Agentic commerce
In short
Commerce in which an AI agent carries out shopping steps on a person’s behalf, within the capabilities and permissions of the service.
Agentic commerce uses AI agents to carry out parts of a shopping process. An agent might compare products, assemble a cart, or coordinate an authorized purchase through connected services.
The degree of autonomy varies. Helping someone choose a product does not automatically include permission to buy it. Payment authorization and merchant acceptance remain separate parts of the transaction.
A practical example
Imagine a buyer requesting a replacement printer cartridge that matches a specific model and delivery deadline. An agent checks the item identity and available offers, then prepares the order details for approval.
If an offer changes before checkout, the workflow needs to handle the new total or delivery date. This illustrative process shows why a plausible recommendation is only one step toward a valid order.
Merchants still run the transaction
In documented commerce protocols, merchants retain responsibilities such as validating availability, calculating totals, processing payment, and fulfilling orders. An AI interface does not remove those requirements.
For a merchant evaluating a channel, distinguish product discovery, checkout support, and post-purchase handling. Verify which actions the integration actually supports and which require the shopper’s confirmation. A protocol implementation alone does not guarantee that an AI platform will recommend or sell the merchant’s products.