The e-commerce industry has undergone a quiet transformation in recent years. Platforms that once prioritized visual flexibility and JSON-based configuration are now rediscovering the value of clean, readable code—both for humans and machines. Shopify, one of the sector's giants, has just taken a bold step that perfectly illustrates this trend: returning to the basics of clean HTML, driven by the rise of artificial intelligence agents.
From its early days, Shopify offered merchants the ability to customize their online stores using themes based on Liquid, its own templating language. Over time, the company added layers of abstraction to give more visual control, converting template instructions into JSON files. This allowed users to tweak every button and block, but at the cost of growing complexity. As Shopify's product director Ben Sehl noted, 'when templates became JSON, they stopped being a great developer surface; they became an auto-saved output.'
Now, the rise of AI has forced a rethink of that architecture. AI agents, like Shopify Sidekick, need to understand the code to help merchants customize their pages. And it turns out that what works for a machine also works for a human: simple code, with explicit syntax and constructive feedback. That is why Shopify has redesigned its Horizon base theme to be mainly HTML, interspersed with easy-to-read Liquid blocks. The new theme reduces lines of code by 93% compared to its predecessor.
This move is not merely nostalgic. Major tech corporations are adopting a similar approach because they have verified that large language models (LLMs) process readable code—like HTML or SQL—better than dense, symbolic formats. Shopify has developed a new parser and an extended syntax for Liquid that supports boolean expressions, infix operators, and array and object literals, all without losing the clarity of HTML.
For a company like Q2BSTUDIO, specialized in custom software development, the lesson from Shopify is clear: simplicity in code not only eases human maintenance but also optimizes interaction with AI systems. When we build solutions for our clients, we prioritize readability and modularity because we know that AI agents and development teams increasingly coexist in the same ecosystem.
The declarative JSON-based architecture forced developers to anticipate how many buttons or sections they would need before starting the design. With plain HTML, if a merchant discovers they need an extra button, they simply wrap it in a div. This flexibility returns control to the creator, without opaque intermediaries. Additionally, Shopify has introduced composable blocks and a standard action system that allow updating specific parts of the page without reloading the entire site—a functionality similar to React's props but without the need for a virtual DOM.
Artificial intelligence is not only changing how merchants design their stores, but also how underlying software is written. Shopify Sidekick, the platform's AI assistant, has already facilitated over 25 million theme edits this year, and 20% of merchants use it. For the assistant to work accurately, the code must be predictable and explicit. Hence, Shopify created 20 new validation rules to ensure generated themes comply with policies on contracts, structure, nesting, and file size limits.
This return to clean code does not mean giving up sophistication. On the contrary, by simplifying the base, greater scalability is achieved. Merchants can now integrate cloud AWS/Azure to deploy their stores with high availability, and development teams can leverage advanced cybersecurity without the code turning into a mess. At Q2BSTUDIO, we combine these capabilities with BI/Power BI to provide businesses with real-time dashboards on their online store performance, all built on a clear, maintainable code base.
Shopify's case shows that, often, the best answer to technological complexity is simplicity. AI agents, far from encouraging opaque, auto-generated code, are driving a new era of transparency. Reading source code is becoming a valuable practice again, both for the experienced developer and for the novice merchant who just wants to change the background color of their page. Next time your team faces a bloated JSON architecture, perhaps it is time to ask: what if we go back to HTML? After all, AI is teaching us that simple is smart.
