Use Cases: Agentic Commerce in Practice
Agentic Commerce is not a theoretical concept — it is already being deployed. Here are the most important real-world implementations, from production systems to promising pilots.
ChatGPT Instant Checkout
Live (USA)
The first complete Agentic Commerce implementation: users describe their purchase intent in ChatGPT, the agent researches products, shows options and completes the purchase directly in the chat — including payment via the Agentic Commerce Protocol (ACP).
Initial partners are Etsy and selected Shopify merchants. Payment is processed through Stripe with SharedPaymentTokens. The user stores their payment method once in the ChatGPT settings and can then confirm with a single click.
Significance: Proof of concept for the entire industry. Demonstrates that end-to-end Agentic Commerce works technically and is accepted by users.
Amazon Rufus
Live (worldwide)
Amazon's AI assistant Rufus is integrated into the Amazon app and helps users with purchase decisions. Rufus answers questions like "What is the difference between these two vacuum cleaners?", compares products within the Amazon catalog and explains features.
Rufus is intentionally not "agentic" in the strict sense: it advises but does not purchase autonomously. The user decides and clicks "Buy" themselves. Also: Rufus stays within the Amazon ecosystem — there is no integration with external shops or open protocols.
Significance: Shows that the world's largest e-commerce player takes AI-powered commerce seriously — but bets on a closed solution.
Google Business Agent
Beta
Google's Business Agent allows users to chat directly with brands and merchants in Google Search. Instead of visiting a website, the user asks questions in the search interface — and the Business Agent responds with information from the product catalog, availability, business hours and more.
Business Agents are based on the Universal Commerce Protocol (UCP). Brands can train their agents with their own data and thus control what information the agent provides.
Significance: Google integrates Agentic Commerce directly into search — the channel where most purchase decisions begin.
Perplexity Shopping
Live (USA)
Perplexity, the AI search engine, offers product recommendations with source citations. Users can purchase products directly via a PayPal integration. The approach differs from ChatGPT: Perplexity shows more transparently where the recommendations come from (which sources, which reviews).
Significance: Shows that Agentic Commerce is not limited to ChatGPT — various AI platforms are implementing their own shopping experiences.
Check24 Sophie
Live (Germany)
Check24's AI assistant Sophie is one of the few German examples of AI-powered commerce. Sophie helps users in the Check24 app compare rates and offers — from insurance to electricity to travel.
Sophie is primarily an advisory tool, not an autonomous agent: she explains, compares and recommends, but the user concludes the contract manually. Nevertheless, Sophie shows what AI-powered commerce advisory looks like in practice.
Significance: German proof of concept. Shows that comparison platforms are a natural entry point for Agentic Commerce.
Microsoft Copilot Commerce
Beta
Microsoft positions Copilot as a context-aware commerce assistant: shopping recommendations based on calendar events ("birthday gift for tomorrow"), email context ("you asked about printer cartridges") and personal preferences.
The integration into the Microsoft ecosystem (Office, Teams, Windows) gives Copilot Commerce a unique context — no other agent knows as much about the user's daily work routine.
Significance: Agentic Commerce is embedded into the daily workflow — particularly relevant for B2B procurement.
B2B: Automatic Reordering
Concept
A scenario that analysts rate as particularly promising: automatic reordering in the B2B sector. The process:
- An agent monitors the office supply inventory (via IoT sensors or manual input)
- Printer paper is running low → agent identifies the need
- Agent checks existing framework agreements and prices from three suppliers
- Agent selects the cheapest available supplier
- Agent creates an order within the defined mandate limits (AP2)
- Confirmation sent to the purchasing manager via email
This scenario is particularly suited for Agentic Commerce because the decision rules are clear (framework agreement, budget, supplier), the products are standardized and the process is recurring.
Frequently Asked Questions
Which use case is the most advanced?
ChatGPT Instant Checkout is the first production implementation with real transactions. Amazon Rufus and Google Business Agent are also live but more focused on advisory than autonomous purchases.
Are there Agentic Commerce use cases in Germany?
Check24 Sophie is a German example of AI-powered purchase advisory. The full Agentic Commerce experience (autonomous purchase by an agent) is not yet available in Germany — ACP and UCP are currently US-focused.
Does Agentic Commerce also work for B2B?
Yes, and potentially particularly well. Recurring orders (office supplies, consumables) and rule-based procurement are ideally suited for agent automation.