From Drag-and-Drop Dreams to AI-Written Code
AI-assisted deterministic automation is the emerging model where large language models translate human intent into conventional code that executes predictably under enterprise governance, replacing visual low-code builders with AI agents that author clean, auditable workflows at scale.
The era of low-code platform replacement has begun, and it is not because visual builders failed—it is because AI made them unnecessary. When AI agents can write real code on demand, the value shifts from drag-and-drop assembly to reliable execution, observability, and control. Tines is explicit about this shift: after eight years of building a no-code automation platform, it launched 3B, a new system where AI authors enterprise workflows but execution happens as conventional code. Shopify is doing something similar on the storefront side, redesigning its base theme so both humans and AI agents can read and modify plain HTML and Liquid templates instead of tangled JSON. The signal is clear: visual abstraction is out; intelligible code plus AI workflow automation is in.
Shopify: AI Pushed the Storefront Back to Source Code
Shopify’s storefront rearchitecture exposes why visual and serialized configuration hit a wall once AI arrived. The company is rolling out a new, yet-unnamed base theme as the likely successor to its JSON-heavy Horizon theme. Instead of opaque configuration blobs, merchants will see plain-text HTML templates interpolated with Liquid, which are easy for anyone with basic HTML knowledge to follow. The result is dramatic: the new theme has 93% fewer lines of code than Horizon.
This is not nostalgia for hand-coding; it is an optimization for AI agents. Shopify’s popular AI assistant, Sidekick, already lets merchants declare what they want—such as changing a background to blue—while the agent edits the theme accordingly. This year, 20% of merchants used Sidekick to make 25 million theme edits. As Shopify’s product director notes, leading models already understand HTML, and agents want what humans want: readable code and explicit contracts. By returning to clean templates and promising a “forever API” for existing Liquid themes, plus future Tailwind CSS support, Shopify is betting that AI code generation beats low-code abstractions for long-term readability and maintainability.
Tines 3B: AI Workflow Automation Without Runtime Guesswork
If Shopify shows AI reshaping front-end customization, Tines shows how AI will replace visual builders in enterprise automation. After eight years of growing a no-code platform, including a $125 million Series C at a $1.125 billion valuation and 1.5 billion automated actions weekly, Tines chose to rebuild from scratch. Its new 3B platform lets employees describe an application, workflow, or agent in natural language; AI then writes the code and deploys it. As CEO Eoin Hinchy puts it, no-code and low-code were “the right solution for a particular moment in time,” but there is “a sell-by date” on the category.
The important part is how 3B executes. The builder starts from a prompt, asks follow-up questions, and generates a workflow broken into steps. Each step becomes code that runs in its own ephemeral Docker container, destroyed and recreated on every execution. “We’re using AI in the creation of these workflows, but that’s it,” Hinchy says; the resulting applications run as deterministic workflows instead of asking a model to reason at every step. That is deterministic code generation in practice: AI is the authoring tool, not the runtime brain.
Governed Enterprise AI Agents, Not Free-Range Bots
The deeper story is governance. Early AI hype centered on autonomous agents wandering freely across systems. Enterprises responded with a hard no. What Tines and Shopify are building instead is a pattern where people state intent in natural language, AI translates that intent into code, and IT retains control over what runs, where, and with which permissions.
In 3B, IT teams define which systems and data an employee can use before they start building. The platform shows which workflows are running, failing, or exposed to the public internet, and it can propose fixes that admins choose to accept. This positions 3B squarely in the arena of enterprise AI agents, competing with other agent platforms while keeping credentials, integrations, and runtime under central oversight. Shopify’s Sidekick follows the same spirit on the storefront side: merchants describe changes and the AI updates readable templates inside a theme architecture designed for clear contracts and feedback. Together, these moves show how new AI platforms combine human-friendly intent capture with deterministic code execution under IT governance.
The End of Visual Builders and the Rise of Deterministic Automation
Low-code visual builders solved a real problem: they helped non-developers assemble workflows when models could not write code. That condition no longer holds. As Hinchy notes, building is now commoditized; the challenge is managing the code, integrations, and runtime. It is telling that Tines is not killing its original product—there is no migration deadline, and the company says it will continue investing in the visual builder. But leadership is clear that prompting, not dragging blocks, is the future interface.
All these improvements should simplify life for ordinary users: Shopify’s merchants assembling storefronts at home, and finance or HR staff who can describe the automation they need but have no idea how to deploy or operate it. Low-code platforms will not disappear overnight, but their role will shrink toward legacy surfaces and niche cases. The center of gravity is moving to AI workflow automation driven by deterministic code generation: intent in natural language, code as the output, and agents executing within clear, auditable boundaries. That is the real low-code platform replacement—and it looks a lot more like reading source code again.
