Shopify Puts AI on Every Screen: Running a Store in the Agent Era

Something has shifted quietly but unmistakably in e-commerce over the past few months. A slice of shoppers no longer start at a search engine or a social feed. They ask an AI assistant — ChatGPT, Gemini — "what fits these requirements?" and arrive at a store through that answer. The share is still small. But the slope is steep. On June 17, 2026, Shopify's major release, "The Everywhere Edition," met that shift head-on with a bold design choice: put AI on every screen of store operations. The update bundled more than 150 changes.

A new axis of competition: getting discovered by agents

The most telling part of the release is a set of tools that let merchants see how their products appear across AI platforms. Aaron Glazer, a Shopify product director, summed up the question merchants now carry.

In plain terms: how does my product get found by an agent? E-commerce has long optimized for human shoppers — photos, copy, search rank. But once assistants like ChatGPT, Google Gemini, and Microsoft Copilot become entry points to shopping, a second kind of merchandising emerges: getting machines to understand you correctly and include you in their recommendations.

Shopify's new dashboard lets merchants track orders, sales, and conversions generated through ChatGPT, Gemini, Copilot, and its own Shop surface, store by store. A search-intelligence feature surfaces the common questions asked in AI shopping, and a knowledge base highlights the things assistants can't answer about a business — store locations, policies, bulk ordering. In short, alongside "SEO for humans," a second front of "optimization for agents" is now woven into the operator's daily routine.

150-plus updates that signal a turn toward AI on every screen

The design philosophy is clear: don't make AI a convenient feature in a separate tab; dissolve it into every corner of the admin. At the center, Sidekick can now be summoned on nearly every admin screen, by typing or by voice. It works in the background without taking over the screen, handles multiple chats at once, and even answers revenue questions from an Apple Watch.

Integration with third-party apps expanded sharply. Sidekick App Extensions let merchants direct work inside major apps — Klaviyo, Loop, Smile, Judge.me — through Sidekick itself. More than 15 partners were live at launch, and the system is open to any developer rather than invite-only. On the marketing side, Campaign Autopilot recommends, launches, and optimizes ad campaigns spanning Meta, Google, ChatGPT ads, and Microsoft's network, within guardrails the merchant sets. Sachin Malhotra, a director of merchant marketing at Shopify, described it as giving merchants a "virtual marketing agency."

An AI sales associate now operates inside Shopify Inbox, drawing on inventory, catalog, and policy to handle inquiries. Native A/B testing for themes and checkout rounds it out. Here is a snapshot of the main features and their availability.

Why measure "sales through AI" now

By the numbers, traffic from AI chat is still under 1% of all web traffic (as of 2026). Yet reporting indicates that at large U.S. retailers like Walmart and Target, AI referrals have climbed from under 1% to more than 1.5% year over year. An IBM survey found 41% of consumers already use AI assistants for product research. When a small number shows up with a rising slope, it pays to build the measurement plumbing before the wave arrives.

This is where Shopify's visibility tools — framed as bringing clarity to what is still a small channel — earn their keep. Without channel-level conversion tracking, a rise in AI-sourced shoppers can't be told apart from luck. Track it early, and you can verify with numbers, not guesses, how tidying up product data affects exposure through agents.

This "measure before you move" sequence is the lesson of the last search-engine era. Given how SEO was scrambled together after the fact, it is rational to build the habit of measurement and product-data hygiene for the AI shopping channel while inflows are still thin.

What Sidekick takes off your plate, and where it asks for a human

Under the hood, Sidekick is a function-calling agent over Shopify's internal APIs. It reads your store's real data, calls actual Shopify functions, and shows you the changes for approval before applying them. A single prompt — "create a 15% discount on the whole summer collection for two weeks and draft an announcement email to subscribers" — assembles a sequence spanning discounts, collections, and email. When it lacks information, it offers multiple-choice options so you can clarify quickly.

The screen below is an example of a "Pulse" card, where Sidekick proactively finds and flags an issue. It detects a product page missing SEO metadata and warns that, as is, the item is nearly invisible to both shoppers and search engines.

What must not be missed: the center of gravity sits not on "full automation" but on "semi-automation with a human in the loop." Changes are shown before they apply, and Campaign Autopilot only operates inside merchant-set guardrails. The system is built so operators can draw the line between what they delegate to AI and what they verify with their own eyes. Designing that line is the crux of using AI hard without causing accidents in production.

How to read the numbers and apply them to your store

Three practical points stand out. First, don't dismiss AI-sourced traffic as "too small to bother with." A channel growing 1.5x year over year changes the precision of your moves six months out simply by having the measurement foundation in place today. Shaping product names, descriptions, and attribute data so machines parse them correctly is an investment that doesn't conflict with human-facing SEO.

Second, treat "early access" and "available" features differently. For maturing capabilities like Campaign Autopilot, rather than handing over the whole budget at once, start with a capped budget and clear guardrails, observe the AI's decision quirks, and widen its remit gradually. Keeping the ability to audit what the AI did after the fact matters as much as it does with an outside agency.

Third, frame assistants like Sidekick as "pre-sorting judgment" rather than "replacing work." Let AI pre-sort routine tasks — discount setup, email drafts — while humans concentrate on final approval and exceptions. Whether you can redirect the freed-up time toward product development and customer relationships, the work machines struggle to replace, decides outcomes more than the choice of tool.

Key takeaways

Shopify's "Everywhere Edition" is the major June 2026 update that spread AI into every corner of the admin and made agent-driven purchases measurable. AI chat traffic is still under 1% of the total, but its climb to 1.5%+ at Walmart and Target, and the 41% of consumers using AI for product research (IBM survey), show the slope of change. The essentials: build measurement and product-data habits while inflows are thin; keep guardrails and human approval on the least mature, early-access automation; and use AI to pre-sort judgment so human time shifts to work that machines can't easily replace. "How do I get discovered by agents?" is becoming, for store operators, a daily question on par with search rank.