Stop Shipping AI Slop. Design with Weavy AI, Claude etc.
Greg Isenberg and designer Suraya Shivji demonstrate how to create beautiful, marketable AI-powered apps by combining functional prototyping with intentional branding and visual design. Rather than relying solely on AI to generate complete products, the episode reveals a workflow that uses AI tools strategically—starting with brand strategy, mood boards, and visual assets before assembling everything into a cohesive interface. The duo builds a voice journaling app called "Cassette" live on camera, showing why most AI-generated apps look generic and how thoughtful design thinking transforms them into products people actually want to use.
Key takeaways
- • Most vibe-coded AI apps fail not because of functionality but because they lack distinctive branding and visual identity—everything looks the same, so differentiation requires intentional design choices made before building.
- • Start by defining how your product should make users feel, not just what it does; this emotional foundation guides all subsequent design decisions and prevents creating generic, tech-forward interfaces when a product needs to feel personal or analog.
- • Use mood boards on Cosmos to gather visual inspiration aligned with your brand vision, then feed those reference images into Weavy AI with image generation models like Flux 2 Pro to extract color palettes, create custom assets, and maintain visual consistency throughout your product.
- • AI image generation models have distinct strengths: Flux 2 Pro excels at general visual assets, Ideogram specializes in typography and logos, and Gemini performs better at complex UI animations and shaders.
- • Rather than fully outsourcing design to AI, keep brand decisions and visual direction in your hands while delegating solved problems like layout and asset generation; this ensures your product feels distinct rather than AI-generated.
- • You can iterate efficiently between design and code by using Claude Code and Cursor for production work while relying on Google AI Studio for one-shot prototypes, choosing your tool based on whether you're starting fresh or refining an existing codebase.
Recommendations (9)
"We're going to use Flux 2 Pro. It's a very good just like general model to use for a bunch of stuff."
Suraya Shivji · ▶ 19:56
"Google AI Studio. It's definitely one of my favorite vibe coding apps. It's really good at oneshotting interfaces."
Suraya Shivji · ▶ 4:35
More from these creators
23 AI Trends keeping me up at night
Stop Vibe Coding. Start Getting Customers.
Paperclip: Hire AI Agents Like Employees (Live Demo)
Firecrawl AI clearly explained (and how to make $$)
I fixed OpenClaw so it actually works (full setup)
Building AI Agents that actually work (Full Course)