For years, we drew a fairly clear line between two ways of building software.
Low-code and no-code platforms let people build visually, with less dependence on engineering. Pro-code gave developers the flexibility and control of working directly with code.
AI is making that distinction much less meaningful.
Low-code is no longer just drag-and-drop
The definition of enterprise low-code has expanded well beyond visual interfaces.
Gartner's description of enterprise low-code application platforms now includes model-driven development, generative AI, reusable components, application deployment and monitoring, governance, APIs, and integration with external development tools.
That changes the value proposition.
The question is no longer simply: How much code do I need to write?
Increasingly, it is: How quickly can I go from an idea to software that can safely run in production?
Webflow is a good example of this shift
Webflow started with visual web development. Look at where the product is going now.
Webflow's AI site builder can turn a prompt into a multi-page site with a reusable design system. The result is not just a disposable prototype. It is a starting point that teams can continue to edit, extend, and publish in Webflow.
The more interesting development for me is AI code components.
Webflow now lets developers, designers, and marketers describe functionality such as a pricing calculator, search filter, animation, or location finder. AI generates a reusable React component that can be refined through conversation or edited as code.
Webflow also supports components created by developers in an external codebase and brought into Webflow. Different people can start from different places, but contribute to the same product.
That is the convergence in practice. A marketer may start with a prompt. A designer may work visually. A developer may work in code. The interface is different. The destination increasingly is not.
Pro-code is moving in the other direction
Professional development is becoming more intent-driven too.
Salesforce's Agentforce for Developers, for example, can generate Apex code from natural language, explain existing code, generate tests, create documentation, and provide code suggestions inside the development environment.
Developers still own the architecture, implementation, and review. But manually producing every line of code is increasingly optional.
So two things are happening at the same time:
Low-code platforms are becoming more extensible and code-capable.
Pro-code environments are becoming more conversational and generative.
They are moving toward the same middle ground.
Generation is becoming the easy part
This is where I think the conversation gets more interesting. If AI makes it easier to generate an application, component, or website, what actually differentiates a development platform?
The answer may be everything that happens after the prompt.
Can the software connect safely to company data? Can it use existing identity and permissions? Can developers extend it when requirements get more complex? Can teams understand what changed? Can it move safely through environments? Can it be monitored? Can the organization control what reaches production?
Retool's 2026 AI governance survey of 307 senior technology and security leaders found that only 8% described their internal-tool governance as strong. Twenty-two percent reported at least one production incident caused by an AI-generated internal tool in the previous 12 months. This is vendor-sponsored research, so the numbers should be read in that context. Retool's own product direction reflects the same problem. It is building governance, permissions, data access, and deployment controls underneath applications, including apps that begin in AI coding tools.
The broader direction is clear: generating software is only one part of the problem. Teams still have to operate it safely.
The old spectrum was easy to understand: No-code → Low-code → Pro-code
I think the emerging model looks more like: Intent → Generation → Refinement → Production
A marketer may enter through natural language. A designer may enter through a visual canvas. A developer may enter through an IDE. An AI agent may start from an issue or task.
Increasingly, those paths can lead to the same underlying software. That changes what platforms compete on. Code generation will not be enough either.
The differentiators move deeper into the platform: extensibility, integrations, data, security, governance, deployment, observability, and reliability.
AI is not necessarily killing low-code. It is making the boundary between low-code and pro-code harder to define. And that may make the labels themselves less important.