Commerce Agent
A Commerce Agent is the mode of an AI agent built for shopping questions: product recommendations, sizing, and order lookups, as distinct from general support questions in the same conversation. It resolves, recommends, or hands off to a support agent depending on what the shopper needs.
Ecommerce support conversations split into two very different jobs: "where's my order" and "what should I buy." A single agent tuned for one usually stumbles on the other, which is why many AI agent platforms split shopping questions into their own mode.
What Is a Commerce Agent?
A Commerce Agent is the mode of an AI agent dedicated to shopping-related questions, product recommendations, sizing and fit, availability, and order lookups, running alongside (not instead of) a general support agent in the same conversation surface. Also called commerce mode or a shopping assistant, it activates when a shopper is browsing or asking what to buy, and can hand off to the support agent the moment the conversation shifts to a service issue like a refund or a WISMO inquiry.
This is a narrower concept than "ecommerce support" broadly. A brand can have ecommerce support enabled site-wide while only a fraction of conversations actually get routed to the dedicated Commerce Agent; the rest are handled by the general Customer Agent even though they touch on an order or product.
Why Commerce Agent Matters
Splitting commerce out from general support gives each mode a clearer job:
- Product recommendations get surfaced by an agent built to reason about catalog and fit, not one tuned for troubleshooting
- Shopping conversations get measured on their own terms (resolutions, recommendations served, escalations) instead of blending into general support numbers
- Genuine support issues that surface mid-shopping (a stuck order, a billing question) get handed to the agent equipped to resolve them, instead of stalling with an agent that only knows the product catalog
How a Commerce Agent Works
A Commerce Agent conversation ends in one of four outcomes:
- Resolved - the shopper's question is answered directly: a sizing question, a "does this come in blue" question, an order status check
- Served recommendations - the agent surfaces relevant products, independent of whether the shopper ultimately buys
- Escalated to support - the conversation shifts to a service issue and hands off to the general support agent
- Unresolved - the agent can't answer and the shopper leaves without a clear outcome
These outcomes are tracked separately from each other. A conversation that serves a strong recommendation isn't automatically "resolved," and an escalation to support only marks that the handoff happened, not whether the follow-up conversation itself succeeded.
Commerce Agent vs Support Agent
| Commerce Agent | Support Agent | |
|---|---|---|
| Typical questions | Product fit, sizing, recommendations, "what should I buy" | Refunds, WISMO, account issues, troubleshooting |
| Primary outcome | Recommendation served or purchase-path question resolved | Issue resolved |
| Handoff direction | Escalates to Support Agent for service issues | Can loop back to Commerce Agent for shopping questions |
Frequently Asked Questions
Does a Commerce Agent replace the Support Agent?
No. They run side by side. The Commerce Agent handles shopping questions; the Support Agent handles service issues, and a single conversation can pass between the two as the shopper's needs change.
Can one conversation involve both agents?
Yes. A shopper might start by asking for a product recommendation, then ask about a delayed order from a past purchase, at which point the conversation escalates from the Commerce Agent to the Support Agent.
What happens if the Commerce Agent can't answer a shopping question?
The conversation is marked unresolved rather than forced into a recommendation or resolution it can't support, keeping the outcome data accurate for measuring where the agent needs improvement.