Shopify's biggest platform shift
is live. Here's what it
actually means.

Let's start by cutting through the language. "Agentic commerce" is not a product category or a subscription tier. It is a description of what happens when the actor doing the shopping is not a human.

In the old model: a shopper opens a browser, types a query, clicks several results, compares products across a few sites, reads reviews, maybe opens three tabs, and eventually adds something to a cart. Every one of those steps is a human making a decision. Your store's job is to be present at the right moments and persuasive enough at each one to close the transaction.

In the agentic model: the shopper tells an AI assistant what they want. The assistant queries data sources — including Shopify's commerce graph — surfaces candidates, evaluates fit, and can complete a purchase end-to-end without the human visiting a storefront at all. The browsing, comparison, and checkout happen inside the conversation. The human delegates the middle steps to the agent.

Shopify shipped infrastructure for exactly this. It's called Agentic Storefronts, and it activated by default for all Shopify merchants in March 2026. This post covers what it actually does, what's related, what's still early, and what you should be doing about it right now.

The infrastructure is live.
Most merchants haven't
thought about it yet.

Shopify Agentic Storefronts does one primary thing: it makes your product catalog — inventory, pricing, descriptions, availability — queryable and purchasable through AI platforms that are not Shopify. When a user asks ChatGPT to find them a gift for a hiker under $75 with Friday delivery, ChatGPT can surface your products because Shopify is feeding your catalog into the conversation layer. The shopper never visits your Shopify storefront. The purchase goes through Shopify's checkout infrastructure regardless.

The technical layer underneath this is the Shopify Catalog — a global product data index that uses AI to categorize, enrich, and standardize product data. It continuously updates pricing and inventory across all connected AI channels. The standard enabling cross-platform transactions is the Universal Commerce Protocol (UCP), which Shopify co-developed with Google. OpenAI and Stripe have their own standard — the Agentic Commerce Protocol (ACP) — for checkout inside ChatGPT specifically.

The platforms where Shopify merchants are now discoverable through Agentic Storefronts: ChatGPT (880 million monthly users), Microsoft Copilot, Google AI Mode, Perplexity, and the Shopify Shop app AI assistant. No separate integration per platform. One setup in your admin, and the catalog is live across the network.

The AOV premium on agentic orders is interesting and worth noting: agent-assisted shoppers tend to have a clear intent and a delegated buying task, which appears to drive less price comparison and higher confidence in completing the purchase. Whether your specific products are winning in that system depends almost entirely on catalog data quality — not ad spend or SEO domain authority. This is early data and may normalize as the channel matures. But it's directionally consistent with the hypothesis that agentic buyers are more decisive than browsers.

Sidekick is the merchant AI.
App Extensions make
third-party apps queryable.

Sidekick and Agentic Storefronts are related, but they're not the same product. Agentic Storefronts faces outward — it puts your products in front of external AI platforms. Sidekick faces inward — it's the AI coworker for the merchant operating their store.

In the Winter '26 Edition, Shopify evolved Sidekick well past its original chatbot framing. It now runs background analysis (Sidekick Pulse), automates workflows, anticipates common operational questions, and can execute complex tasks like theme edits through conversation. The framing Shopify is using internally is closer to "co-founder" than "assistant" — the distinction being that Sidekick has store context and acts on it, rather than waiting to be prompted for each discrete task.

The development that matters most for app builders: Sidekick App Extensions. Currently in developer preview, these allow third-party apps to expose their data and functionality to Sidekick so merchants can query and act on it through conversation.

Sidekick App Extensions are in developer preview — meaning the architecture and API patterns are published, active development is happening, and the feature is not yet generally available for production apps. The timeline to general availability hasn't been officially confirmed. What has been confirmed is that this is the direction Shopify is building in, and early participants in the developer preview will be ahead of the pattern when it ships at scale.

SimGym: testing your storefront
against AI shoppers before
real ones see it.

SimGym is the third piece from the Winter '26 Edition worth understanding. It is in Research Preview — meaning it is live for interested merchants to use, but it is explicitly early-stage. Shopify is not treating this as production-ready infrastructure yet.

What it does: SimGym trains AI models on real commerce sessions across Shopify's platform to simulate customer personas — approximations of how real shoppers behave and respond to a storefront. Merchants can run simulated purchase sessions against planned changes before launching them. A major redesign, a new product layout, a checkout flow change — SimGym gives you a way to catch obvious conversion problems before real shoppers hit them.

The value case: A/B testing catches friction points, but it requires actual traffic and real time. SimGym is directional signal that arrives faster and without the cost of running a live experiment to failure. You're not running it instead of user testing or A/B testing — you're running it before, to eliminate the easy stuff.

Calibrated expectations matter here. Simulated shoppers are approximations built from aggregate behavior patterns. They will not catch every edge case a real human encounters, and they have no lived experience with a specific brand's context. SimGym is not a replacement for watching a real customer navigate your store. It is a complement — a faster, cheaper first pass before you invest in more precise testing. Treat the outputs accordingly.

What to actually do —
in the order it
matters.

Agentic Storefronts is already on for your store. Your products are in the feed. The question is how visible and purchasable they are when an AI agent is doing the evaluation. Most merchants who check their AI visibility for the first time find they're largely invisible for their highest-intent queries — not because the feed is broken, but because the product data quality doesn't give AI systems enough signal to confidently recommend them. This is the same data-quality problem covered in our 10-point product page CRO audit — and it turns out that agentic readiness and conversion optimization require the same fixes.

Here's what changes that, in priority order:

App builders have a window.
It is closing as
more people read this.

For app developers in the Shopify ecosystem, Sidekick App Extensions are a new distribution surface — and the pattern for who wins it is already visible from prior Shopify platform shifts.

When a merchant asks Sidekick "how do I improve my product page conversion?" — the apps that have built Sidekick extensions will be part of the answer. The apps that haven't built the integration will not. That's not speculation about future platform behavior; that's how every conversational AI recommendation layer works. The model surfaces what it can query. It can't surface what it can't see.

The apps that built native Shopify theme integrations early got disproportionate installs when theme stores became a major discovery channel. The apps that built Shopify Flow triggers early became part of merchant automation workflows in ways their competitors couldn't access. Sidekick extensions follow the same dynamic — and it mirrors what I documented in the Partner Program retrospective: the apps that move early on new Shopify surfaces consistently capture install velocity before the pattern becomes obvious to everyone.

What's different this time: the gap between "developer preview" and "generally available" is a genuine window for preparation, not just feature gating. The architecture is published at shopify.dev. You can build the integration now, get merchant feedback in the preview environment, and ship a polished extension the day it goes GA — rather than scrambling after launch when every competitor is doing the same thing at the same time.

The categories where Sidekick extensions will be highest-value: analytics and reporting apps (data that merchants want to query conversationally), loyalty and retention tools (segment queries, churn analysis), inventory management (SKU performance, reorder alerts), and any app where the primary user interaction is looking at a dashboard to answer a recurring question. If your app has a dashboard, a Sidekick extension that makes that dashboard queryable is a near-term development priority worth scoping. Distribution strategy on these extensions will follow the same patterns covered in the App Store SEO ranking guide — early completeness signals compound.

What we don't know yet —
and why it matters
to say so.

There's a version of this post that reads like a Shopify press release with consulting copy grafted onto it. I want to avoid that. The direction here is clear. The specific timeline and impact curves are not.

My read: the direction is clear and Shopify is building the right infrastructure. The merchants and app builders who treat this as a "future problem" are making a choice — not avoiding a hype cycle. There's a real window here, and the actions in Section 04 and 05 are not contingent on believing the most optimistic projections.

But the timeline for agentic commerce driving meaningful volume impact on a typical DTC merchant's revenue is closer to 18–36 months than next quarter. I'd be skeptical of anyone telling you to pause everything else and rebuild your operations around this channel tomorrow. What you should do tomorrow is the product data audit. What you should not do is panic-pivot.

The things that make a product discoverable by AI agents are the same things that make it discoverable by humans using text search: clear language, specific claims, trustworthy signals. The investment is not wasted if the agentic adoption curve is slower than projected. You're building better product content either way.

Shopify built the plumbing for agentic commerce. Your products are in the network. The question — for merchants and for app builders — is whether you're building the right things on top of that infrastructure while the field is still thin, or whether you're going to be catching up when it matters.

The features are named. The direction is clear. The uncertainty is honest. Now you have what you need to make a decision.