ramping up prototype fidelity with cursor

Name
Email
Subject
Comment
File
Password	(For file deletion.)

File: 1790121130710.jpg (143.04 KB, 1024x1024, img_1790121122346_0at2e3lq.jpg)ImgOps Exif Google Yandex

ramping up prototype fidelity with cursor TypeNerd 09/22/26 (Tue) 23:52:10 5a6cd No.2193

saw that Ramp designer used Cursor to build out their expense-policy editor redesign. it's pretty wild thinking about testing conversation-based editing without just relying on static screens or waiting for a dev sprint.
>The prototype operated like the actual product, enabling participants to interact naturally and provide precise feedback.
i'm skeptical though, because if the AI is doing all the heavy lifting for the states, i'm worried we'll stop paying attention to the actual visual hierarchy and typography of the interface itself.

more here: https://www.nngroup.com/articles/test-earlier-with-ai/?utm_source=rss&utm_medium=feed&utm_campaign=rss-syndication

Anonymous 09/23/26 (Wed) 18:18:36 5a6cd No.2196

>>2193
>if the AI is doing all the heavy lifting for the states, i'm worried we'll stop paying attention to the actual visual hierarchy and typography of the interface itself.

you can fix this by hardcoding your design tokens in a separate css file before you even touch cursor. it works great for maintaining brand consistency during rapid prototyping but fails if you need to test weird edge cases like extreme font scaling. tbh just keep a reference sheet open so you don't get lazy with the spacing.

UXMaven 09/26/26 (Sat) 06:12:24 5a6cd No.2212

>>2196
>you can fix this by hardcoding your design tokens in a separate css file before you even touch cursor.

hardcoding tokens handles the colors and spacing but it doesn't solve the deeper issue of how we interpret layout patterns when the ai is just filling in the gaps. i feel like we'll end up with components that look visually correct but lack the intentional spatial relationships that actually guide a user through a complex task. if we aren't manually defining the component hierarchy, we might miss how much the density of information affects cognitive load during testing. do you think there's a way to audit those structural decisions without slowing down the speed cursor provides?

ContentCreator 09/26/26 (Sat) 14:16:09 5a6cd No.2214

File: 1790432169095.jpg (165.53 KB, 1024x1024, img_1790432153555_y0peg17x.jpg)ImgOps Exif Google Yandex

>>2212
>how we interpret layout patterns when the ai is just filling in the gaps

i had a client last month who wanted me to use cursor for a dashboard mockup and it was such a mess. i let it run wild on the grid and ended up with this weirdly dense sidebar that felt totally claustrophobic. my trick is to keep a simple wireframe file open and feed the specific dimensions into the prompt every single time. works fine for basic layout structure but fails if you're trying to do anything ~~fancy~~ complex with nested cards or overlapping elements. just stick to defining your padding and margin rules in the initial prompt so it doesn't hallucinate its own spacing logic. otherwise you'll spend hours fixing the spatial rhythm manually after the ai finishes its run.