Designing a Dashboard with AI: Steve Schoger Shows His Process
Published on 13.04.2026
Designing a Dashboard with AI
TLDR: Steve Schoger, the designer behind Refactoring UI and Tailwind UI, records himself designing a dashboard with AI assistance. The video is a practical, unfiltered look at how AI fits into a real design workflow — not a polished marketing demo, but actual work.
Steve Schoger is one of those rare people who can make you feel like you've been designing wrong your whole career, in the most constructive way possible. If you've read Refactoring UI or spent any time inside Tailwind UI, you know his eye for detail is exceptional. So when someone like him sits down to try designing a dashboard with AI, I pay attention.
The video is framed around the actual process of building a dashboard UI with AI in the loop. What makes this worth watching is that Schoger isn't trying to sell you on AI being magical. He treats it as a collaborator with specific strengths — generating layout scaffolding, suggesting component structures, getting you past the blank canvas problem — while making clear that the decisions that actually make a design good still come from a trained human eye. The AI can produce something plausible. Whether it's right is a different question entirely.
There's a real tension in AI-assisted design that this kind of video surfaces honestly. Tools like Cursor, v0, and various AI design assistants have gotten very good at producing things that look finished at first glance. The problem is that "looks finished" and "is good" are not the same thing. Spacing decisions, information hierarchy, the way a user's eye moves across a dashboard — these require taste and intent, not just pattern matching from a training corpus. Schoger's workflow shows where AI saves time and where it would cost you time if you just accepted what it produced.
For dashboard design specifically, the stakes are higher than a landing page. Dashboards communicate state, trends, and action items to users who need to make decisions quickly. A bad layout isn't just ugly — it's cognitively expensive. Watching how an experienced designer interacts with AI output, corrects it, and builds on top of it is genuinely instructive.
Key takeaways:
- AI tools are useful for overcoming the blank canvas problem in UI design, but the output requires critical evaluation from someone with design training
- Dashboard design demands strong information hierarchy — AI can scaffold structure but struggles with the intentional visual weight decisions that guide user attention
- Steve Schoger's process shows a practical model: use AI to generate, then apply judgment to refine rather than accepting the first plausible-looking result
- The gap between "looks reasonable" and "is actually good" is where design expertise lives, and that gap doesn't close just because AI generated the first draft
Why do I care: As a senior frontend developer, I spend a meaningful amount of time either building dashboard interfaces myself or working alongside designers who are building them. The question of where AI fits in that workflow is not academic — it affects every sprint. Watching someone with Schoger's caliber of design experience navigate these tools honestly, without pretending AI is either useless or a replacement for skill, is exactly the kind of signal I want. It also helps me have better conversations with designers about what AI can reasonably handle versus where I should not expect it to substitute for their judgment.