The biggest takeaway from my design work over the past two years is just how much our workflow has actually changed. We used to follow a linear path: Requirement → Wireframe → Visuals → Handoff → Development. Now, I often start by running core interactions through Large Language Models (LLMs) to build a prototype first. Whether it’s a complex user flow, component states, or intricate interaction logic, I use AI to quickly generate a functional demo.
By "using" the product myself at such an early stage, I catch issues that were previously invisible in static design files. Most logical flaws aren't caught during design reviews anymore—they are caught the moment you actually interact with them. Exposing these issues early prevents massive rework during the final QA stage.
However, this new shift brings its own set of challenges. The UI generated by LLMs still lacks the polish of a professional design file. Relying solely on prompting to tweak visual details feels like a game of chance; it’s hard to achieve precision.
My current, more stable workflow looks like this:
Validate the interaction quickly using an LLM.
Export the structure back to Figma to refine the UI details.
I’ve recently been using a Figma plugin called Pixlore. It converts AI-generated page structures directly into Figma layers with clean component hierarchies. This approach gives me the best of both worlds: using AI to validate the solution early on, and then using Figma for high-fidelity refinement. It makes the entire process significantly smoother.
Here's how to use it—give it a go!
Try it：https://pixlore.newportai.com/despilot-server/operation/trace/promotion/dt-jcy-20260311
For further actions, you may consider blocking this person and/or reporting abuse
Top comments (2)
Cool！I‘ll try it!
GO ahead！