AI Summary
5 min readGreg Isenberg, host of The Startup Ideas Podcast, walked into this episode wanting to know whether Gemini 3.0 in Google AI Studio could actually design well — not just generate a passable layout, but take feedback, handle reference images, and produce something he would actually want to use. He tested it on three real projects: a personal website, a SaaS dashboard, and a mobile app. The results were strong enough that he gave the tool scores between 8.3 and 9 out of 10 across the three tests, and the episode is essentially a live walkthrough of how he got there.
The Personal Website: Reference Images and Taste Over Technical Skill
Isenberg started by redesigning his own personal site, gregisner.com. He gave Gemini 3.0 a screenshot of his current site and a deliberately vague prompt: make it more colorful, more interesting, and design it like a Microsoft XP experience. The tool built a full React app with Tailwind CSS in about 111 seconds. The result was a working Windows XP-style interface with a notepad, application icons for his newsletter and guides, and a tagline pulled from his existing copy. "This is crazy," Isenberg said. "How does this even... there's no way this works on mobile, right? Actually, it works on mobile, surprisingly."
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of The Startup Ideas Podcast
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 (00:00) **Episode Intro & Challenge Setup** - Greg announces he will test Gemini 3.0 in Google AI Studio by designing three types of apps: a personal website, a SaaS app, and a mobile app.
- 2 (00:54) **Test 1: Redesigning a Personal Website** - Greg uploads a screenshot of his current site and asks Gemini to make it more colorful and interesting.
- 3 (03:13) **Observing Gemini's Design Process** - Greg notes that AI Studio shows the model's "work" as it generates files (metadata, index.html, React entry point).
- 4 (05:33) **Sponsor Break: Idea Browser** - Greg pitches his own product, Idea Browser, which sends daily startup ideas backed by data trends.
- 5 (06:40) **Reviewing the Personal Website Output** - Gemini has created a Windows XP-style personal site with a notepad, app icons, and guides.
- 6 (08:23) **Iterating on the Design with Feedback** - Greg asks Gemini to fix the app icons to look more like real Windows/Mac icons and to change the white background.
- 7 (15:48) **Test 2: Designing a SaaS Dashboard** - Greg uploads a Dribbble reference image of a clean SaaS dashboard and asks Gemini to build an analytics/AI app for restaurants in a similar style.
+ Full timestamped outline available in the app
Show Notes
On today’s episode I stress-test Gemini 3.0 in Google AI Studio to see how good it really is as a designer, not just a code generator. Across the episode, I ask Gemini to redesign my personal website in a Windows XP–inspired style, build a restaurant analytics SaaS dashboard, and create a workout mobile app inspired by the “Brain Rot” app. Along the way, I experiment with prompts, visual annotations, and reference images to see how well Gemini takes feedback. By the end, he’s rating each build.
Timestamps
00:00 – Intro
00:54 – Personal Website
15:48 – SaaS
21:52 – Mobile App
26:35 – AntiGravity
27:17 – Final rating and takeaways
Key Points
- Gemini 3.0 can now generate full, styled web and mobile UIs (not just “purple Tailwind vibe-coded” layouts) when given strong prompts and references. greg-take-02
- A Windows XP–themed personal site, built from a screenshot and a short prompt, impresses Greg enough that he considers redoing his actual homepage. greg-take-02
- Visual annotation inside Google AI Studio (drawing on the canvas and commenting) is a powerful way to refine icons, backgrounds, and layout without “speaking designer.” greg-take-02
- A restaurant analytics SaaS dashboard (“Chef OS”) shows how combining Dribbble shots + Teenage Engineering hardware as references pushes Gemini toward more tactile, “real button” UI. greg-take-02
- The “Gains” workout app, modeled on the Brain Rot app, demonstrates that AI can remix an existing product pattern into a new behavior-change app with streaks, goals, and a reactive mascot. greg-take-02
- Greg’s big takeaway: good ideas + taste + references + Gemini 3.0 let non-designers ship highly differentiated experiences, raising their odds of standing out. greg-take-02
The #1 tool to find startup ideas/trends - https://www.ideabrowser.com
LCA helps Fortune 500s and fast-growing startups build their future - from Warner Music to Fortnite to Dropbox. We turn 'what if' into reality with AI, apps, and next-gen products https://latecheckout.agency/
The Vibe Marketer - Resources for people into vibe marketing/marketing with AI: the
More from this podcast
The Startup Ideas Podcast →