Playbook
Tasteful Skills
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.
Back to Podcast Digest
AI Made Junior Designers Look Like Most Seniors: Hannah Ahn
TL;DR
At Superpower, trust isn't a UX nice-to-have — it's clinical risk management — Hannah Ahn says health design carries consequences far beyond SaaS, so every feature has to clear compliance, copy clarity, and emotional communication because a bad lab result delivered poorly can genuinely harm a user experience.
Superpower pushed the whole company onto Claude Code in January, and the breakthrough came from non-coders — one video lead with zero coding background spun up a GitHub repo and built a Spotify-styled Premiere Pro plugin for shared media assets, which became the internal proof that anyone could automate their own workflows.
AI has made idea generation cheap, so design leadership now means saying 'slow down' — Ahn describes a 'mega acceleration' in landing pages and prototypes, then having to add more reviews so the team didn't ship 'slop' just because AI made drafts instantly available.
She hires for 'love of the game' before AI literacy — Ahn argues that people who genuinely care about design will naturally become AI-native, and she looks for signals like portfolios that reflect current taste, craft, and difficulty of execution rather than generic work that could have passed 10 years ago.
AI is flattening the ladder for junior designers, but also letting strong juniors look like seniors — Ahn says she's seeing 20-year-olds with portfolios comparable to designers in their 30s, often with spikes like design engineering, which makes agency and self-belief more important than waiting for the old apprenticeship path.
In AI products, trust comes from the extra 1%, not average flows — her core test for designers is whether they care enough to shape the tiny details, interactions, evidence, and emotional cues that make users feel a company has rigor, especially in health where AI can summarize data well but shouldn't be prescribing medicine.
The Breakdown
Why health design feels different from Canva
Ahn opens with the core shift: at Superpower, trust is the product. In Canva-style SaaS, a fuzzy message might be annoying; in health, bad communication around labs or genomics can deeply affect how someone interprets their body, so the team layers clinical compliance, careful copy, and emotional clarity into every release.
From product manager to designer because PM got boring
She didn't come up through design school at all — no college, straight into product management at Canva, with an 'internet kid' background learning Photoshop in high school. Canva taught her how good products get made across user, business, and engineering constraints, but she admits the day-to-day of docs and data got dull; she wanted to be the one shaping what the product actually felt like.
The team runs on autonomy, short toes, and a very high taste bar
Superpower's design team is only about five people right now, with plans to nearly double it, and Ahn says what works is autonomy. She loves the 'short toes' culture — if you see a problem, fix it — and treats brand guidelines as 'the floor, not the ceiling,' so designers can bring their own flavor as long as the standard stays high.
What she actually looks for when hiring designers
Ahn keeps splitting the hiring question into two layers: what a person is good at, and whether that shape fits the team composition. She doesn't expect every designer to build internal tools or Figma plugins; some are there for design engineering, others for feel, content design, imagery, or art direction — but across the board she wants AI literacy, speed, and especially people who clearly love making things.
Why 'love of the game' matters more than AI-native branding
For her, AI literacy usually follows genuine obsession with the craft. She scans for whether a portfolio feels current, whether the designer tracks where taste has moved over the last decade, and whether the work has that hard-to-fake edge of difficulty — her example being how creative taste shifted from stock photography to illustration to whatever now feels harder and more distinctive to make.
January was the Claude Code turning point
The company made an all-hands push in January to get everyone into Claude Code, GitHub, and workflow automation, led in part by a 21-year-old agents lead who had been automating Shopify and dropshipping systems since he was 18. The most convincing internal case study came from their video production lead, who had never coded before but built a Premiere Pro plugin for media asset deployment in a weekend — Ahn's reaction was basically, 'holy shit.'
AI sped up output so much they had to add more review
Ahn says the last few weeks have felt like 'mega acceleration': landing page strawmen and concept drafts are now absurdly easy to produce. But instead of rewarding instant shipping, she's had to tell the team to use the saved time for polish, because just because AI can generate something fast doesn't mean it deserves to go live.
In health AI, the real differentiator is care
She draws a clean line on what AI should do: aggregate medical records, genomics, and blood data into rigorous insights better than most doctors can, yes; prescribe medicine and own the risk, no. For trust-heavy products, she doesn't think there are magic interface patterns so much as signals of care — the evidence, the tiny details, the emotional weight of interactions, and the brand's broader credibility all add up to whether users believe you should be helping with their health.
Share
Keep Reading
Two Ways to Follow
The Weekly Echo. The inbox-shaped summary of what mattered.
New editorials announced here.
