Getting real about agentic commerce
Forrester says autonomous shopping is coming fast — but no one’s ready for it yet, as brands race to define what agentic commerce actually means in practice
Agentic commerce — where AI agents act, decide, and even purchase on behalf of users — is emerging fast, but the reality is still experimental and uneven. Forrester analysts unpack what “agentic” really means in commerce, where it’s happening today, and what’s coming next as answer engines and merchants start to collaborate more closely.
Key takeaways
- Loose definitions, high expectations: Agentic commerce refers to AI-driven systems that can plan, adapt, and act autonomously in buying or selling contexts — though few examples today live up to that promise.
- Owned vs. non-owned experiences: Owned experiences happen on brand-controlled sites and apps, where AI tools guide product discovery but rarely enable checkout. Non-owned experiences occur in answer engines like ChatGPT, Gemini, or Perplexity, which are beginning to support direct purchases.
- Assistive, not autonomous — yet: Most so-called agentic commerce tools are still “assistive” rather than fully agentic, helping shoppers find products but not completing transactions independently.
- Fragmented checkout reality: Current “buy in chat” experiences are clunky — disconnected from retailer accounts, loyalty programs, or inventory systems — reflecting an early stage of integration between agents and merchants.
- Rising B2B and agent-to-agent interactions: In early tests, business agents are beginning to negotiate prices or terms autonomously, signaling the coming phase of automated commercial negotiation.
- Caution before scaling: Forrester advises vendors to test carefully and gather real-world feedback before rolling out agentic experiences broadly — and tells retailers not to panic; no one has this fully figured out yet.
Get the full story at Forrester