All podcasts / Dive Club / Summary

Katie Dill - The new era of design at Stripe

2026-02-23 - 54 min - source - Read full transcript
Rid (host)Katie Dill

Key insights

AI prototyping collapses the gap between having an idea and showing it, changing how ideas move through the org.
Dill says designers no longer have to pitch a concept, find engineers, and convince them to build a prototype - they can build it themselves and share a working link. This removes a whole layer of persuasion and coordination that used to precede any exploratory work.
ai-native-design-practice
Stripe built an internal tool, Proto Dash, so prototypes look and feel like the real dashboard from the start.
Proto Dash lets anyone spin up an AI prototype already skinned in Stripe's actual product context, which Dill says matters because no product is ever experienced in isolation - putting a new idea in its real surrounding context immediately clarifies how it will actually feel to a user.
ai-native-design-practice
Even 'AI-made' polished visuals require heavy manual finishing before they are ship-ready.
Dill describes crit notes on Stripe's marketing imagery calling out lighting that was slightly off and shadows that weren't exactly right - the team combined AI generation with manual work in Photoshop's 3D controls to get realism right, arguing that subconscious cues from these details affect whether users trust the polish of the whole product.
craft-and-beauty-as-utility
As AI makes competent execution cheap, the differentiator shifts to disproportionate investment in the details that matter most.
Dill frames this via the 'seven out of ten experience becomes free' idea: rather than only using AI efficiency to move faster, she wants teams to use freed-up time to push specific moments (like a UI transition) from a 7/10 to something far higher, echoing the Atlantic essay's point that hand-cared-for work commands a premium over mass-produced baseline output.
craft-and-beauty-as-utility
Patrick Collison and Tyler Cowen's 'new aesthetic' grant call names a broader problem: modernism's minimalist vocabulary was copied without the craft that originally justified it.
Dill argues that architecture and digital product design both inherited clean, minimal forms from genuinely exceptional originals, but many later imitations became 'lifeless' and 'sterile' because the underlying care and detail were dropped, leaving only the surface simplicity.
craft-and-beauty-as-utility
Copying a feature's existing visual system can undersell what the feature actually does.
When Stripe launched financial accounts, the initial design reused standard dashboard form fields and drawer interactions. A product lead flagged in review that the UI 'doesn't feel as magical as this product actually is' - the team then redesigned it around fluid, direct-manipulation containers that better matched the underlying capability of moving money between currencies and destinations.
design-leadership-at-scale
High standards mean being willing to pull already-shipped work rather than let mediocrity stand.
Dill says Collison has asked for changes to billboards that were already live, and that she'd rather have the hard conversation of stopping or redoing something than ship what the team knows isn't great, even when a team has already invested significant time and morale in it.
design-leadership-at-scale
Design leadership at scale means building the system and culture that produces great work, not personally touching every project.
Dill describes her role shifting over five years from being intimately involved in individual projects (which she now considers a possible failure state) to shaping the conditions and community that let good work happen without her, while deliberately staying close to users and to the tools her team uses daily so she doesn't lose touch.
design-leadership-at-scale
In hiring, curiosity now specifically means genuine, hands-on engagement with fast-moving AI tools, not general intellectual interest.
Dill doesn't expect candidates to have deep AI tenure, since the field moves too fast for anyone to have mastered it, but she looks for demonstrated genuine experimentation because AI product thinking is inherently non-deterministic and requires comfort with ambiguity that spills into how someone thinks about users generally.
hiring-for-curiosity-and-agency
Agency is the other half of curiosity: acting on an idea rather than just having one.
Dill cites internal examples - one engineer built Proto Dash on his own time after noticing a tooling gap, and a content designer redesigned platform-business navigation after recognizing Stripe's UI was built around single entities rather than complex platform businesses like Shopify - as the bottoms-up pattern she wants more of.
hiring-for-curiosity-and-agency
Beauty and fine detail are treated as functional signals of trustworthiness, not decoration, especially for a company handling money.
Dill points to small touches like animated dividing lines and hero imagery built from a designer's own hiking photos as deliberate 'moments of magic' that communicate care and technical craft, arguing this matters concretely for Stripe because partners are trusting the company with their money and business.
craft-and-beauty-as-utility

Media referenced

Companies

Techniques and frameworks

Summary

Katie Dill, Stripe's head of design, walks through the redesigned Stripe.com and the broader shift in how design works at the company as AI reshapes the practice. The conversation opens on specific choices in the new site - a sentence-length H1 instead of a shrinking-with-scale generic tagline, a non-literal "data viz" section built from a designer's own hiking photographs, and animated dividing lines that Dill calls her favorite detail that "totally does not have to exist." These choices anchor a recurring theme: beauty and fine detail aren't decoration for Dill, they're functional signals of care and trustworthiness, which matters more for a company that moves people's money.

The episode's structural argument is that AI is raising the floor - baseline, competent execution is becoming fast and nearly free - which means the real differentiator is what teams do with the freed-up time and effort. Dill repeatedly returns to a "raise the ceiling" framing: rather than only using AI to ship more, she wants Stripe to point outsized craft at the specific moments that matter, echoing a 1927 Atlantic essay she cites on how machine-age mass production made hand-cared-for objects (her Hermes-versus-Zara example) command a premium. She connects this to Patrick Collison and Tyler Cowen's public call for grants exploring "new aesthetics," framing today's minimalist, often sterile digital and architectural design as a diluted copy of genuinely exceptional modernist originals that dropped the underlying craft.

Practically, this plays out through Stripe's internal AI prototyping tool, Proto Dash, which lets designers, PMs, and engineers build dashboard-realistic prototypes rapidly and in context, replacing the old cycle of pitching an idea, finding engineers, and waiting weeks for a mockup. Dill gives a concrete example: Stripe's financial-accounts launch initially reused standard dashboard form fields, and only after a product lead noted in review that the UI "doesn't feel as magical as this product actually is" did the team rebuild it around fluid, direct-manipulation interactions that better matched the feature's actual capability. She's also candid that AI-generated visuals are rarely ship-ready out of the box - the team's marketing imagery required manual fixes to lighting and shadows in Photoshop before it read as real.

On hiring and team evolution, Dill says curiosity and agency remain her top signals, but curiosity now specifically means hands-on engagement with fast-moving AI tools (not general intellectual interest), and agency means building something rather than waiting for permission - she cites an engineer who built Proto Dash on his own initiative and a content designer who redesigned platform-business navigation after spotting a gap. She also describes her own role shifting from being personally involved in most projects (which she now calls a potential failure state) to building the systems and culture that let good work happen without her, while staying grounded by personally using the same AI tools her team uses, including "vibe coding" her own side projects on weekends.

Throughout, Dill pairs a demand for creative risk-taking with an unwillingness to ship mediocre work - she describes Collison asking for changes to billboards that were already live, and says she'd rather have a hard conversation about pulling something than let a team ship what everyone privately knows isn't great. The episode closes with Dill's personal product stack: Framer for websites, Granola for crit notes, Jitter for animation, Lovable for building ideas in code, Mobbin for design inspiration, Paper for creative design work, and Raycast as her everyday shortcut tool.

Notable Quotes

"I feel like there's no more of the oh, you know what would be great if... I got to go find some couple of people I'm going to have to convince them of this so they can build the prototype. It's just like I had an idea for this thing. So here it is, here's the link, go check it out, what do you think?" - Katie Dill

"As things get easier to make, people are aware that it is easier to make things that are the seven out of ten, the baseline experience." - Rid

"The cost of the seven out of ten experience is going to be borderline free in the long run... what if we just took five times the amount of resources and time and creativity and just pointed it directly at the way that these panels move?" - Rid

"It's more about how do I help put the conditions in place where this is an organization where great work gets done. I am not often the right person to make the great work happen." - Katie Dill

"We do some serious work... if there are bugs in our product or we have a typo, that is a big deal because we move people's money and our partners rely on us to do the work well." - Katie Dill