Shopify's Winter Editions dropped a wave of updates, but most brands are still trying to work out what actually matters for their 2026 strategy. In this Shero Live, Beth Shero, CEO and co-founder of Shero Commerce, is joined by Gavin McKew, Director of Shopify Practice at Shero Commerce, to walk through three shifts reshaping commerce right now.
If you're scaling on Shopify, running a B2B or hybrid brand, or just trying to plan your roadmap around what's actually practical, not just buzz, this conversation is for you.
In This Episode:
- Why Shopify's new Universal Cart Protocol, built with Google, is the first real infrastructure for agent-to-answer-engine commerce
- What "agent ready" data actually looks like, and why salesy product copy gets deprioritized by AI agents
- How Shopify's new AI testing tool, Sim Gym, simulates shopper behavior to speed up CRO
- Which B2B features finally went native in this release, and why that matters for agencies
- Why ChatGPT's announced 4% transaction fee is a preview of new costs in agentic checkout
Watch the video for an in-depth dive
Episode highlights
Gavin McKew opens by framing 2026 as the year agentic commerce moves from buzzword to infrastructure. Last year, brands weren't sure if customers were actually buying through AI or just researching.
What changed, he explains, is that the frameworks and protocols needed to build the rails for agent-to-agent commerce are now in place, making it technically possible to demonstrate how discovery, checkout, and fulfillment can work through an answer engine rather than a traditional storefront. He's careful to note this is still early. Today's agents mostly act on behalf of the buyer, searching for a product or a price. The more interesting shift, he argues, comes later, when agents start representing sellers too.
That shift depends entirely on data quality. Gavin describes how structured, complete product data matters more than ever, since answer engines actively deprioritize salesy language in favor of product specs and problem-solving context. He recommends brands audit every piece of their tech stack- apps, PIM, ERP, anything touching product data- to confirm it can actually feed that data to answer engines and that the provider ships updates regularly.
The conversation moves to Shopify's new Universal Cart Protocol, built with Google, which lets merchants control which answer engines (ChatGPT, Gemini, Perplexity, Claude) can access their catalog data, down to the variant level. Gavin flags a real gap here: he tested the same variant selector on two different providers and got inconsistent results, meaning one could tell an answer engine a shirt comes in four colors while the other only surfaces one.
Beth shifts the discussion to Sim Gym, Shopify's new AI-assisted testing tool, which simulates shopper behavior to run CRO experiments faster and with fewer live tests. Gavin notes it's a solid first version rather than a polished product, more a signal of a new app category than a finished tool.
They close on B2B, where several previously custom-built features are now native, and on ChatGPT's newly announced 4% transaction fee, a reminder that agentic checkout isn't free for merchants.