Empathy, Half-Built Products and Staying Power: Scott Belsky on Product Sense in the Age of AI

Belsky explains that his long run as chief product officer at Adobe rested on a belief that the boundaries between product, design and engineering are at best artificial and at worst constraining. In his view many products win not because of the technology but because of the user's experience of it, so he oversaw all three functions together.

Asked how he built his own product sense, he says the biggest mistake teams make is falling in love with a solution instead of developing empathy for the person suffering the problem. Empathy often hands you the solution, while passion for a guessed solution may be "30 degrees off". The most humbling moments of his career came from sitting shoulder to shoulder with customers and watching how they go about their whole day, not just how they use the product.

That context matters because people use your product amid meetings, pings, children and Netflix. Understanding it lets a maker feel what the customer feels and, in effect, become the customer, which is where many of the best products come from.

Belsky describes the first 30 seconds of any product as a moment when users are lazy, vain and selfish. They want it done fast, want to look good to colleagues or friends, and do not want to read or watch anything. If a team gets people through that moment, he says, it earns the chance to build a lasting relationship, yet most teams spend only their last stretch of effort on the first mile. Lenny clarifies that this covers onboarding, orientation and defaults.

Even at Adobe's scale he sees plenty of opportunity there, because every cohort of customers is different. Early adopters are forgiving, while later pragmatists are sceptical and unwilling to tolerate friction. Photoshop, which once cost hundreds of dollars and now starts around ten dollars a month, draws many more people who have the desire but not the skills, which forces the onboarding to be redesigned. He hopes AI will let products eventually meet each user where they are.

On growth, he notes that people rarely talk about a product doing what they expected. They talk about the unexpected, such as the Easter egg a Tesla owner found on the dashboard, so he asks why teams do not deliberately optimize for surprise and delight.

Lenny suggests that new consumer products almost never last, and Belsky mostly agrees. The winners he points to had something structural: Uber built a network effect from excess capacity that was always there but untapped, and Pinterest, where he was the first seed angel and product advisor, offered a new insight into consumer psychology by letting people collect and represent their interests rather than chase likes.

Pinterest also drove heavy traffic to the source of every pin, which led those sites to add pin buttons themselves. Belsky contrasts this with many recent creative consumer apps, which he describes as clever momentary interfaces that effectively act as R&D funded by venture capital for platforms that already have network effects and distribution. When such features work, they get absorbed by TikTok, Instagram or the native Apple camera.

He adds that his answer has changed over a decade. After super apps in China and a more technologically literate public, people tolerate more tabs and hidden features, which is one reason incumbents can simply copy novel capabilities instead of letting them become standalone apps.

Asked how to decide what goes into an MVP, Belsky says to optimize for the problems you want to have. You want customers to get through the funnel, feel successful and then complain that they need another platform or a sharing feature. So only build what prevents them from reaching that point, such as signup, account connection and Google login, and postpone everything else.

He learned the lesson at Behance in 2008, when he hedged by launching groups, a tip exchange, portfolios and work-in-progress sharing all at once. When the team killed the Tip Exchange, project publishing rose, and when it killed groups, publishing rose again. Making the product about one thing made the core crank run at roughly ten times the velocity, and he now asks teams adding a feature what they can replace or remove.

Removing the colour controls on portfolios brought 24 hours of angry messages and then silence, while portfolios looked cleaner and more projects were published. Belsky says the best product leaders he has worked with are naturally reductionist, and Lenny likens it to advice that teams move faster once fewer people are on them. Belsky agrees that tough decisions usually feel like a relief afterwards.

Belsky calls himself an optimist and says human potential has been held back by mundane, repetitive labour, the "work" in workflow. AI, he argues, moves us towards flow. He recounts a conversation with Airtable's Howie about an IBM leader who said he would not hire about 8,000 people because AI could do the work. Their counterpoint was that more productive engineers have led companies to demand more of them, so firms with three products might build five, seven or thirty.

He adds that tools like ChatGPT mostly mine and regurgitate the centre rather than finding the edges that become the centre, and says he is "long the experience economy". He expects more people to focus on non-scalable work that moves the needle for customers, with less grunt work for everyone.

On organisations, he sees AI collapsing the stack. People will answer their own data questions instead of waiting on an analyst, and cross-functional work will become easier, much as in small flat teams that outrun large dispersed companies. He suggests it could also push companies towards promoting on ingenuity, while analysts are freed from redundant requests.

For product managers, Belsky says great designers and product leaders protect time to explore many possibilities, cull them, get feedback and present two or three. Generative AI should expand that surface area. He cites a Hollywood director who asks ChatGPT for five scenarios and uses none, because they show him what he does not want.

At Adobe, Belsky says, one obsession has been getting designers into the room for customer research and debates about the value proposition, something that traditionally happened only with PMs. He calls this the cheat code of his career. He also describes Photoshop experiments with generative extension of images, and imagines designers getting predictive thumbnails that let them take leaps rather than incremental steps.

He defines the "golden gut" as the stream of "I wonder if" instincts while designing a flow: splitting a page into three steps, removing steps entirely or using a presumptuous default. In his view that separates junior from experienced product thinkers. An experienced one may accept that 10% of people get confused if 90% get through far faster.

His advice for product managers worried about AI is one word: play. Experience makes people set in their ways, so they should try the tools and pursue every curiosity. Belsky started his Implications newsletter to force himself to understand the fast-moving field. He and Lenny agree that writing helps you think, and that Belsky protects slivers of time, such as 40 minutes after a run, to write.

The years in the middle of a venture are messy, Belsky says, because they are full of lows: anonymity, uncertainty and anxiety. He notes that human life expectancy was about 25 only a century or so ago, so we are biologically wired to want constant rewards and affirmation. To endure, teams need a culture they care about and a way to short-circuit the reward system with micro goals and milestones that everyone agrees to celebrate.

He likens a product leader to someone driving a team across the country with the back windows blacked out. If people cannot see that the traffic is clearing and they just crossed a state line, they go stir-crazy, so leaders must "merchandise progress" and be the steward of the narrative.

On when to quit, he asks founders how much conviction they have now, compared with the beginning, in the problem and the solution. Those with more conviction after all the validation and failed attempts are simply in the messy middle and should stick with it. Those who say they would not have started if they had known are told to quit or pivot, because life is short. He cautions against making a bold decision on a bad day, but says that lost conviction means the fuel is gone.

As an angel investor, Belsky values founders who listen, learn and want to shake things up, and who care about the mission more than the money. He is wary of promoters who sugarcoat the truth, and prefers leaders who are optimistic about the future but pragmatic and somewhat pessimistic about the present, and who will share the real obstacles with their team and investors.

On the product side he looks for an "object model" that he is confident will scale. In every screen a user should be able to tell how they got there, what to do now and what to do next. When a team that claims to be design driven shows a demo with no clear breadcrumbs, he considers it a red flag. He is agnostic about sector and looks for product and design-oriented teams making things that need to exist.

For the current resource-constrained moment he says that resourcefulness brings you further than resources. Over the last seven to ten years, teams threw money at servers and headcount, while now they must refactor databases and customer service processes or use AI. He ends with the reminder that anything extraordinary is an exception, and that if everyone says you are crazy you are either crazy or onto something.

Belsky describes three waves of transformation driving Adobe's strategy. The first is creative confidence: adults lose the confidence they had as five-year-olds, and tools like Adobe Express and the generative AI offering Firefly help people feel capable right away.

The second is that creative professionals can explore ten times the surface area of possibility, which he says will not replace them but lead to solutions ten times better. The third is personalization, with apps, marketing and commerce experiences that know who we are, down to showing only our shoe size, a change that Adobe's digital marketing business is positioned to enable.

In the lightning round he recommends Tony Fadell's *Build* and books on human nature and power. He says that with Vinod Khosla's prediction that most jobs could be replaced by AI, he is encouraging his children to find passions, since taking initiative on a passion is a replicable muscle. He also names the app Queue and the presentation tool Tome, in which he invested, as recent favourites.