Introduction: The Future of Shopping—Can We Trust AI to Handle Purchases?
Artificial intelligence has seamlessly integrated into our lives, from asking voice assistants like Siri for weather updates to using maps for navigation. Yet one major area has remained largely manual: shopping. Despite the rise of recommendation engines and one-click ordering, we still finalize most purchases ourselves. That began to change in late 2025 with Instant Checkout, a feature introduced in ChatGPT, which allows AI to purchase products on behalf of users directly within the chat.
This new functionality, powered by the Agentic Commerce Protocol (ACP), is transforming how we think about shopping. In this blog, we dive into how Instant Checkout works, why it represents a significant shift towards agentic commerce, and the challenges that brands must overcome to build trust and credibility in this emerging era of automated shopping.
How Does Macaron's Instant Checkout Work?
The Agentic Commerce Protocol: The Backbone of Instant Checkout
At the heart of Instant Checkout is the Agentic Commerce Protocol (ACP), an open, draft standard co-developed by OpenAI and Stripe. ACP allows AI agents, like ChatGPT, to not only recommend products but also execute purchases. When a user interacts with the AI, it can initiate a transaction, gather shipping details, and complete the payment—all within the chat interface, without the need to visit external websites.
Initially, Instant Checkout is connected to U.S.-based Etsy sellers, and soon will expand to over a million Shopify merchants. For consumers, the process is straightforward: the AI facilitates a purchase for a single item, and prices remain the same. Merchants pay a fee to OpenAI for each completed transaction, but consumers incur no additional charges.
The protocol enables an AI agent to create, update, and complete checkout sessions with merchants. When you request to buy an item, the AI uses the createCheckoutSession endpoint to retrieve the price, shipping methods, and any applicable taxes from the merchant. If you change details like shipping address or product quantity, the AI updates the session accordingly. Once confirmed, the payment is processed through Stripe and the order is finalized.
The Growth of Instant Checkout
While Instant Checkout currently supports only single-item purchases, OpenAI plans to expand its functionality to multi-item carts and integrate with more merchants and regions in the future. This feature marks a clear step towards an AI-driven shopping experience that could redefine e-commerce.
What is Agentic Commerce? Shifting from Assisted to Autonomous Shopping
How Agentic Commerce Is Different from Traditional E-Commerce
Traditional shopping tools like recommendation engines and one-click payments have made shopping more convenient. However, these systems remain assistive—they help users, but the consumer still needs to finalize the purchase. Agentic commerce takes this further by allowing AI to act entirely on behalf of the consumer. Imagine telling ChatGPT, "Buy the best blender under $200," and the AI takes care of everything, from selecting the product to completing the transaction.
This shift introduces a major change in how we interact with e-commerce platforms. Agentic commerce enables consumers to fully delegate the shopping process to an AI, potentially reducing friction and streamlining purchases.
Trust Issues: Can Consumers Trust AI to Make Purchases?
Building Trust in AI Shopping Agents
Despite the potential for AI to transform e-commerce, the major barrier to widespread adoption is trust. A 2025 Bain & Company survey showed that only 10% of consumers have used AI to make a purchase, with just 24% feeling comfortable allowing AI to complete a transaction. However, 64% of consumers are open to using AI to make a purchase, and 73% would consider using AI to research products. While adoption is still in its early stages, these numbers indicate that a large portion of consumers is ready to embrace the idea once they feel more confident in AI’s capabilities.
Consumer Concerns: Privacy, Security, and Control
Consumers are often hesitant to trust AI with their financial information. According to a Salesforce report, 63% of consumers believe trust is more important due to advances in AI, and 68% are concerned about communicating with AI agents. Younger generations, such as Gen Z, show more openness, with 43% willing to let an AI agent handle purchases. Additionally, 53% would use an AI to avoid repeating themselves, and 51% would value faster service through AI.
Ensuring Transparency and Control
Building trust will require businesses to ensure transparency and provide users with control over their transactions. Clear information about how data is used, visible opt-in processes, and strong privacy protections are essential to reassuring consumers. By addressing these concerns, businesses can pave the way for broader adoption of agentic commerce.
Potential Issues: What Could Go Wrong in AI-Powered Purchases?
Misunderstandings and Erroneous Orders
AI may make mistakes, especially if the user’s request is ambiguous or unclear. For example, if you ask for the "best blender under $200," the AI may choose an option that doesn’t meet your preferences. Similarly, AI could misapply shipping methods or discounts, leading to frustration, especially with time-sensitive orders.
To mitigate these risks, the Agentic Commerce Protocol includes a confirmation step before completing purchases, ensuring that users have the opportunity to review and correct orders before finalizing.
Unauthorized Purchases and Fraud Risks
One concern is the potential for unauthorized purchases, especially if a device is shared among multiple users. Children might inadvertently place an order using a voice assistant, or a hacker could exploit vulnerabilities. To address these concerns, businesses should implement multi-factor authentication, secure payment processing, and allow users to easily modify or cancel transactions.
Subscription Traps and Hidden Fees
AI agents could also unintentionally enroll users in subscriptions or mislead them about the nature of the purchase. Clear dialogues, transparency in subscription terms, and robust cancellation options will be key in preventing these issues.
The Economics of Agentic Commerce: Why Companies Are Embracing This New Model
OpenAI, Stripe, Etsy, and Shopify's Strategic Investments
The rise of agentic commerce opens up new revenue streams for companies like OpenAI, Stripe, Etsy, and Shopify. OpenAI will earn transaction fees for purchases processed through ChatGPT, and merchants will benefit from access to a vast audience without building their own conversational shopping experiences. Stripe, as the payment processor for ACP, is strategically positioned to become the go-to infrastructure for agentic transactions.
For Etsy and Shopify, Instant Checkout allows them to tap into the growing demand for AI-driven shopping without significant upfront investments. As more consumers embrace agentic commerce, these companies will capture a larger share of the e-commerce market.
Building Trust in Agentic Commerce: Best Practices and Principles for Businesses
Transparency and Education
Businesses must be transparent about when users are interacting with AI and how their data will be used. Clear opt-in processes, privacy policies, and data handling procedures will foster trust and improve user confidence.
User Control and Confirmation
Even if AI becomes autonomous, users should have the ability to review orders, set spending limits, and easily cancel or modify purchases. Ensuring that users can maintain control over their transactions will increase trust and make consumers feel more comfortable with agentic commerce.
Security and Privacy Safeguards
To protect sensitive financial information, businesses must adopt end-to-end encryption, tokenization, and secure authentication mechanisms. Real-time fraud detection and transparent error handling will also be necessary to ensure that consumers feel secure when using AI for transactions.
The Future of Autonomous Shopping: Will AI Handle All Our Purchases?
Moving Towards Full Autonomy
Currently, consumers must confirm each transaction, but the end goal of agentic commerce is full autonomy. Imagine an AI that not only handles your everyday purchases but also manages your subscriptions and replenishes supplies automatically. This level of autonomy could redefine e-commerce, but it raises questions about consumer control and liability.
Liability, Autonomy, and Ethical Considerations
If an AI agent makes an erroneous purchase or is compromised, who will be held responsible? As agentic commerce scales, regulatory frameworks will need to evolve to address these concerns, ensuring that consumers are protected while maintaining the benefits of automation.
Conclusion: Embracing the Future of Autonomous Shopping with Macaron
Instant Checkout in ChatGPT represents a major step forward in the evolution of e-commerce. By allowing AI to handle transactions, we are entering the era of agentic commerce, where AI agents can autonomously make purchases on behalf of users. However, the success of this model depends on building trust through transparency, robust security, and user control.
The next decade promises significant advancements in AI-powered shopping, and Macaron is poised to lead the way by empowering users with innovative, secure, and efficient tools for managing their purchases.
Top comments (0)