Agent-ready structure
Themes, templates and content models that an agent can read and change predictably, without guessing how a page is put together.
AI
AI agents are starting to read and change Shopify themes, content and operations. For enterprise brands, especially those running several stores, the question isn't whether agents can make changes. It's whether you can control and trust those changes.
Point of view
AI could make good structure matter more, not less. From my writing on Shopify's {% block %} preview
Many multi-store brands share one theme across every store and keep each store's layout and content separate, so a release never overwrites store-specific work. If agents start editing page layouts directly in code, that line can blur. A change meant for one store could end up in the shared theme, or clash with the next release. It also becomes harder to tell who, or what, changed something, and why.
Focus areas
Themes, templates and content models that an agent can read and change predictably, without guessing how a page is put together.
Clear boundaries between shared code and store content, review gates, and an audit trail for changes made by people and by agents.
Bringing coding agents into team workflows with conventions, documentation and checks they can follow, so output is consistent and reviewable.
Questions
This website is an example on a small scale. It includes a self-documenting scaffold that an AI agent can follow to make safe, consistent changes.
I write about this regularly on LinkedIn, and I'm always interested in how other teams are approaching it.