Generating Frontend Interfaces with LLMs: Comparing Tools (As of 10/2025)
· Dylan Halpern
LLM-based coding tools have come a long way over the past few years, and dreams of all-in-one app builders from a chat interface feel closer than ever. But how close are we? Which tools do the best, and what are the trade offs?
To understand the affordances of frontend app builders (as of October 2025), we tested three workflows:
- Mockup-driven: Make mockups either by-hand or using image generation, and then pass the mockups to an AI agent to build the application.
- Full-service AI system: Use a FE-specific AI tool like v0 or Lovable to design, develop, and preview your app.
- AI-assisted development: Use AI agents like Cursor or Claude Code to design and develop your app, with your assistance.
The first approach mirrors more traditional static design → frontend workflows — one our team rarely uses given our emphasis on data science and engineering rather than visual design. Our team usually roughs out interface designs and user experiences as we go, given small project size and focused engagements. If an LLM tool can provide a reasonable quality design and UX, the expanded capacity for our engineers focusing on backend work and our data scientists would be a serious boon.
We found that for our development cycles, using a desktop coding agent and managing the deploy more traditionally is often the best fit. A few of the browser-based chat app builders have potential. Generating static mockups using LLM tools was not a good fit; for people with workflows already integrated with tools like Figma, translating to interactive code may be viable.