Ron Goldin - Building your ideas as a design leader

2026-06-09 - 54 min - source - Read full transcript

Vid (host)Ron Goldin

Key insights

Direct, embodied field research still beats reports for driving organizational action; video that transmits felt experience persuades where written memos get shelved.

Ron's Uber Eats delivery research (biking food deliveries himself across New York, Mexico City, Tokyo, and Bangkok) surfaced the real cause of courier idle time - couriers converging on already-popular restaurants and starving less-visible 'hotspots' of demand. He credits a raw video shown at an all-hands, including a full minute of uncomfortable silent waiting, with driving Uber's broader dog-fooding program, contrasting it with the more common fate of research memos: nodded at, filed, forgotten for two or three quarters.

research-and-empathy

Ambitious executive mandates without direction ('make it better, no direction') are a normal, even generative, condition for senior design leaders, not a failure of process.

At Shopify, CEO Tobi Lutke told Ron's team to redesign an already industry-leading checkout with no specifics on what 'better' meant. Ron frames this ambiguity as core to Lutke's leadership style (always tearing things down and rebuilding) and says it produced one of his most productive stretches, including checkout redesign, OpenAI agentic commerce integration, and Coinbase crypto checkout within six months.

design-leadership-in-ai-era

A non-engineer designer built a working prototype that reversed a VP-level 'this can't be done' verdict by vibe-coding the feature himself.

After a design VP said automatically pulling brand colors and vectorizing a logo from a merchant's website wasn't feasible, and an engineer he asked was noncommittal, Ron built the tool himself in Cursor: it scraped a site's CSS for likely brand/accent colors and converted a low-res logo JPEG into a recolorable, transparent SVG. He calls this an underused 'superpower' designers now have to demonstrate feasibility rather than requesting it.

ai-native-design-workflow

Demonstrated working prototypes ('the aura of inevitability') make it far harder for stakeholders to say no than a proposal alone.

Referencing a Cleo quote about early Facebook, Ron and the host discuss how showing a functioning prototype, not just describing an idea, shifts organizational momentum: once something visibly works, the conversation moves from 'should we' to 'how do we ship this,' which Ron says is now achievable by designers directly rather than requiring engineering buy-in first.

ai-native-design-workflow

AI tooling has made the 'player-coach' design leader model close to universal at senior levels; nearly every leadership role Ron interviewed for expected hands-on building, not just management.

He reports being unable to recall a single senior design leadership role in his recent job search that didn't expect hands-on contribution, including one early-stage startup that put him on a live call to verify he could use Figma's auto-layout. He frames this as a structural shift toward flatter orgs and increased 'wingspan' per manager, making hands-on building a de facto requirement rather than a nice-to-have.

design-leadership-in-ai-era

Personal side projects built outside of work are increasingly weighted by employers as evidence of product, technical, and entrepreneurial capability, sometimes over traditional work experience.

Ron says companies now look at things candidates built independently (like his company Matchmaker) not as 'extra credit' but as direct evidence of initiative, resilience, product thinking, and technical savviness, and that this is becoming a more explicit hiring signal than it was previously.

founder-building-outside-work

Building a product from scratch is dramatically faster now, but iterating through multiple failed platform choices before landing on the right one is still normal.

Ron and his wife rebuilt Matchmaker three times: a no-code tool called Softr that ended up 'ugly as hell,' then Lovable, chosen deliberately over hand-coding so his non-technical wife and her business partner could run the product without him. He notes a comparable later project went from 'a few months' (Lovable's first version) to about a day for a similarly scoped build, crediting both tool improvements and his own accumulated skill at directing AI.

founder-building-outside-work

Writing a rough strategy doc/PRD first and feeding it to Claude for a build plan produces materially better AI-built output than prompting ad hoc.

Ron's workflow: write a messy, typo-ridden PRD covering the product's goal, audience, and critical user journeys, then hand it to Claude to generate a structured plan and prompt sequence for the build platform. He notes Claude proactively identifies gaps and asks clarifying questions (e.g., about what 'brand' means in the product), and that maintaining a persistent project-memory context across sessions measurably improves output consistency.

ai-native-design-workflow

Building software has stopped being the hard part; growing an audience for what you built remains as hard as it always was.

Despite reaching a fully functional, revenue-taking, incorporated business with Matchmaker, Ron says the team is still in early stages specifically because distribution, not construction, is now the bottleneck - a shift he sees widely discussed as more people realize anyone can vibe-code a working product from scratch.

founder-building-outside-work

Product thinking, scrappy user research, and brand/visual point-of-view are the three skills Ron sees as durable design differentiators as AI commoditizes execution.

He argues product managers no longer need designers purely for execution, so senior designers must own problem framing (what user/business problem, what experience fits). He also flags that AI-generated interfaces increasingly look identical because they're trained on the same corpus, making early, intentional branding and visual point-of-view a competitive necessity rather than a late-stage polish step.

hiring-and-team-skills

Motion design is an underexploited differentiator that current AI tools cannot yet produce well, making it a high-value, scarce skill.

Ron argues that with static/flat layouts converging across products, motion is one of the few remaining channels for brand expression, but current AI tools can't yet surprise you with excellent motion work, and most engineers (even at top companies) are weak at performant motion implementation. He calls people who combine visual/motion craft with a builder mindset - 'brand engineers' - a scarce, top-tier hire right now.

hiring-and-team-skills

A light technical background (reading terminal/agent output, understanding what code is doing) meaningfully improves a designer's ability to debug and direct AI builders.

Ron credits a minor in computer science with letting him recognize when an AI agent (e.g., Lovable) is stuck in a loop or recursing incorrectly and intervene to help it debug itself, arguing this technical literacy is a durable advantage even though AI nominally 'does the coding' now.

hiring-and-team-skills

Companies

Uber - Ron's early design role at Uber Eats, where he did courier deliveries himself in New York, Mexico City, Tokyo, and Bangkok to research the driver/courier experience firsthand.
Shopify - Ron led online checkout at Shopify, redesigning what the company marketed as the highest-conversion checkout in the industry per a direct mandate from CEO Tobi Lutke; also led launches with OpenAI's agentic commerce chat experience and Coinbase cryptocurrency checkout.
Google - Named as one of the companies where Ron led design, alongside Shopify and Uber Eats.
Matchmaker - The startup Ron co-founded with his wife Melanie and a third co-founder, a software product matching brands with partnership/influencer opportunities, built and rebuilt three times using Softr, then Lovable.
Softr - The no-code, Framer-like tool Ron and his wife first tried to build Matchmaker's MVP with before moving to Lovable; described as ending up 'ugly as hell.'
Lovable - The AI app-builder Ron chose over hand-coding for Matchmaker specifically so his non-technical wife and her business partner could run and extend the product without him.
Cursor - The AI coding tool Ron used at Shopify to vibe-code an internal prototype (brand color/logo extraction for checkout) after an engineer said the idea couldn't be done; also an early tool in his personal-project toolkit.
Figma - Ron's persistent visual design tool of choice even amid heavy AI/vibe-coding use; he builds high-fidelity mocks and small design systems in Figma and feeds them into Lovable/Claude, and still sketches fixes there rather than describing them in words.
Claude - Ron's primary AI planning and building partner: he feeds it rough PRDs to generate build plans, asks it to flag gaps in his thinking, values its long-term project memory, and highlights its surprising skill at animating SVGs.
Jitter - Sponsor mention; a motion-design tool with animated, reusable components that Ron says lowered the barrier to doing polished motion work compared to After Effects.
Dsn (Diesign/Destin) - Sponsor mention; a design/prototyping tool ('Surfaces' feature) that lets designers prototype directly on top of an existing production interface pulled from prod.

Techniques and frameworks

Empathy-building through self-directed field research - At Uber Eats, Ron personally did food deliveries by bike in New York and internationally (Mexico City, Tokyo, Bangkok), recording the experience on an early Insta360 camera and Google Cardboard to transmit the courier's pain and confusion directly to leadership.
Video-as-persuasion over written memos - Instead of writing a report, Ron cut hours of raw delivery footage into a 3-minute video shown at an Uber Eats all-hands, deliberately leaving a full minute of dead time to make the audience feel the courier's wasted waiting, which he says spearheaded Uber's broader dog-fooding program.
Prompt-first PRD-to-plan workflow with Claude - Ron writes a rough, typo-ridden strategy doc/PRD, drops it into Claude, and asks it to generate a build plan and a series of prompts for the target platform (Lovable, Cursor); Claude asks clarifying questions and flags gaps before building starts.
Staged, sprint-like AI build sessions - Rather than building an entire product at once, Ron breaks builds into small phases (e.g., core screens first, settings and chat later), checking output and managing credit burn each session, likening each short AI session to a compressed sprint.
Persisted project memory / context store - Ron keeps a running store of project context (what the product is, who it's for) that he references across AI sessions so the model doesn't lose track of the product identity; he notes Claude handles long-term context better than Lovable.
Dual-tasking between AI generation and manual Figma fixes - When AI output is close but not quite right, Ron chooses between describing the fix in words to the AI or just sketching/adjusting the component directly in Figma, since manual pixel-level correction is often faster than a text back-and-forth.
Vibe coding to bypass 'no' answers - When a design VP said a feature (auto-extracting brand colors and vectorizing a logo from a merchant's website) couldn't be done, Ron built a working internal prototype himself in Cursor rather than waiting on an engineer, then presented the finished tool as proof.

Summary

Ron Goldin, who has led design at Google, Shopify, and Uber Eats and recently co-founded the startup Matchmaker with his wife, joins host Vid for a wide-ranging conversation about building products and careers as design leadership shifts under AI. The episode opens with Ron's origin story as a researcher-minded designer: in his early days on Uber Eats, he personally did food deliveries by bike across New York, then Mexico City, Tokyo, and Bangkok, recording the experience with an early Insta360 camera and Google Cardboard headset. The footage, rather than a written report, revealed that couriers were clustering around already-popular restaurants while nearby "hotspot" demand went unserved, and a 3-minute cut shown at an Uber Eats all-hands (including a deliberate minute of silent, dead-time waiting) is credited with spearheading the company's broader dog-fooding research program.

The conversation moves to Shopify, where CEO Tobi Lutke tasked Ron's team with redesigning an already industry-leading checkout with essentially no specific direction, a pattern Ron frames as core to Lutke's leadership style and one of his most productive stretches (checkout redesign, an OpenAI agentic-commerce integration, and Coinbase crypto checkout within six months). It was there that Ron had his personal "light bulb" moment with AI: after a design VP said automatically extracting brand colors and vectorizing a merchant's logo from their website "can't be done," and an engineer was noncommittal, Ron built the feature himself in Cursor as a non-engineer, then presented the working prototype as proof. He and the host connect this to a "Cleo" idea about the "aura of inevitability" - a working demo makes it far harder for an organization to say no than a proposal ever could.

A large stretch of the episode covers Ron's co-founded startup, Matchmaker, a tool matching brands with partnership and influencer opportunities, built with his wife Melanie and a third co-founder. The product went through three builds: a rough MVP in the no-code tool Softr that ended up "ugly as hell," then a rebuild in Lovable, chosen deliberately (over hand-coding, despite Ron's CS background) so his non-technical wife could run and extend the product without him. Ron replicates his design process inside AI tooling: writing a messy strategy doc/PRD, feeding it to Claude to generate a structured build plan, building in small staged sessions rather than all at once, maintaining persistent project context across sessions, and toggling between describing fixes to the AI in words versus just sketching the correction directly in Figma. He notes that building software has gotten dramatically faster (a comparable later side project took about a day versus Lovable's original few months) but that growing an audience remains just as hard as ever.

On hiring and team-building, Ron argues that nearly every senior design leadership role he interviewed for recently expected hands-on building ("player-coach"), including one startup that live-tested his Figma auto-layout skills on a call. He sees personal side projects (like Matchmaker) increasingly weighted by employers as real evidence of product, technical, and entrepreneurial capability, sometimes outweighing traditional experience. Looking forward, he names three durable design skills as AI commoditizes execution: product thinking (defining the real user/business problem, not just executing), scrappy hands-on research, and brand/visual point-of-view, since AI-generated interfaces increasingly converge on similar-looking output. He singles out motion design as a particularly underexploited differentiator, since current AI tools can't yet produce excellent motion and most engineers are weak at performant motion implementation, making people who combine visual craft with a builder mindset - "brand engineers" - especially scarce hires.

The episode closes on a lighter technical note: Ron credits his light CS background with helping him debug AI agents that get stuck in loops, and describes being newly impressed by Claude's ability to animate SVGs beyond what he thought was achievable, a recurring theme of surprise at what current AI tools can already do versus what he assumed was impossible.

Notable Quotes

"I'm a non-engineer, but I just built the ability to create an SVG that can be colored to any color you want to... and a merchant never will have to lift a finger." - Ron Goldin

"Building software isn't as hard as it used to be, but growing an audience is as hard as it always has been." - Ron Goldin

"Prototypes win arguments. Prototypes get people psyched... moving away from slide decks, moving away from PRDs, moving into something that people can actually feel and react to." - Ron Goldin

"I think we joke about the purpleness of AI, but I do think you have a lot of non-designers right now looking at the outputs of AI tools and they're like, 'Oh, that's pretty good.' But a lot of them look very similar, because they're literally all being trained on the same world of stuff." - Ron Goldin