Agentic Commerce:It's changing how your customers shop. Are you ready?

Agentic commerce lets customers shop through a completely different experience: instead of opening a browser and navigating to a storefront, they open ChatGPT, Claude or Gemini and describe what they want. The agent searches, compares, and can complete the purchase on their behalf, often knowing more about the customer's preferences and context than any single retailer ever could.

As PJ, our Co-Founder & COO puts it: the agent will know as much as the customer has chosen to share with it, and it's able to go across every brand and retailer at once. Ask for a men's shirt, size L, in blue, for a summer wedding, and the agent doesn't just filter a catalogue, it interprets the occasion, cross-references it against dozens of retailers, and comes back with a shortlist.

For the customer, that's a genuinely better experience. For the retailer, it raises a question that goes beyond marketing: how do you prepare for customers shopping through a channel you don't control?

The reality check: this is early, not finished

It's worth being precise about where things stand, because the hype cycle and the reality are running at different speeds. Forrester's assessment in mid-2026 is that adoption remains limited, most of what's live today is conversational discovery, not autonomous purchasing. Very few consumers currently let an agent complete a transaction without direct human oversight; answer engines are reshaping how products get found, but they're mostly still referring traffic onward rather than closing the loop on checkout.

That said, the direction of travel is not ambiguous. Shopify reported AI-driven traffic to its stores up roughly 8x year-over-year in Q1 2026, with orders originating from AI search up nearly 13x. The infrastructure to support full agentic transactions is being built quickly and by serious players: the Universal Commerce Protocol (UCP), co-developed by Shopify and Google, standardises cart creation, checkout and payment across AI platforms so merchants don't need a custom integration for every new agent that appears. Shopify's own Catalog system now structures and syndicates product data for AI consumption in real time, and Shopify reports that well-structured data drives roughly 2x the conversion rate in AI chat interactions compared to unstructured listings.

The takeaway for enterprise merchants: the discovery half of agentic commerce is already reshaping traffic and conversion today. The checkout half is still forming. Retailers who wait for it to fully mature before acting will be optimising their data and governance under time pressure, in exactly the kind of "big bang" scramble that enterprise transformation is usually trying to avoid.

Why this is unsettling for enterprise retailers and where the real risk sits

The anxiety here isn't irrational, and it's worth naming precisely rather than waving it away with enthusiasm about "the future of shopping." There are three distinct risk categories enterprise leadership teams need to separate:

Loss of channel control and brand fidelity. When a customer's entire interaction happens inside a third party's conversational interface, the retailer no longer controls the presentation layer, the thing enterprise brands have spent years and significant budget getting right. Inconsistent pricing signals, an agent surfacing the wrong variant, or merchandising decisions made by an algorithm outside the brand's control can quietly erode the trust a storefront was built to protect.

Financial exposure. An unconstrained agent optimises for whatever it's told to optimise for, frequently price or perceived relevance, not necessarily margin. Without explicit guardrails, agentic channels can drive volume at the expense of profitability, or expose promotional and discount logic to comparison in ways a retailer never intended to make so visible.

Compliance, security and accountability gaps. Every regulatory and accessibility obligation a retailer already carries still applies when an agent is the one initiating the interaction. Overly broad API access or credentials granted to accelerate integration can become the security gap an agentic protocol wasn't designed to guard against. And when something goes wrong, a wrong order, a pricing error, a fraudulent transaction, questions of merchant-of-record status and liability are still being worked out across the industry, not settled.

None of this is a reason to sit it out. It's a reason to treat "getting ready for agentic commerce" as a governance and architecture question, not a marketing one.

What preparing looks like

Based on where the platform and the risk landscape both stand right now, enterprise merchants should be doing five things in parallel:

Treat product data as infrastructure. Structured, complete, accurate product data, clean titles, full attributes, real-time pricing and inventory, is the single highest-leverage thing a retailer controls in an agentic channel. This is effectively the new SEO discipline (some are calling it generative engine optimisation, or GEO), and it's foundational regardless of how quickly checkout-level autonomy matures.

Participate in the emerging standards rather than waiting for them to settle. UCP and similar protocols are being shaped now. Retailers engaging early get a say in how their brand is represented in these channels; retailers who wait will be adopting whatever defaults get set without them.

Build explicit guardrails before granting any agent autonomy. Margin floors, discount limits, scoped API permissions, and audit trails aren't optional extras, they're the difference between agentic commerce as a controlled channel and agentic commerce as an unmanaged liability.

Fix the organisational silos. Forrester's clearest finding is that the biggest blocker to enterprise readiness is its internal coordination between digital, IT, legal, and marketing teams who haven't yet had to make joint decisions about a channel none of them fully own. That's a change-management problem before it's a technical one.

Start narrow and internal. The lowest-risk way to build organisational muscle is to experiment with conversational AI on channels the retailer already controls, before extending trust to third-party agents operating at scale.

The Familiar Choice, Again

This is precisely the pattern enterprise transformation keeps repeating: the brands that treat platform-level shifts as a governance and architecture exercise and start early, end up ahead of the brands that wait for certainty and then scramble. Agentic commerce is the next version of that same choice. The retailers asking "how do we prepare for a channel we don't control" now are the ones who'll have working answers by the time it matters.