Design thinking, craft, and building beautiful products. Insights from designers and product leaders on Lenny's Podcast.
“One of the things that people don't easily understand is that most people come to your product with an emotional frame and a lot of people want to appeal to their logical brain right away. And the reality is, don't do that. You have to understand the mindset that people are in.”
“When I was working on user acquisition and designing registration for Facebook, that knowledge was incorporated into the design of the product in ways that I think other companies not caught onto yet. And I know for a fact that a lot of that thinking that went into designing how you think about a phone number and a device and its use among one, has been helpful for Facebook's growth.”
“I like to joke today that AI products, and it's a half joke, they're actually really hard to use because you have to be very thoughtful about when it could help you. And if you're not prompting a model to help you, it's probably not helping you at that time. And if you think of how many times the average user is prompting AI today, it's probably tens of times. But if you think of how many times people could actually get benefit from a really intelligent entity, it's thousands of times per day.”
“Opinionated software is basically software platforms that essentially have either best practices or maybe some rules integrated into the system. Greenhouse is another example from a recruiting perspective in terms of structured recruiting. It's principled, anchored in some values. And because of those values, there will be certain rules built in that you can't really work around.”
“Meeting developers where they are, integrating with their tools, taking security to them instead of pulling them out of their workflows. Flow is just this incredibly important concept for developers. You want to strive to keep them in that flow for as long as possible.”
“The real world has entropy and it's hard and it's messy. Computers are deterministic, but humans aren't. And so building products that have a little bit more flex or a little bit more fail safes in case those things happen becomes a little bit more paramount.”
“We strive for this concept of what we call perceived simplicity, which is there are advanced features in the product and they are easily discoverable when you look for them, but they're effectively hidden if you're not looking for them.”
“Build for your best user, not your worst user. It's actually really easy to get stuck focusing on abuse, or all the ways in which the product won't be used well. And then you end up shaping the product in really weird funky ways to make up for that. The worst users should be a fraction of your users anyway.”
When building algorithmic products, PMs must define what algorithms should handle versus what requires human judgment -- algorithms optimize but lack understanding of long-term effects and user intent.
Adriel FrederickOpenAI's extreme velocity comes from combining top talent density with radical bottoms-up autonomy - most companies cannot simply copy this model
Alexander EmbiricosCategory creation is only worth pursuing when your product's scope far exceeds existing categories -- if buyers already have budget and language for what you do, elevate the existing category instead of inventing a new one.
Barbra GagoProduct and operations teams function best as a twin turbine engine — they need mutual respect and a strong bidirectional feedback loop to maximize efficiency
Brian TolkinUnder-communicating upward is a common career limiter. Executives need context on challenges and trade-offs to support you effectively—proactively share what's blocking you rather than silently struggling.
Casey Winters 1.0Stripe's success came not from one or two pivotal decisions, but from a culture that consistently produced many good decisions through rigorous first-principles thinking and a strong writing culture.
Eeke de MillianoDesign your UI to match the actual performance of your algorithms -- fault-tolerant interfaces with escape hatches are essential because prediction accuracy is never 100%, and users need graceful fallbacks when recommendations miss.
Gustav SöderströmAI will not replace developers but will transform how they think - junior developers can now focus on systems architecture from the start instead of spending years learning basic code syntax
Inbal S