Agentic Commerce: Reality Check, Why AI Shopping Still Can’t Replace Checkout.

Photo of author

By JUNED

AI agents already help people search products and plan trips, but they still don’t “close the deal” at scale.
Bernstein analysts say the real disruption in agentic commerce will take time because consumer habits, checkout friction, and trust still anchor transactions to retailers and travel sites.

What’s happening right now

Bernstein’s message sounds simple: AI has entered daily life for search and information, but shopping behavior changes slowly.
Even with new shopping tools, partnerships, and early agent-led checkout experiments from ChatGPT and Google, most people still browse with AI and then buy the old fashioned way directly on merchant apps and websites.

One line from the data hits hardest: AI-driven referrals to eCommerce and online travel sites remain small typically below 1% of total web traffic.
Independent benchmarking points in the same direction, with Conductor data summarized by Search Engine Land putting AI referrals at about 1% of all web traffic across major industries.searchengineland

Bernstein also notes that only a small share of ChatGPT usage ties directly to purchasing, and it remains well below the share of commercial queries seen on traditional search engines.
That gap explains why “agentic commerce” feels everywhere on Twitter but still looks like a side dish in actual checkout metricsAgentic Commerce Reality Check, Why AI Shopping Still Can’t Replace Checkout.

Why this news matters (for businesses, not just AI fans)

“Disruption” usually means someone loses sleep.
Right now, Bernstein frames this phase as broadly positive for listed internet companies because fears of instant disintermediation haven’t materialized.

That matters for three groups:

  • Retailers and travel platforms: They still own the customer relationship because users keep visiting them directly.
  • AI platforms: They chase scale and engagement first, and they add commerce features to push users from “compare” to “buy” inside the AI interface.
  • Investors and operators: They should treat agentic commerce as “early experimentation,” not a finished replacement for search ads, affiliate links, and marketplace discovery.

 

Here’s the human reality: people don’t change buying habits just because a new button appears.
They change when the new flow beats the old flow on speed, confidence, and customer support especially when money leaves the wallet.

Agentic commerce disruption: Why checkout stays “sticky”

Bernstein says AI already improves search and discovery, but it still has limited impact on checkout and payment.
That checkout gap isn’t a small technical detail; it’s the whole business model.

When a user buys, they ask “Can I trust this?” more than “Can I find this?”
Checkout includes returns, delivery promises, fraud protection, warranties, and customer service areas where consumers already trust familiar retailers.

Agent-led checkout needs deep merchant integrations and consistent product data to avoid the nightmare scenario: the agent buys the “right” product… in the wrong size… from the wrong seller… with the wrong delivery date.
If that happens once, the user won’t say “nice try.” They will say “never again.”

Agentic commerce disruption: Why travel moves first

Bernstein expects higher-consideration purchases—like travel or big-ticket discretionary goods to see the earliest gains from AI assistance.
That prediction makes sense because travel planning already looks like a research project: dates, budgets, reviews, maps, timing, and “is this place actually real?” checks.

AI can compress that messy research into a cleaner shortlist, which feels valuable even if the final booking still happens on a travel site.
So the near-term win may look boring: better discovery and planning, not full automation of buying.

What changes next (and what to watch in 2026)

True agentic commerce, in Bernstein’s view, remains “some distance away” because adoption depends on delivering a clearly better experience than current apps and websites.
This creates a simple scoreboard: consumers decide with their actions whether AI agents become central to commerce or stay complementary.

Three shifts can move the needle:

  • In-chat purchases that feel safe: OpenAI has already described “Instant Checkout” in ChatGPT, where users can complete purchases without leaving the chat, while merchants handle orders and fulfillment in their existing systems.openai
  • Better product truth: Agents need accurate inventory, shipping, and pricing in real time, or they will hallucinate the most expensive part of the internet: delivery promises.
  • Clear incentives for merchants: Early partnerships look “mutually beneficial” right now, not zero-sum, according to Bernstein.

The bigger implication: brands must prepare for AI-assisted discovery even before they see massive “AI referral traffic.”searchengineland
In plain terms, an agent can influence the purchase long before a click shows up in analytics.

FAQ 

1) What is “agentic commerce disruption”?
It describes a shift where AI agents help people discover products and, eventually, complete purchases on their behalf, potentially changing where transactions happen.

2) Are AI agents already changing shopping?
Bernstein says AI agents play a bigger role in search and planning, but they haven’t meaningfully disrupted where people actually buy yet.

3) How big is AI-driven traffic to online stores today?
Bernstein says AI-driven referrals to eCommerce and travel sites typically stay below 1% of total web traffic.

4) Does any outside data support that “~1%” view?
A Conductor benchmark summarized by Search Engine Land reported AI referrals at about 1.08% of all web traffic across major industries.searchengineland

5) Why doesn’t discovery automatically lead to checkout?
Bernstein says search/discovery improves faster than checkout/payment because consumers change buying habits slowly and still prefer familiar platforms.

6) Which categories could benefit first?
Bernstein expects higher-consideration purchases like travel and big-ticket discretionary goods to gain earlier from AI help.

7) Are AI platforms partnering with retailers or trying to replace them?
Bernstein says partnerships look mutually beneficial so far, not zero-sum, and rapid disintermediation hasn’t happened yet.

8) What would make agentic commerce go mainstream?
Bernstein says adoption hinges on a clearly better experience than existing apps and websites, and consumers will decide through real usage.

Conclusion + direct answer

Agentic commerce disruption has started in discovery, not in checkout, so businesses should plan for gradual change not a sudden collapse of retailer and travel platforms.
Direct answer: Bernstein says AI agents will influence shopping more over time, but transactions won’t shift fast because AI referrals stay small (typically under 1%), and consumers still trust existing checkout flows .searchengineland

Leave a Comment