you back to the store's own site. The first attempt stumbled. It is trying again. That matters — even the leader has not solved the agentic checkout experience. Open-sourcing ACP still forced every rival to answer. Etsy stock popped 16% on the news.

2. Perplexity Buy with Pro

Perplexity built agentic checkout into its answer engine with PayPal. You buy in one click, get free shipping, and never leave the result. It shipped free to all U.S. users.

The proof point is the interesting part. Per Shopify's own analysis, Perplexity shoppers carry a 57% higher average order value than shoppers on other AI platforms. Agentic buyers are not bargain hunters. They are higher-intent, and they spend more.

3. Amazon Rufus

Rufus is the incumbent defending its store. It now takes real agentic actions — auto cart-add, reorder, price alerts, and auto-buy when an item hits your target price. With 300 million-plus users, Amazon credited Rufus with $10–12 billion in incremental annualized sales in Q4 2025.

In 2026 Amazon folded Alexa+ and Rufus together as rivals pushed into shopping, per GeekWire. It also rolled out AI custom-merch design on June 8, 2026 — a competitive note worth holding for later.

4. Walmart Sparky

Sparky is Walmart's assistant, and its trick is where it lives: inside the Walmart app, inside ChatGPT, and inside Google Gemini. One retailer, three surfaces. Sparky users show a 35% higher average order value than non-users.

This is the omni-agent move — a retailer choosing to appear inside third-party agents instead of guarding its own walled garden. Distribution is flipping, and Walmart flipped first.

5. Google AP2 and UCP

Google brought the protocol. The Agent Payments Protocol (AP2) arrived September 16, 2025 with 60-plus launch partners, including Mastercard, PayPal, Coinbase, Amex, and Salesforce. At NRF in January 2026 it added the Universal Commerce Protocol (UCP), letting agents reach merchant catalogs, carts, and checkout through one open standard.

AP2's design is the clever bit. Every purchase becomes three signed Mandates — Intent (what you asked for), Cart (what the agent assembled), and Payment (what gets charged) — spanning cards and stablecoins. In May 2026, Google donated AP2 to the FIDO Alliance for community governance. That makes it the interoperability anchor of the whole field. (Google's AP2 announcement.)

6. Visa and Mastercard Rails

Trust is the gating problem. When an agent buys the wrong thing, who pays? The card networks are racing to own that answer.

Visa Intelligent Commerce authenticates, authorizes, and tokenizes agent-initiated payments, with Intelligent Commerce Connect in pilot. Visa and OpenAI announced a strategic collaboration on June 10, 2026. Mastercard Agent Pay brings agentic tokens and agent-aware identity so agents can start, authorize, and complete a transaction. The distinction to remember: AP2 is an open governance protocol, while Visa Trusted Agent and Mastercard Agent Pay are network-specific implementations of agent-authorized payment.

7. Instacart and DoorDash

Groceries and restaurants prove agentic commerce works for the highest-frequency category — replenishment. Instacart launched as a grocery partner through ChatGPT Apps on December 8, 2025. DoorDash is an ACP partner. You order food and weekly staples inside the chat.

Discretionary retail gets the headlines. Replenishment gets the repeat behavior. That is where agents earn a permanent seat.

8. Agentic Travel

Travel is the stress test — multi-vendor, high-ticket, and full of changes. Expedia's AI agent handles bookings, cancellations, and support in one interaction. Trip.com's TripGenie builds multi-leg itineraries across flights, hotels, and activities, adapts on the fly, and completes the booking in the same flow.

Expedia, Booking.com, Sabre, PayPal, Amazon, and Google are all building into this space. If an agent can book a three-city trip without melting down, it can handle most of retail.

9. B2B Procurement Bots

The biggest dollars are the ones you never see. Enterprise agents now run with delegated authority — they research vendors, shortlist, check compliance, negotiate, and trigger purchase orders within policy, no human review per step.

Named examples: Lio raised a $30M Series A from a16z in March 2026 for exactly this. JAGGAER runs Autonomous Commerce across source-to-pay. Results are concrete — one retail chain cut stockouts 31% and excess inventory 18%; one aircraft maker trimmed active inventory 30% for roughly $700M in added EBIT. Gartner projects agents will handle 90% of B2B buying by 2028, routing around $15 trillion in spend.

One detail brands should underline: agents favor suppliers with clean, structured, machine-readable data. PDF catalogs go invisible. → See how this is rebuilding online retail end to end.

10. Custyle AI MerchAgent

Every example above automates buying something that already exists. This one is different. Custyle bills itself as the world's first AI MerchAgent™. You type a prompt, drop an image, or describe a vibe. A nine-agent AI Crew handles creative direction, design, mockup, virtual try-on, build method, and fulfillment through partners like Printful and Printify. Idea to real merch in about a minute. No design skills.

The timing is the tell. Amazon launched AI custom-merch design on June 8, 2026 — the largest store on earth validating the category. Custyle was there first and is purpose-built for it. This is the creation-side of agentic commerce, and it is the differentiated frontier, covered in full below.

Agentic Commerce Examples Compared

Example	Surface	Transaction layer	Payment	Proof metric	Category
ChatGPT Instant Checkout	ChatGPT	In-chat (ACP), pivoting to apps	Stripe	50M shopping queries/day	General retail
Perplexity Buy with Pro	Perplexity	In-chat one-click	PayPal	+57% AOV vs other AI	General retail
Amazon Rufus	Amazon app	Auto-buy, reorder, alerts	Amazon Pay	~$12B incremental; 300M users	General retail
Walmart Sparky	App + ChatGPT + Gemini	Multi-surface assistant	Walmart	+35% AOV	Mass retail
Google AP2 / UCP	Gemini + partners	Open protocol, 3 Mandates	Cards + stablecoins	60+ launch partners	Infrastructure
Visa / Mastercard	Network rails	Agent auth + tokenization	Visa / Mastercard	Visa × OpenAI, Jun 2026	Payments
Instacart / DoorDash	ChatGPT Apps	In-chat ordering	ACP / Stripe	Launched Dec 8, 2025	Grocery / local
Expedia / TripGenie	Travel apps + agents	Itinerary + booking	PayPal / cards	Multi-vendor booking	Travel
B2B procurement (Lio)	Enterprise S2P	Delegated PO execution	Enterprise rails	−30% inventory, +$700M EBIT	B2B
Custyle AI MerchAgent	Custyle.ai	Generative create → fulfill	Printful / Printify	Idea → merch in ~1 min	Custom merch (creation-side)

What These Examples Reveal

Five patterns run through the list.

Protocols are fighting, then converging. ACP from OpenAI and Stripe competes with Google's AP2 and UCP, which competes with the Visa and Mastercard schemes. Yet FIDO Alliance governance, taken on in May 2026, is pulling them toward interoperability. Expect a messy middle and a standardized end.

Distribution is flipping. Retailers now show up inside other companies' agents. Walmart Sparky lives in ChatGPT and Gemini. The storefront is no longer the destination — the agent is.

Agentic shoppers spend more. They convert at 8x social media, and they carry higher carts — Perplexity at +57%, Sparky at +35%. This is quality of intent, not just volume of clicks.

Trust is the real bottleneck. Signed mandates, tokenization, agent identity — the field is racing to answer one question: who is liable when an agent buys wrong? Solve that, and the rest scales.

Structured data decides visibility. Both procurement bots and consumer agents reward clean, machine-readable catalogs. Unstructured listings go dark. Forecasts agree the prize is enormous — McKinsey sees agentic AI influencing $3–5 trillion in retail by 2030, Morgan Stanley pegs U.S. agentic spend at $190–385 billion, and Bain estimates $300–500 billion. (McKinsey's agentic commerce outlook.)

The Creation-Side Frontier

Read the list again and one split jumps out. Nine of these examples make it faster to acquire a product that already exists. One makes it possible to create a product that never did.

That is the line worth remembering. Most agentic commerce helps an AI buy you something off a shelf. Custyle's AI MerchAgent helps it make something from a sentence — design, mockup, build method, and fulfillment, handled by a nine-agent Crew in about a minute. Call it generative agentic commerce.

It also sidesteps the visibility problem. As agents reward structured, machine-readable catalogs, products born digital-native are agent-ready by default. There is no PDF to go invisible. When Amazon entered AI custom merch in June 2026, it confirmed the category rather than created it.

The takeaway is not which agent wins checkout. It is that agents are about to do two jobs, not one — buy what exists, and make what doesn't. Most of 2026's examples cover the first. The second is where the next category gets built. Want the short version of the whole shift? → Start with what agentic commerce is, then come back for the frontier.

FAQ

What is an example of agentic commerce? ChatGPT Instant Checkout is the clearest one — you describe a product and buy it inside the chat in a few taps. Other examples include Perplexity Buy with Pro, Amazon Rufus auto-buy at a target price, and Walmart Sparky across app, ChatGPT, and Gemini.

What's the difference between agentic commerce and AI recommendations? Recommendations stop at discovery — the AI suggests, you click and buy. Agentic commerce adds transaction execution. The agent builds the cart and completes the purchase for you, within limits you set, sometimes with no per-step approval.

Which companies are leading agentic commerce in 2026? OpenAI, Google, Amazon, Walmart, and Perplexity lead on the consumer side. Visa and Mastercard own the payment rails. In B2B, names like Lio and JAGGAER lead delegated procurement. Salesforce reports agents already influenced $67 billion during 2025 Cyber Week.

Can AI agents create products, not just buy them? Yes. Generative agents like Custyle's AI MerchAgent design and fulfill custom merch from a prompt — idea to real product in about a minute, no design skills. It is the creation-side of agentic commerce, distinct from agents that only buy existing inventory.

Is agentic commerce actually making money yet? It is. Beyond Salesforce's $67 billion Cyber Week figure, Shopify reports AI-search orders grew 15x year over year, and Amazon credited Rufus with $10–12 billion in incremental annual sales. Forecasts run from $190 billion to over $3 trillion by 2030, depending on how you count.