Stop Shipping AI Slop. Design with Weavy AI, Claude etc.

SUMMARY

Greg Isenberg interviews designer Suraya Shivji on transforming generic AI-built apps into branded masterpieces using a workflow with Google AI Studio, Claude, Cosmos, Weavy AI, Flux, Ideogram, and Figma to create a voice journaling app called Cassette.

STATEMENTS

Vibe-coded apps built with AI tools like Google AI Studio often look identical and unappealing, making it hard to attract downloads.
Branding is essential for creating products that resonate emotionally and encourage user engagement beyond mere functionality.
Google AI Studio excels at quick one-shot prototypes for interfaces without needing an existing codebase.
Claude serves as an ideal brainstorming partner for defining emotional user experiences and generating brand guidelines.
Defining how a product should make users feel—such as calm and analog for overthinkers—is a foundational step in design.
Vibe coding typically focuses on "what" the app does, but neglecting "how" it feels leads to generic results.
Outsourcing all creative thinking to AI results in sameness; designers should retain control over unique vision.
Mood boards built with Cosmos help visualize analog, vintage aesthetics to guide visual assets.
Weavy AI's node-based interface allows visual experimentation with AI models like Flux for generating palettes and assets from references.
Iterating between Claude for prompts and Weavy AI for outputs enables rapid refinement of design elements.
Brand guidelines function as prompts to ensure consistency across colors, buttons, and logos.
Using textures and aging elements, like cassette tapes, adds purpose and engagement to digital interfaces.
Ideogram excels at generating typography-heavy assets like logos due to its strength in text rendering.
Compositing AI-generated assets in Figma with blend modes ensures cohesive color matching.
Pasting refined Figma designs back into Google AI Studio dramatically improves AI-generated outputs.
Inspiration from single images or references can anchor an entire product's visual identity.
Negative prompts in AI tools help avoid unwanted styles like glossy gradients or corporate minimalism.
Prototyping evolves with AI, shifting focus from building to curating unique emotional experiences.
Tools like Claude Code and Cursor are preferred for integrating designs into existing codebases.
Mobile vibe-coding platforms are improving but often lack depth in backends compared to full AI workflows.
Spending time on inspiration gathering prevents reliance on default AI aesthetics.
Analog metaphors, like cassette tapes for history logs, make digital products feel personal and lived-in.
Designers now benefit less from manual skills and more from curating AI outputs effectively.
The workflow democratizes professional design, allowing non-designers to create agency-level apps.

IDEAS

AI tools enable anyone to prototype apps in minutes, but without branding, they blend into a sea of sameness.
Emotional framing—targeting overthinkers seeking calm—transforms journaling from a tech chore into an analog escape.
Brand guidelines aren't rigid corporate documents; they're flexible prompts that evolve through AI iteration.
Node-based tools like Weavy AI turn complex AI model chaining into intuitive visual experimentation.
A single mood board image can spawn color palettes, textures, and even core product metaphors like aging interfaces.
Cassette tapes as history elements cleverly mimic real-world wear, making repeated use rewarding visually.
Switching models mid-workflow—Flux for general assets, Ideogram for logos—optimizes for specific strengths.
Negative prompts act as design guardrails, preventing AI from defaulting to trendy but mismatched styles.
Figma's blend modes automate color harmony, bridging the gap between disparate AI-generated pieces.
Uploading refined assets to AI prototyper loops back improvements, creating a feedback cycle for better results.
Inspiration from 1980s analog tech evokes nostalgia without gimmickry, grounding digital apps in tactile warmth.
AI democratizes logo design, reducing weeks of work to seconds, but taste curation remains human.
Overusing analog in digital feels forced; tying it to functionality, like usage-based aging, adds authenticity.
Claude's role as a "brainstorming partner" outsources refinement while keeping the designer's vision intact.
Vintage palettes with subtle textures create cohesion that feels premium yet approachable.
Prototypes should prioritize "feeling good" over feature lists, mirroring how users remember favorite products.
Weavy AI's low credit cost makes high-fidelity asset generation accessible for solo creators.
Compositing in Figma provides a "north star" for AI iterations, reducing prompt ambiguity.
Mobile apps benefit from simple typography paired with bold, thematic icons for intuitive navigation.
AI's rise shifts designer value from execution to reaction—spotting and selecting the best outputs.
One visual anchor, like a textured audio system, can unify an entire app's aesthetic narrative.
Gathering endless inspiration via Cosmos prevents creative blocks and ensures originality.
Vibe coding's future lies in hybrid workflows: AI for speed, human curation for soul.

INSIGHTS

True product differentiation arises not from features but from evoking specific emotions, turning users from passive scrollers into engaged adopters.
AI workflows invert traditional design by starting with feeling over function, ensuring outputs align with human desires rather than technical checklists.
Node-based tools unlock AI's potential by making model orchestration visual and iterative, much like sketching ideas before committing to code.
Single-source inspiration, like a mood board image, amplifies creativity exponentially, as it cascades into cohesive palettes, assets, and metaphors.
Negative prompting in AI mirrors real design critique, filtering out clichés to preserve a unique voice amid algorithmic tendencies toward uniformity.
Aging digital elements with use fosters emotional attachment, bridging virtual and physical worlds to combat the disposability of modern apps.
Compositing in Figma acts as a human-AI mediator, refining raw generations into polished visions that AI alone can't achieve.
Branding's invisibility in success—feeling intuitive rather than overt—highlights how subtle guidelines create memorable, download-worthy experiences.
Democratized tools shift power from elite agencies to individuals, but success hinges on curating outputs with personal taste over blind generation.
Looping refined designs back into prototypers creates a virtuous cycle, where each iteration builds on human-guided AI enhancements.
Analog metaphors in digital design gain depth when purposeful, transforming gimmicks into narratives that encourage habitual use.
The designer's edge in AI eras lies in sensory intuition—reacting to vibes—rather than manual labor, redefining expertise as curation.

QUOTES

"Anyone can build an app now. It's amazing. So much has changed, but now everything looks the same."
"You don't really think of your favorite products in the world and think, 'Oh, it does this.' You think, 'Oh, it makes my life easier and I just feel so good using it.'"
"Brand guidelines are really just a prompt in itself and it's a prompt we're going to bring into Weavy."
"Sometimes if you overuse analog things digitally, but there's no purpose to them, it feels kind of gimmicky."
"The nice part about product design is you don't need like a face to look the same. You just need colors, shadows, lighting to look the same."
"Spend lots of time on Cosmos. Spend lots of time gathering your own inspiration."
"AI is ruining gradients."
"It's all the same exact thing. And even for the new product I'm working on right now, like I literally just found one image that I really liked and now the entire product is based on this like one shadow image."
"Claude is like the perfect brainstorming tool."
"If your app looks like everything else, it's just really tough to actually expect anyone to download it."

HABITS

Start every design process by defining the target user's emotional experience using prompts in Claude.
Build mood boards extensively in Cosmos by searching themes like "vintage cassette" and saving inspiring images to desktop.
Iterate prompts in Claude before inputting into Weavy AI to refine outputs like color palettes or buttons.
Experiment with multiple AI model runs in Weavy AI, comparing results to develop personal taste in aesthetics.
Use Figma's blend modes and background removal to composite and harmonize AI-generated assets quickly.
Alternate between tools like Claude for ideation and Weavy for visualization to maintain creative flow.
Reference movies or TV for emotional arcs when framing product feelings, working backward from desired user state.
Download and organize favorite AI outputs immediately into Figma for easy assembly and iteration.
Spend extended time gathering inspiration before building to avoid generic AI defaults.

FACTS

Suraya Shivji sold her previous company to Snap, leveraging her design expertise in AI workflows.
Weavy AI provides 200 free credits monthly, with a $15 paid plan offering 1,500 credits for heavy use.
Flux 2 Pro is favored for general image generation due to its speed and reliability with text prompts.
Ideogram V3 specializes in typography, making it ideal for logos and text-heavy designs.
Google AI Studio generates full app prototypes in under two minutes from basic prompts.
Cosmos serves as an AI-enhanced Pinterest alternative, clustering images thematically for mood boards.
Claude has improved significantly in iOS and Swift code generation over the past six months.
Apple's design under Jony Ive drew heavy inspiration from Dieter Rams' 1950s-1960s work.
Vibe-coding platforms like V0 are built on advanced models like Claude Opus 4.5 for better backends.

REFERENCES

Google AI Studio: For one-shot prototyping of app interfaces.
Claude: Brainstorming partner for brand guidelines, prompts, and emotional framing.
Weavy AI: Node-based tool for running Flux 2 Pro, Ideogram V3, and generating assets visually.
Flux 2 Pro: AI model for color palettes, textures, and general image generation.
Ideogram V3: Model excelling in logos and typography.
Figma: For compositing assets, using blend modes, and creating iPhone frames.
Cosmos: Alternative to Pinterest for building mood boards with vintage imagery.
Cursor: Code editor with Gemini model for working in existing codebases.
Claude Code: Primary tool for integrating designs into codebases, especially iOS.
Origami: Past node-based design tool that Suraya avoided but sees AI unlocking.
Late Checkout Agency: Helps Fortune 500s build AI products (mentioned in promo).
Idea Browser: Tool for finding startup ideas and trends.
The Vibe Marketer: Resources for AI-driven marketing.
Dieter Rams: 1950s-1960s designer inspiring Apple's aesthetics via Jony Ive.
Movies and TV shows: Used as references for product emotional arcs.
Pinterest: Traditional mood board tool, contrasted with Cosmos.
V0 and Anything.com: Emerging mobile vibe-coding platforms.

HOW TO APPLY

Prototype the basic app structure in Google AI Studio using a simple prompt describing core screens like home, recording, and history to establish functionality.
Define the target user and desired emotional feel in Claude, such as calm and analog for overthinkers, to frame the product's vibe.
Generate initial brand insights in Claude by pasting emotional prompts, extracting key phrases like "analog warmth" and what to avoid, like productivity tools.
Build a mood board in Cosmos by searching terms like "vintage cassette" and saving 10-20 images that evoke the desired aesthetic to your desktop.
Load mood board images into Weavy AI as nodes, then use Flux 2 Pro to extract color palettes with prompts like "extract colors from this reference, make textured and vintage."
Create custom assets in Weavy AI by chaining nodes: input palettes and references into Flux for buttons (e.g., "analog 80s record button using this palette") and history elements.
Develop logos in Weavy AI using Ideogram V3 with Claude-generated prompts, including negative prompts to avoid glossy or corporate styles.
Composite all assets in Figma: add iPhone frames, apply blend modes for color matching, remove backgrounds, and add simple typography for screens.
Paste Figma screenshots and refined prompts back into Google AI Studio to regenerate the app, iterating as needed for a cohesive, branded prototype.

ONE-SENTENCE TAKEAWAY

Elevate vibe-coded apps from generic prototypes to branded delights by prioritizing emotional design through AI-curated inspiration and iteration.

RECOMMENDATIONS

Begin designs with emotional user profiling in Claude to ensure apps evoke calm or nostalgia, avoiding feature-only focus.
Invest time in Cosmos mood boards to source authentic visual references, preventing AI's default sameness.
Use Weavy AI exclusively for asset generation, starting with Flux for palettes and transitioning to Ideogram for text elements.
Always incorporate negative prompts in AI tools to steer away from overused trends like gradients or minimalism.
Tie visual metaphors, like aging cassette tapes, to app functionality for purposeful, engaging narratives.
Composite in Figma before final AI prototyping to create a clear reference, reducing iteration frustration.
Alternate Claude for prompt refinement with direct Weavy experimentation to build taste intuitively.
Skip rigid brand documents; treat guidelines as living prompts updated through dialogue with AI.
For mobile, keep typography simple and use blend modes to harmonize AI assets seamlessly.
Gather inspiration from one powerful image to anchor the entire design, cascading it across elements.
Prototype with Google AI Studio first, then loop back refined assets for superior outputs.
Experiment with multiple model runs in Weavy to compare and select vibes aligning with emotional goals.
Reference analog eras sparingly but meaningfully to add warmth without gimmickry.
Use free Weavy credits for initial explorations, scaling to paid for production workflows.
Build for personal use cases initially to naturally intuit desired feelings and refine backward.

MEMO

In an era where AI can conjure a functional app in minutes, the real challenge lies in making it unforgettable. Designer Suraya Shivji, who sold her startup to Snap, joins podcaster Greg Isenberg to dissect this paradox during a live build of "Cassette," a voice journaling app for overthinkers weary of digital overload. Their workflow transforms a bland Google AI Studio prototype—complete with generic buttons and lists—into a tactile, analog-inspired gem that feels like a cherished relic. Starting with emotional framing, Shivji emphasizes that products endure not for what they do, but for how they make users feel: calm, unpolished, and disconnected from phone fatigue.

The process kicks off in Claude, where prompts unearth the app's soul. "This is for the overthinker that needs to exhale," Shivji explains, distilling the essence into guidelines that reject productivity nagging or social feeds. These aren't stuffy corporate mandates but fluid directives: analog warmth over sleek tech, permission to ramble without judgment. From there, Cosmos replaces Pinterest as a mood-board haven, pulling vintage cassette imagery that evokes 1980s nostalgia—faded tapes, textured plastics, earthy palettes. Shivji saves clusters of these visuals, noting how endless scrolling uncovers subconscious inspirations, much like how Jony Ive drew from Dieter Rams to redefine Apple.

Enter Weavy AI, the episode's star: a node-based canvas that demystifies AI image models. Shivji loads mood-board photos and chains them through Flux 2 Pro, extracting palettes with prompts like "textured vintage colors from this reference." Iterations yield warm browns and subtle grains, leaning into cassette textures for cohesion. Buttons emerge next—an adorable red record icon inspired by 80s gear—followed by history logs as stacked tapes, their spines dated like personal archives. Negative prompts banish glossy gradients, ensuring outputs stay grounded. Shivji switches to Ideogram for the logo, generating tape-label styles that scream authenticity, all for pennies in credits.

Compositing in Figma brings it home. iPhone frames host the assets: splash screen with logo, recording view with timer, library of "used" tapes aging with entries. Blend modes automatically tint elements to match palettes, while background removal cleans edges. It's not pixel-perfect—AI quirks like wonky dates persist—but the vibe sings. Shivji pastes screenshots back into Google AI Studio, prompting a rebuild: "Design based on this logo, cassette history, record button." The result? A leap from robotic to resonant, proving the loop elevates one-shots.

This hybrid dance—AI for speed, human curation for heart—democratizes agency-level design. Shivji admits her edge now lies in reacting to generations, not crafting from scratch; tools like Claude Code or Cursor handle code integration. Yet warnings abound: over-relying on AI breeds sameness, while purposeless gimmicks flop. Their Cassette, with its evolving wear, invites daily use, turning journaling into a ritual.

Beyond tutorials, the session reveals AI's broader shift: from execution to evocation. Platforms like V0 inch toward full backends, but Shivji's method—start with inspiration, end with iteration—empowers solo creators. As Isenberg notes, it's like self-driving cars: pick your automation level, but steer with vision. For aspiring vibe-coders, the message is clear: beauty isn't accidental; it's orchestrated emotion in digital form.