Creating Taste: Brandon Jacoby on AI-Amplified Design
Top designers create taste by knowing when to break patterns and invent new ones; AI amplifies those who build custom tools and decide ruthlessly, enabling indie practices to push founders past 'good enough.'
Decisiveness Unlocks Design in Chaos
Brandon Jacoby, who designed at early Cash App, Capital (first design hire), and X, describes X as a pressure cooker that shattered his preconceptions. Unlike big companies where ideas languish in committees, X demanded instant decisions—no barriers, no historical metrics sacred. In one design review, a months-tracked metric was upended on a whim because someone questioned its relevance, leading to rapid experimentation despite user risks. This first-principles approach, embodied by Elon Musk, forced Jacoby to question every micro-decision in interfaces, shedding 'calloused' constraints designers accumulate.
The opportunity? Thriving in 0→1 environments where speed trumps polish. Before X, Jacoby optimized onboarding at Cash App for 45 million users via A/B tests and micro-changes. At X, scale amplified stakes—onboarding couldn't break. Tradeoff: Comfort with iteration yields to ruthless discernment. Result: Designers learn 'decisiveness' as a core trait, essential in AI eras generating endless options.
"You just have to question every requirement and if someone says anything that would get in the way of questioning a requirement, why is that a requirement in and of itself?" – Brandon on X's culture, highlighting how it dismantles passive constraints for bolder designs.
Onboarding: Patterns vs Invention
Nikita Bier, X's former Head of Product, schooled Jacoby on onboarding in a Hawaii hotel room doc-dump—'Pandora's Box' of insights. Problem: Teams reinvent wheels on utilitarian flows despite proven patterns from massive growth teams. Jacoby, onboarding-weary from Cash App, realized 90% of products share effective patterns; blindly innovating wastes time.
Decision chain: Evaluate context—follow patterns for scale/reliability (e.g., X's precious flow), invent for differentiation. Nikita's doc revealed overlooked basics, implemented post-Jacoby's exit. For juniors: Master 'know when'—revert to mean or break it. Example: Tolan app's creative onboarding as rare invention.
Tradeoffs: Invention risks breakage; patterns stifle taste. Post-X, Jacoby applies this to client products, prioritizing utilitarian flows first.
"There are patterns that work that show up in 90% of the products out there... know when to reinvent the wheel, know when to follow patterns." – Core onboarding lesson from Nikita, distinguishing junior growth from senior discernment.
Post-Burnout Pivot to Indie Autonomy
Burned out from X's grind, Jacoby explored for 3 months: Helping friend startups, incubating ideas across design, brand, product. Surprise: This fluidity—most fulfilling in years. Timing perfect with AI models exploding, enabling non-technical hackers like him to prototype deeply without engineers.
Options considered: Return corporate? No—craved autonomy. Rejected full pivot to code (still loves Figma 'dragging rectangles'). Chose indie practice: Help founders 'push past good enough' in 0→1. Why? AI fills gaps (logistics agent, overnight recaps, Figma AI for blocks), letting him focus on judgment.
Mentor Owen Jennings (Block, 4 years) shaped craft-consequence balance. Result: Solo practice blending UI, brand, direction—AI enables small-team scale.
Tradeoffs: Uncertainty vs freedom; burnout recovery via exploration. Now positions as taste-creator for startups, using decisiveness from X.
AI Amplifies Builders Who Hack Tools
AI doesn't replace taste—it empowers custom tooling. Jacoby builds bespoke agents for client work (e.g., logistics, effects), echoing pre-AI trailblazers at Cash App who 'ran through walls.' Spectrum of amplification: Winners unbound by barriers, questioning reality like at X. Losers? Stuck wishing for skills—solution: Prompt Claude for WebGL, particles.
When to reach: AI for speed/exploration (cloud code prototypes), Figma for expression. John Lasseter's Pixar mantra: "The technology inspired the art, the art challenged the technology." Now, tech limits vanish—build what’s in your head.
Examples: Particle effects from vague ideas; agents for unknowns. For indies: Use across stack (email, Figma Make, code). Tradeoffs: Over-reliance caps human judgment; decisiveness separates rockets from capped ceilings.
Types amplified: 'Tastemakers' balancing deep prompts with flow—zoom in/out. Jacoby's practice: AI supplements, job unchanged—mold valuable products.
"The single biggest way that AI specifically has helped with client work is actually the ability to like build my own design tools." – Jacoby on empowerment, shifting from off-the-shelf to custom AI for creative freedom.
Seeing Taste vs Creating Taste
Core distinction: 'Seeing taste' copies trends/patterns; 'creating' breaks rules strategically. Tastemakers discern: Push boundaries or flow? AI generates options but lacks this balance—deep prompts for novelty vs accepting 'not everything needs to be new.'
Shaped by moments: Lasseter quote, X/X mentors, Cash App trails. For 0→1: Quiet constraints, invent onboarding sparingly. Indie future: AI lowers barriers, decisiveness wins. Push past 'good enough' via craft (pixels) + consequence (impact).
"I think the tastemakers know when to break the rules... knowing that balance... is what creates taste." – Defining 'creating taste,' AI's current limit, and indie edge.
Key Takeaways
- Question every requirement ruthlessly—emulate X's no-committee speed to escape design calluses.
- Onboarding rule: 90% patterns, 10% invention—discern via context to avoid wheel-reinvention.
- Post-burnout: Allocate 3 months exploration; AI enables non-tech autonomy across roles.
- Build custom AI tools (agents, effects)—ask Claude for unknowns to unbound creativity.
- Amplify via decisiveness: Balance deep AI iteration with flow; reject 'human touch forever' debates.
- Indie positioning: Target 0→1 founders; blend UI/brand with taste-creation beyond good enough.
- Taste creation: Zoom in/out—break rules when patterns fail, flow otherwise.
- Tech-art loop: Let AI inspire, challenge it—per Lasseter, limits gone.