News
AI SHIFT:

Shopify Reworks Theme Code For AI Agents And Readability

Newsroom brief

Shopify is preparing a cleaner storefront theme that shifts from JSON-heavy configuration toward mostly HTML and Liquid, as AI-assisted theme editing makes readable code and explicit contracts more important.

Verified against source materialEdited by SendTech Times AI & Enterprise Desk
Shopify Reworks Theme Code For AI Agents And Readability
Image source: The Register

A storefront theme rewrite at Shopify is turning AI adoption into a code-architecture constraint.

The Register's account describes a new, still-unnamed base theme that moves the commerce platform away from Horizon's JSON-heavy structure and back toward mostly HTML mixed with Liquid templates.

The operating problem is that storefront customization now has two readers: merchants and AI agents.

The Register account states that Shopify Sidekick is used by 20% of merchants to edit themes, and that those users have made 25 million edits this year.

Cleaner Theme Files Replace Dense Configuration

The new base theme is being shown in developer forums as a likely successor to Horizon, the current theme for customized store pages.

Its templates directory is designed to expose plain-text files that can be understood by reading the page structure, rather than chasing long serialized configuration blocks.

Ben Sehl, product director for storefronts, framed HTML as a better surface for models because leading systems already understand it and because it is expressive, local and token-efficient.

That design choice turns page structure back into something a developer, merchant or agent can inspect directly.

JSON gave merchants more controls inside the Online Store Editor.

The cost was a developer surface that became harder to reason about from one file, especially when an assistant needed to understand what a page contained before changing it.

Agent Editing Changes The Design Target

The redesign treats readability as infrastructure for agent-assisted commerce rather than as a cosmetic code preference.

The architecture is being organized around easy-to-read source, explicit contracts and feedback that an automated editor can act on without breaking a theme.

Liquid gains a new syntax and parser, while composable typed blocks sit beside ordinary HTML.

Expanded documentation tags carry examples and contract instructions, and a composable block tag passes values, configurations and overrides into nested page components.

Standard actions cover site events such as cart updates.

Sehl wrote that this primitive can remove thousands of lines of reactivity code from Horizon, shifting more behavior into a smaller set of declared operations.

Readable Code Becomes A Deployment Guardrail

Policy controls sit alongside the cleaner files.

The generated-theme rules cover contracts, structure, validation, complexity, nesting and file-size limits, turning machine-written storefront changes into something that can be checked against declared boundaries.

The broader lesson is that models often handle semantically rich, human-readable formats better than dense symbolic configuration for non-coding tasks.

A cited developer example found SQL easier for models than domain-specific tool languages, while Anthropic's prompting guidance favors XML for semantically rich structure.

For merchants, the result could be a storefront system that is easier to inspect before and after an AI assistant changes it.

The material evidence gap is commercial timing and validation: the available record stops short of a release date for the new theme or independent performance data for Sidekick-generated edits.

Share this article
inXf

Related articles

More
GitHub And Google Back ARD As AI Agents Search For Tools
AI

GitHub And Google Back ARD As AI Agents Search For Tools

GitHub, Google, Microsoft and other companies are backing Agentic Resource Discovery, a specification meant to help AI agents find, verify and connect to tools, skills, MCP servers and other resources without hard-coded integrations.

Oracle Adds AI-Native Builder For Fusion Agentic Applications
AI

Oracle Adds AI-Native Builder For Fusion Agentic Applications

Yahoo Tech, republishing Verdict, said Oracle introduced an AI-native builder inside AI Agent Studio for Fusion Applications. Oracle said the builder supports no-code, low-code and pro-code work, runs inside Oracle Fusion Cloud Applications, and can extend over 1,000 existing AI agents and 22 Fusion Agentic Applications.

Anthropic’s Conway Points Claude Toward Always-On AI Agents
AI

Anthropic’s Conway Points Claude Toward Always-On AI Agents

Anthropic is preparing a Claude expansion that includes Conway, Orbit, Operon, memory upgrades and multilingual voice mode. The move signals a shift from chat-based AI toward persistent assistants that can connect with external services and manage workspaces. Enterprises, developers and research teams could be affected if Claude becomes a broader agent platform.

Grep Adds LLM Agent To Monito As Online Proctoring Shifts Toward Context Review
AI

Grep Adds LLM Agent To Monito As Online Proctoring Shifts Toward Context Review

Grep said its Monito online proctoring product now uses an LLM agent to analyze context around suspected cheating events. The company cited internal tests showing more than 30 percent shorter post-exam review time and nearly 20 percent fewer false alerts. The key issue is whether agent-based proctoring can improve review efficiency while preserving human final judgment and candidate fairness.

AI Agent Rollouts Require Testing Before Live Customers
AI

AI Agent Rollouts Require Testing Before Live Customers

No Jitter reported that enterprise AI agents need guardrails, simulations, answer checks and visibility before customer-facing deployment, as vendors add tools to catch regressions and rollback failures.

OpenAI Says Cars24 Runs Million AI Conversation Minutes Monthly
AI

OpenAI Says Cars24 Runs Million AI Conversation Minutes Monthly

OpenAI said Cars24 uses its APIs, ChatGPT Enterprise and Codex across customer conversations and internal workflows, including more than a million AI conversation minutes a month. The case study did not disclose OpenAI API spend, audited conversion lift, model versions or customer-retention figures.

Creatio Adds AI Agent Governance To 10x CRM Platform
AI

Creatio Adds AI Agent Governance To 10x CRM Platform

No Jitter reported that Creatio 10x adds AI Studio, AI Twin and prebuilt CRM agents while keeping agent access under enterprise permissions and consumption guardrails. The report did not name customers, measured cost savings, consumption thresholds or independent security-audit results.

Block’s Builderbot Shows Where AI Coding Tools Hit The Enterprise Wall
AI

Block’s Builderbot Shows Where AI Coding Tools Hit The Enterprise Wall

Block says its Builderbot framework coordinates AI agents across internal repositories, Slack threads, issue trackers and continuous-integration workflows. The company says the system runs over 200,000 commands each day, merges about 1,500 pull requests each week and accounts for roughly fifteen percent of company code changes. The stronger claim is not code generation alone. Block is testing whether agentic software work can handle permissions, context, CI failures and customer-data isolation inside a large engineering organisation.

Keep Reading

More Stories

Latest
Ramp And FIS Add AI Controls For Spend And Payments SecurityFintech & Digital PaymentsJul 27, 2026Ramp And FIS Add AI Controls For Spend And Payments SecurityRamp launched controls for AI spending while FIS joined Anthropic Project Glasswing to test Mythos 5 against payment-infrastructure vulnerabilities, leaving adoption and measured security results outside the public record.UAE Statistics Programme Targets AI-Ready Official DataCapital & PolicyJul 27, 2026UAE Statistics Programme Targets AI-Ready Official DataThe UAE has approved a national programme to improve official statistics as government agencies seek higher-quality data for economic policy and artificial intelligence systems.SourTrade Malvertising Makes Browsers Assemble Windows MalwareCybersecurityJul 27, 2026SourTrade Malvertising Makes Browsers Assemble Windows MalwareThe Hacker News reported that Confiant analysed SourTrade, a malvertising campaign that uses fake trading pages and browser-side assembly to vary Windows malware files for retail trading and crypto targets.Lumen Tracks Proxy Botnets Approaching 60 Million IPsCybersecurityJul 27, 2026Lumen Tracks Proxy Botnets Approaching 60 Million IPsCyberScoop reported that Lumen Black Lotus Labs is tracking malicious residential proxy botnets approaching 60 million victim IP addresses, with takedowns failing to stop rapid rebuilds.Empery Invests $20 Million In Cardinal Data PowerCloud & Data CentersJul 27, 2026Empery Invests $20 Million In Cardinal Data PowerEmpery Digital invested $20 million in Cardinal Data Power, backing a West Texas data centre campus plan with a 750MW first phase and potential expansion beyond 5GW.AI Distillation Debate Moves From Labs To WashingtonAIJul 27, 2026AI Distillation Debate Moves From Labs To WashingtonCNBC reported that AI distillation has become a policy fight after Moonshot AI's Kimi K3 raised questions about open-weight models, proprietary model output and U.S. restrictions.Hugging Face Adds 4-Bit Nunchaku Loading To DiffusersChips & SemiconductorsJul 27, 2026Hugging Face Adds 4-Bit Nunchaku Loading To DiffusersHugging Face added Nunchaku Lite support to Diffusers, letting developers load 4-bit diffusion checkpoints with `from_pretrained()` while using Hub-delivered CUDA kernels.CFTC Warning Narrows Prediction-Market Contract FilingsFintech & Digital PaymentsJul 27, 2026CFTC Warning Narrows Prediction-Market Contract FilingsCoinDesk reported on July 26 that the CFTC warned prediction-market operators against broad template certifications, tightening the filing burden for event contracts while courts still test the regulator’s authority.B Capital Names Ex-G42 AI Chief To Doha Investment RoleAIJul 27, 2026B Capital Names Ex-G42 AI Chief To Doha Investment RoleB Capital has appointed former G42 executive Dr Andrew Jackson as General Partner and Chief AI Officer in Doha, tying the firm’s Gulf office to AI investment governance and Stargate UAE experience.Ajman AI Agent Renews Trade Licences Through AjmanOneEconomyJul 27, 2026Ajman AI Agent Renews Trade Licences Through AjmanOneMiddle East AI News reported that Ajman used agentic AI to renew a trade licence through AjmanOne, moving a government service from chatbot support toward an automated transaction flow that still depends on governed data and system integration.AMD Helios AI Racks Bring Epyc CPUs To Nvidia Compute ChallengeChips & SemiconductorsJul 26, 2026AMD Helios AI Racks Bring Epyc CPUs To Nvidia Compute ChallengeData Center Knowledge reported that AMD moved Helios into production with MI455X GPUs, Epyc processors, Pensando networking and ROCm software, while vendor performance claims still lack third-party benchmark validation.Dubai Airports Adds Smart Gate Pre-Check Before Summer PeakEconomyJul 26, 2026Dubai Airports Adds Smart Gate Pre-Check Before Summer PeakEconomy Middle East reported that Dubai Airports launched a Smart Gates Eligibility Pre-Check at DXB, letting passengers confirm eligibility through Pocket Flights or terminal QR codes before passport control.