The interviewer introduces Scott Belsky, highlighting his impressive career and influence in the tech and design world. Scott is described as a builder, author, investor, and "maybe the world's most famous designer" who founded Behance in 2006 as an "OG New York Tech company."

The introduction emphasizes Scott's journey from founding Behance to becoming Adobe's Chief Product and now Chief Strategy Officer, as well as his success as one of New York's most prolific angel investors and bestselling author of two books.

"I think I'm motivated by frustration. As I look back at all of the experiences that I've had, I feel like most of them were inspired by a sense of frustration."

Scott explains that his work has consistently been driven by identifying problems and seeking solutions—Behance was inspired by frustration with the disorganization of the creative world, while his book "The Messy Middle" emerged from noticing how people focus on beginnings and endings but neglect the crucial middle part of journeys.

Timestamp: [0:00-1:32]

Scott shares his career philosophy, focusing on finding the sweet spot where skills, interests, and opportunities converge.

"I've always been obsessed with the intersection of design, technology, and business really, and I've been always trying to find in my career a way of getting closer and closer to that intersection."

When asked for career advice, Scott consistently recommends that people look for the overlap of three key elements:

1. Skills they have or can easily develop
2. Genuine interests that captivate their attention and keep them engaged
3. Opportunities that present themselves

Scott emphasizes that this convergence of skills, passions, and opportunities has been the guiding principle throughout his career journey.

Timestamp: [1:32-2:41]

When asked about building a tech company in New York back in 2006 and what has evolved in the ecosystem since then, Scott expresses his enthusiasm for the city and highlights two critical advantages that make New York uniquely positioned for today's tech landscape.

"As more and more skill is being offloaded to compute... I think that taste is going to become more important than skill. And then that begs the question of well, how do you develop taste? And I feel like taste comes from culture."

Scott explains that as AI and technology democratize technical skills, taste becomes the differentiator. He argues that taste development requires cultural exposure—art, fashion, diverse industries—making New York an ideal environment for cultivating this crucial quality.

His second point addresses how to navigate the rapid pace of technological displacement, where new innovations constantly replace previous solutions in quick succession. Scott uses the example of competing video models (Runway, Sora, Pika, Luma) to illustrate this "giant game of slap a hand" where technologies rapidly supplant one another.

"The answer is always to double down on customer empathy. If you can zero in... when you really go back to spending time with what the customers are struggling with and thinking about it, it does give you a bit of like a compass or a North Star."

New York's advantage lies in access to diverse customers across industries like healthcare, marketing, real estate, and finance—all within a three-mile radius. This proximity to customers, combined with rich cultural resources, positions New York ideally for building customer-focused technologies.

Timestamp: [2:41-5:56]

The conversation turns to Behance's early days, which took six years to build before being acquired by Adobe. Scott discusses the importance of startups doing "things that don't scale" as a competitive advantage.

"One of my favorite things to think about with new companies is that they will do things that don't scale and are remarkably unscalable in the beginning. And by the way, that's the competitive advantage in some ways because big companies don't do things that don't scale."

Scott explains that large companies avoid unscalable activities because they can't get executive approval, resources, or airtime for such approaches. This creates an opportunity for startups to break through by embracing labor-intensive but effective methods that larger competitors won't attempt.

For Behance, the mission was to "organize the creative world at work," inspired by seeing talented creative friends who:

• Lived their careers "at the mercy of circumstance"
• Never received proper attribution for their work
• Watched agencies take credit and awards for their contributions
• Maintained outdated portfolios because they lacked time and motivation

Scott believed this system was fundamentally broken and assembled a team to address these problems.

Timestamp: [5:56-7:41]

Scott shares how an early focus group for Behance nearly derailed their vision when participants rejected the idea of "another social network" alongside MySpace, DeviantArt, and LinkedIn. This taught Scott a valuable lesson about customer research.

"Don't ask customers what they think of your product or your idea. Ask them what they're struggling with."

When they shifted their questions to focus on pain points, they discovered that creative professionals struggled with attribution, self-marketing, and portfolio management—confirming the need for a professional platform to showcase work.

However, when they approached top creatives to join the platform, many declined due to time constraints. This led to a brilliant "non-scalable hack":

"We created a blog and we went to the same group of creatives that we really admired that we wanted everyone to see were on Behance... we want to interview you for a blog and they're like 'oh I'd love to be on your blog'... and we said 'oh and please send us examples of your work and we'll make a portfolio for you for the blog' and everyone did it."

This strategy allowed Behance to:

1. Build 100 high-quality portfolios for the most admired creative professionals
2. Launch with 300-400 impressive projects (since each portfolio had 3-4 projects)
3. Create the illusion of a robust, active platform from day one

This unscalable approach provided the critical mass needed to attract other users, demonstrating how startups can use creative methods to overcome initial adoption challenges.

Timestamp: [7:41-9:55]

• Success often stems from frustration—Scott's major ventures were inspired by identifying problems that needed solving
• Career fulfillment comes from finding the intersection of skills, interests, and opportunities
• As AI democratizes technical skills, taste becomes more important—and taste is developed through cultural exposure
• New York offers unique advantages for tech companies through cultural richness and proximity to diverse customers
• Customer empathy provides a "North Star" during rapid technological change
• Startups gain competitive advantage by doing "things that don't scale" that larger companies won't attempt
• When researching product ideas, ask customers about their struggles rather than their opinions on your solution
• Creative "hacks" like Behance's portfolio-building approach can overcome initial adoption challenges

Timestamp: [0:00-9:55]

Books:

• The Messy Middle - Scott's book focusing on the crucial but often overlooked middle part of the entrepreneurial journey

Companies/Platforms:

• Behance - Creative portfolio platform founded by Scott in 2006, later acquired by Adobe
• Adobe - Company where Scott serves as Chief Strategy Officer
• MySpace - Referenced as an existing social network when Behance was launching
• DeviantArt - Referenced as an existing creative platform when Behance was launching
• LinkedIn - Referenced as an emerging professional network when Behance was launching

AI/Tech Companies:

• Runway - Mentioned as part of the rapidly evolving AI video generation landscape
• Sora - Referenced in discussion of rapidly changing AI video models
• Pika - Referenced in discussion of rapidly changing AI video models
• Luma - Referenced in discussion of rapidly changing AI video models

Concepts:

• Law of Displacement Speed - Scott's term for how quickly new technologies replace previous solutions during platform shifts
• Customer Empathy - Emphasized as the compass needed during rapid technological change

Timestamp: [0:00-9:55]

Scott discusses what makes Adobe unique in its approach to acquisitions, highlighting how it's kept products like Behance and Frame thriving, unlike other companies that often discard acquired products.

"The True Unicorn in sort of big companies is successful M&A and that's been a real focus of mine for seven years now, because it's how companies change."

Scott explains that Adobe was itself built through significant acquisitions:

• Macromedia - which enhanced Adobe's web capabilities
• Omniture - which built Adobe's digital marketing capabilities

The company has a history of not just acquiring businesses that transform the company but also elevating acquired talent to leadership positions. Many current Adobe leaders joined through acquisitions, creating a culture that embraces outside perspectives.

"I think we're particularly good at allowing people to change us and coming in with better ideas as opposed to some companies like Google and others that are kind of known for acquiring teams and being like 'forget everything you knew, you're going to do it the Google way.' We kind of do the opposite."

Scott also notes that Adobe attracts people who are genuinely passionate about specific domains like photography, 3D technology, or immersive experiences. This specialist passion helps with talent retention after acquisitions.

Timestamp: [10:01-12:06]

When asked whether acquired founders stay in "founder mode" or transition to "manager mode," Scott gets visibly energized by the question. He reflects on the challenge of maintaining founder-like thinking within a large organization.

"The fantasy I have sometimes of totally going into founder mode is hard to execute because you have so many stakeholders and so many opinions and so much process."

Scott reframes the concept of "founder mode" as being less about the person's title and more about their approach to communication and decision-making:

"I think founder mode has less to do with whether it's the founder and more to do with whether people are being direct and truthful. I love when people say what they feel, and when you can have a culture where people can actually speak up and be like 'I don't buy this' or 'I would never use this.'"

He contrasts this honest, direct approach with the tendency for teams to follow the "happy path"—an idealized scenario where users interact with a product exactly as designed, which rarely happens in reality.

"In the real world, the happy path doesn't exist. Every customer in some ways is an exception."

For Scott, the essence of founder mode includes:

• Directness and transparency
• Making decisions that prioritize long-term benefits over short-term comfort
• Fighting against the pressure to show immediate results for long-term bets

This becomes particularly challenging in a public company environment where quarterly results and annual plans often conflict with multi-year innovation horizons.

Timestamp: [12:06-14:14]

The conversation shifts to how large companies like Adobe approach AI adoption compared to startups, with Scott offering insights from his unique perspective working both at Adobe and supporting hundreds of startups.

Scott first identifies the paradoxical advantage startups have in not having an existing customer base:

"The thing that startups don't have yet, that is an advantage in some ways, is customers. Startups can do things any way they want, and they don't have to go back and explain and bring along a community of customers that are trying to adapt to change."

He illustrates this challenge through Adobe's experience transitioning Lightroom to the cloud:

"There were so many changes that we needed to make to Lightroom to make the new version that would work in the cloud and on web and on mobile that we actually had to fork the product. We called the previous one Lightroom Classic, which remains a huge business for us, and we have this new business called Lightroom, which is also a huge business."

This dual product approach—maintaining the legacy product while developing a new version—is "really expensive" and not something startups would typically consider. However, Scott notes that established companies have the advantage of helping existing customers navigate technological shifts, as Adobe did when transitioning graphic designers to web design.

Another burden large companies face is the need for more rigorous development approaches:

"A lot of startups are using these open source models that are trained on the open internet, and you can get something to market in a week. We actually had to train our own model using all licensed data in a very clean and transparent way with all the safety guards and everything to allow us to do it the right way. That took a lot longer."

Timestamp: [14:14-17:11]

Scott shares his perspective on specific areas where startups can effectively compete against large companies in the AI space.

"I definitely subscribe to the idea that a lot of great consumer products—novelty precedes utility. I think anyone who's building AI in the consumer space needs to have a real playful, imaginative approach where it's all about finding something that people love. Big companies will not do that."

He believes large companies don't sufficiently prioritize "surprise and delight," instead focusing on features, roadmaps, and business cadences. This creates space for startups to bring more imaginative approaches to market.

Scott also sees opportunities in disrupting incumbents with outdated approaches:

"Look at government tech, look at companies that do mapping technologies or transcription or all these sorts of things. It's just very hard for them to transition their customers, just like the drama I just described that we deal with at Adobe. Startups have an opportunity to just start fresh and do it the way you imagine it should be done."

However, he offers strategic advice for startups to differentiate themselves:

"Focus very much on differentiation, not on just the interface layer but also the data layer. Most of you will not be able to differentiate on the model layer because you're going to leverage models that are being built and available. I don't encourage many folks to build their own model these days."

Instead, Scott recommends startups focus on two key areas of differentiation:

1. Interface Layer: "How can you be simpler? How can you really take into account the workflows of your customer and the struggles of your customer and do something on the interface level that your competitor never could or would?"

2. Data Layer: "How can you either import or capture or help the customer leverage data of their own that you then can deliver an experience no one else can because you have that data sort of advantage?"

Timestamp: [17:16-19:13]

The interviewer connects Scott's insights to his first book and asks how consumer founders can transition from having a clever interface that hooks users initially to building a sustainable platform or product.

Scott identifies this as "probably one of the most important questions in product" and introduces the concept of the "first mile experience":

"Every product has what we call a first mile experience, which is the part of your product that the most customers will see, and it's all drop off from there."

He begins to explain that successful products must empathize with where customers are in the first 30 seconds of their experience, whether they're consumers or enterprise customers.

Timestamp: [19:13-19:58]

• Successful acquisitions are rare in large companies; Adobe stands out by embracing acquired talent and allowing them to influence company direction
• "Founder mode" is about directness and truthfulness rather than title—having people speak candidly about products
• The "happy path" fallacy: Companies design for ideal user journeys, but in reality "every customer is an exception"
• Large companies struggle with bringing existing customers along during technological shifts, sometimes requiring parallel product versions (like Lightroom and Lightroom Classic)
• For AI consumer products, "novelty precedes utility"—startups should focus on playful, imaginative approaches that big companies won't prioritize
• Startups should focus differentiation on the interface layer (workflows and simplicity) and data layer (unique data advantages)
• Building sustainable products requires careful attention to the "first mile experience" and understanding customer context in their first 30 seconds

Timestamp: [10:01-19:58]

Companies:

• Adobe - Where Scott serves as Chief Strategy Officer
• Macromedia - Acquired by Adobe to enhance web capabilities
• Omniture - Acquired by Adobe to build digital marketing capabilities
• Google - Referenced as having a different acquisition integration approach than Adobe
• Apple - Mentioned as limiting how deep specialists can go in their domains

Products:

• Lightroom - Adobe's cloud-based photography product
• Lightroom Classic - The legacy version of Lightroom that Adobe maintains in parallel

Concepts:

• First Mile Experience - The initial part of a product that most customers see, where drop-off is highest
• Happy Path - The idealized scenario where users interact with a product exactly as designed
• Founder Mode - Scott's concept of directness, transparency, and long-term thinking
• Interface Layer - One area where startups can differentiate through simplicity and workflow optimization
• Data Layer - Another area where startups can create unique advantages through proprietary data
• Model Layer - The AI foundation that Scott suggests most startups should leverage rather than build

Timestamp: [10:01-19:58]

Scott continues his explanation of the "first mile experience" of products, offering insights into user psychology and why this initial interaction is so critically important yet often undervalued.

"I guarantee you they're lazy, vain, and selfish. They want to get through it fast, they want to look good to their boss, they want to look good to their friends, they want to feel good about themselves. There needs to be some quick hit of feeling successful in that first mile in order for them to engage further."

Scott equates this quick feeling of success with "product-led growth," where the product experience itself drives adoption. He points out a common development paradox: the onboarding experience that every single user will encounter is often treated as an afterthought.

"It's the final mile before you launch where you're like 'oh wait, what should the onboarding be?' and 'what should the splash page?' 'oh and what should those defaults be?' That's like a happenstance conversation towards the end of shipping, when in fact that's the only thing that every customer will ever see, and it's all downward from there."

Scott makes a provocative claim that underscores the importance of this first impression:

"If you can really just nail the first mile experience of your product and after that it's all kind of shitty, you're probably one of the top 1% products out there, really, if you can just get people through."

He goes on to describe a pattern in product evolution that leads to deterioration of that first-mile experience:

1. Companies create simple, thoughtful interfaces that attract users from competitors
2. As they grow, they begin listening more to power users than new users
3. Investor pressure for revenue leads to more complexity and enterprise features
4. The product adds complexity, choices, and processes, neglecting the first-mile experience
5. Meanwhile, the user base shifts from early adopters to pragmatists who need a different onboarding experience

Scott's recommendation is simple but powerful:

"The short hack to bottom line this is to spend consistent time forever on that first mile experience of your product."

Timestamp: [20:04-23:05]

The conversation shifts to Scott's second book, "The Messy Middle," which focuses on the challenging period between a company's exciting beginning and its eventual success or failure. The interviewer frames this in the context of companies that raised significant funding during the zero-interest-rate environment and are now struggling with flattened growth.

Scott defines the messy middle:

"The messy middle is when you feel like you're really just working amidst constant anxiety, ambiguity, uncertainty, and oftentimes anonymity—like no one even knows or cares what you're doing. And that can go for some period of time."

He expresses particular concern for companies in a specific predicament:

"I actually feel most bad for the companies that are profitable with low growth because they're just in this middle land where there's no end in sight. People are getting impatient, the founders want to move on, but it's working well enough where you don't have to shut it down. No one wants to shut down a profitable entity, and it just kind of meanders—the zombie company metaphor."

Scott estimates there are "probably a few thousand zombie companies right now" with substantial revenue ($10-30+ million annually) but "no end in sight."

For teams navigating the messy middle, Scott offers several strategies:

1. Focus on endurance: "One of the greatest competitive advantages of startups is simply sticking together long enough to figure it out."

2. Build genuine culture: Foster "genuine affection for one another," celebrate small wins, and recognize non-monetary accomplishments.

3. Merchandise progress: Leaders need to continually narrate the journey and progress to their teams.

"Without your voice telling this story that people are on, it's almost like your team is in the back seat with the windows blacked out driving cross-country, and you have no idea if you're sitting in traffic on the Brooklyn Bridge or if you're actually making progress in your journey."

4. Take bold actions: For zombie companies especially, Scott encourages bold moves rather than continuing in stasis.

For founders struggling with whether to persist or quit, Scott offers a simple framework:

"You had a lot of conviction in the idea and in the end state of the world should you be successful in the beginning. Then you went and learned a lot, a lot of truth. From what you've learned, have you gained or lost conviction? If you have gained conviction but you're still struggling and you're still in the messy middle, it's par for the course... But if you've lost conviction, there is just no pride in continuing. Quit."

Timestamp: [23:05-27:03]

The conversation turns to AI's impact on design, with the interviewer asking Scott to explain the concept of "AI knobs" he has previously discussed.

Scott explains this concept in the context of how algorithmic feeds work in social networks and content platforms:

"Everything we see now is based on this interest graph—it's no longer a follower/friend graph really. And the way those graphs work is that they are trying to tell us things, show us things that are always kind of bringing us to the edge of what we think or know or are fascinated by."

He describes how this can be beneficial with benign interests:

"This is a wonderful thing if you have a good interest like soccer. Then you'll find something that tells you about a soccer team that you didn't know as much about, and then you'll go deeper and find something about that soccer player you didn't know about. It's bringing you to the edge of what you knew, and that's a successful use of these algorithms."

However, Scott highlights the danger when this same dynamic applies to more sensitive topics:

"When applied to things like politics or conspiracy theories or other things like that, it's really dangerous because these algorithms are constantly bringing us to the edge of what's truth."

He explains the engagement mechanics behind this problem:

"If you just got pushed back at you the mainstream narrative of both parties, it would be boring and you wouldn't engage. So these algorithms say, 'Oh, but when I throw in something that this crazy person said, everyone clicks on it. Oh, let's give them more of that.' And then before you know it, they're a conspiracy theorist."

Scott reveals that "product leaders are in control of what I like to call the knobs in these algorithms" that determine optimization for conversion, retention, and engagement. He acknowledges the complexity of the problem, as the same algorithmic approach can be appropriate in one domain (sports) but harmful in another (politics).

When asked whether consumers should have more control over these algorithmic "knobs," Scott suggests greater transparency would be helpful:

"I think we should have consumers have a better understanding of why they're seeing what they're seeing. Imagine when you saw conspiracy content, it sort of said, 'Well, you're seeing this because people like you clicked on stuff like this.'"

Scott concludes by acknowledging this is an evolving area where he's still developing his thinking, but emphasizes the emerging ethical responsibilities for product leaders in the AI era:

"There's a whole new set of responsibilities that product leaders have in the age of AI that are very much like the earlier thoughts around designer code of ethics or even the notion of the Hippocratic oath for doctors. There's something that we need to figure out that we haven't yet."

Timestamp: [27:03-29:59]

• Users are "lazy, vain, and selfish" in their first product interaction—they need a quick win to continue engaging
• The "first mile" of a product is often treated as an afterthought, despite being the only part every user will experience
• As products grow, they often sacrifice simplicity for power-user features and monetization opportunities
• The "messy middle" of company building involves constant anxiety, ambiguity, uncertainty, and anonymity
• "Zombie companies" with modest profitability but low growth present a particularly difficult situation with no clear exit
• Team endurance through the messy middle is a competitive advantage—cultivate culture, celebrate small wins, and narrate progress
• The "conviction test" helps founders decide whether to persevere or quit: if you've lost conviction despite what you've learned, it's time to quit
• Content algorithms bring users to the "edge" of their interests, which is beneficial for harmless topics but dangerous for politics or conspiracy theories
• Product leaders control algorithmic "knobs" that optimize for engagement but may have unintended social consequences
• The AI era demands a new ethical framework for product leaders similar to medical ethics or design ethics

Timestamp: [20:04-29:59]

Books:

• The Messy Middle - Scott's book about navigating the challenging middle phase of building companies and products

Concepts:

• First Mile Experience - The initial part of a product that every user encounters, critical for engagement
• Product-Led Growth - Growth strategy where the product experience itself drives user acquisition and retention
• Conviction Test - Scott's framework for deciding whether to continue with a struggling venture
• Interest Graph - Modern algorithmic content selection based on user interests rather than social connections
• Follower/Friend Graph - Traditional social media organization based on explicit connections
• AI Knobs - Scott's concept for the controllable parameters in algorithmic content systems
• Zombie Company - Profitable businesses with low growth that continue without a clear exit strategy
• Hippocratic Oath - Medical ethics framework referenced as a potential model for AI product ethics
• Designer Code of Ethics - Ethical frameworks for designers referenced as analogous to needed AI ethics

Timestamp: [20:04-29:59]

The conversation turns to the controversial question of whether AI will democratize or commoditize the field of design, and whether it will eliminate jobs or enhance them.

Scott begins with a bold prediction:

"I think that the prompt-based era of creativity is almost over. I think that it was a very trite kind of low-hanging fruit way of leveraging this technology where you say 'I want a sunset' and boom, you get a sunset."

He suggests the next generation of creative AI tools will be control-based rather than prompt-based, reintroducing the same kinds of creative controls designers are familiar with from traditional tools:

"The next generation of these tools are very—as opposed to being prompt-based—they're control-based. So in some ways, they're bringing back the same levers and chisels, if you will, that people have used in products like Photoshop and Illustrator in the past. They're just AI-based."

Scott argues that AI is having two major effects on creative fields:

1. Lowering the floor: Making creative capabilities accessible to more people

   "We can't deny the fact that someone who was an accountant who always wanted to show her analysis with graphs but never had the design sensibility to do so or the skills, no longer has to go and find a designer to do it. She can do it herself. That's a very empowering thing."

2. Raising the ceiling: Enabling professionals to accomplish more sophisticated work

   "I see creators whose job was to test color palettes and packaging for a consumer product, and they would spend three weeks doing like 12 different color palettes... and now with this tool called Generative Recolor, they can literally test a thousand in an hour."

Scott rejects the notion that this technology should be limited to preserve traditional design jobs:

"I don't subscribe to this idea that we should kill this technology so that she still has to go hire a designer. That's not fair. We should have technology lower the floor and allow more people to partake in the creative discipline."

He observes that professionals are actually doing "higher order work" with these tools, citing examples of packaging designers creating more variations and video editors using generative extend to avoid costly reshoots.

Scott acknowledges the tension between those concerned about career implications and those excited about the possibilities, describing himself as caught in the "crossfire" between these perspectives.

Timestamp: [30:04-33:26]

When asked how he would build Behance differently today in the age of AI, Scott offers a reassuring perspective on what remains constant in creative careers.

"I think that at the end of the day, people hire for skill and taste, and experience. I don't think that's ever going to change."

Scott suggests that the fundamental purpose of Behance—helping creative professionals showcase their capabilities and receive attribution for their work—remains relevant regardless of technological shifts:

"I think that people want attribution for the work that they do. I think they want to be able to show what they're capable of, and that's why people get opportunity."

However, he expresses concern about intellectual property and creative style theft:

"I am concerned about people leveraging other people's styles without permission."

Scott acknowledges the long tradition of creative influence and inspiration:

"Creativity is the world's greatest recycling program. Everyone's always—everything that is an input is an output. Everything's constantly being recycled, and every legitimate artist will admit that they were very deeply influenced by other artists."

Yet he draws a distinction between influence and direct appropriation:

"When you actually take someone's work and build a model around it and then say 'I want to do this in the style of so-and-so,' I think that's something where people should be compensated."

Scott reveals that his team is advocating for legislation called "The FAIR Act" against AI impersonation, which would address cases where an AI system is trained on and mimics a specific person's creative style.

He also hints at a potential future where Behance could function as a "marketplace for styles" where creators could monetize their distinctive approaches, allowing them to "make money while they sleep" by having their styles licensed through the platform.

Timestamp: [33:26-35:14]

The interviewer follows up on Scott's earlier point about taste becoming more important as technical skills become more accessible, asking the key question: "Can you teach taste? How can someone learn taste?"

Scott acknowledges that this question has been on his mind and is particularly relevant to educators trying to prepare students for future careers:

"I've been meeting with educators in the creative space, but also K through 12 educators, and they're thinking about how do I future-proof my kids, the students, for productive careers. And then they hear someone like me say that taste is going to be more important than ever, and they're like 'Thanks, but that doesn't help me. What do I do? How do I develop taste?'"

Scott offers several thoughts on the origins of taste:

"Taste, much like creativity, comes from human experience. I think it comes from curiosity. I think it certainly comes from exposure to different disciplines, to culture."

He shares an insight from Behance's early development:

"One of the very early decisions we made in Behance was to really promote in the algorithm of what people see every day on Behance to show people stuff outside of their fields. Photographers don't just want to see other photography—they want to see architecture, they want to see graffiti, light art. They want to see all these different [disciplines] because that's where their ideas come from."

Scott emphasizes the importance of interdisciplinary exposure:

"We're all ultimately stimulated by things at the intersection of different fields and disciplines. I think that's where taste comes from."

This perspective suggests that developing taste requires deliberate exposure to diverse creative domains and cultivating curiosity that crosses traditional boundaries.

Timestamp: [35:14-36:35]

The interviewer brings up a concept Scott has discussed elsewhere: in a world of AI-powered hyper-personalization, there may be a new opportunity for "luxury software experiences." The interviewer asks Scott to define this concept.

Scott shares a personal anecdote from the early days of Uber to illustrate his point:

"A few blocks from here on 17th Street was the Behance office, very early days, 2007 maybe, maybe 2008. Garrett [Camp] had just reacquired his company from eBay, a company called StumbleUpon. He whips out this notebook and he's like, 'Oh, I'm working on this side project that helps people summon a limo car.'"

Scott recalls being surprised that Garrett was already working on a new idea while running StumbleUpon, but continues to describe how Garrett was grappling with Uber's early positioning:

"I remember in the early days he was sort of thinking about whether the brand narrative should be 'everyone's taxi' or 'everyone's personal driver'—being able to have something that is not accessible to anyone else. It was an interesting decision as a brand decision, which affected their branding and everything else. They wanted to make it like 'everyone's private driver,' this notion of something that only rich people can have."

Scott explains how this connects to our fundamental desire for personalized experiences:

"I think that we as consumers always want that. We always want special treatment. We always want to feel special. When you come into a restaurant and they remember your favorite drink or they remember your allergies, that's an amazing sensation."

He notes that the hospitality industry has long tried to create these personalized moments through "crappy software," but modern AI and agents present a new opportunity:

"We have this opportunity now with agents and modern browsers and AI and all these different tools to be known wherever we go. The fact that we go to a website that still asks our gender and what size our shoe is and all this stuff is just crazy to me."

Scott envisions a future where personal preferences follow users seamlessly across digital experiences:

"I want everyone to know my shoe size if they're in the shoe business. And I want every restaurant in the world to know that I'm a vegetarian... I want that. So I think that there's this opportunity to have a hyper-personalized world where every digital experience recognizes who we are."

He concludes with a powerful summation of this vision:

"We have this preferences profile that follows us around, and a lot of the great brand moments are feeling special as a service."

Timestamp: [36:35-39:53]

• The "prompt-based era of creativity" is giving way to "control-based" AI tools that reintroduce familiar creative controls in an AI context
• AI is simultaneously "lowering the floor" (democratizing creative capabilities) and "raising the ceiling" (enabling more sophisticated professional work)
• Creative professionals using AI tools are often doing "higher order work" rather than being replaced
• Even as technology evolves, people will continue to hire for three timeless qualities: skill, taste, and experience
• The challenge of AI and creative styles requires new approaches to attribution and compensation, potentially including a "marketplace for styles"
• Taste, which becomes more valuable as technical skills are democratized, develops through human experience, curiosity, and exposure to diverse creative disciplines
• The future of digital experiences may involve "hyper-personalization" where preferences follow users across platforms, creating "feeling special as a service"
• The tensions in Uber's early branding (between "everyone's taxi" and "everyone's personal driver") illustrate how aspirational positioning can drive product development

Timestamp: [30:04-39:59]

Companies/Products:

• Photoshop - Traditional design tool referenced when discussing control-based vs. prompt-based AI tools
• Illustrator - Traditional design tool referenced alongside Photoshop
• Behance - Scott's platform for creative professionals, discussed in context of AI and design careers
• Uber - Company Scott advised in early days, originally called "Uber Cab," used to illustrate luxury positioning
• StumbleUpon - Web discovery service founded by Garrett Camp before Uber
• eBay - Company that previously acquired StumbleUpon
• Pinterest - Mentioned briefly as a company Scott invested in
• Generative Recolor - AI tool mentioned that allows designers to test color variations rapidly
• Premier Pro - Video editing software mentioned in relation to AI's "generative extend" capability

People:

• Garrett Camp - Co-founder of Uber and founder of StumbleUpon, featured in Scott's anecdote

Concepts:

• Prompt-based creativity - The current paradigm of AI creative tools that Scott believes is ending
• Control-based AI tools - The next generation of creative AI that will reintroduce familiar creative controls
• The FAIR Act - Proposed legislation against AI impersonation that Scott is advocating for
• Marketplace for Styles - Scott's concept for how Behance might evolve to let creators monetize their styles
• Luxury Software Experiences - Personalized digital interactions that make users feel special and recognized
• Feeling Special as a Service - Scott's description of the aspirational quality of personalized experiences

Timestamp: [30:04-39:59]

The interviewer asks Scott about how people today can build the kinds of impactful networks that existed in the early days of Silicon Valley, when "the circles were very small" and everyone knew each other.

"I'm thinking back to South by Southwest, sitting in literally like the bathroom of a big hotel room with a lot of the folks you just mentioned. Gary Vaynerchuk sent me a picture of like this group—Kevin [Systrom], Garrett [Camp] was over there, and like all these people, but these were all nothing businesses, right? We were all just like a group of people who had shared interests and like minds."

Scott emphasizes that what made these early networks special wasn't just the connections but the quality of interaction:

"That was the fun part of that chemistry also was that we always debated and argued."

He offers practical advice for building meaningful professional circles today:

1. "Find the people that you really respect, that you're building things that you admire"
2. "Make a consistent effort to be each other's sort of board"
3. "Get together... debate and challenge each other"

Scott shares an amusing anecdote about Kevin Systrom (Instagram's founder) declaring at South by Southwest that he would "never sell this company" shortly before the Facebook acquisition, illustrating how these communities fostered honest, challenging conversations.

He concludes by noting that building communities like South Park Commons—where like-minded people challenge each other, respect each other, and consistently show up—is essential for fostering innovation.

Timestamp: [40:07-42:34]

During the audience Q&A, an attendee raises a thought-provoking question about how many AI products simply automate existing workflows rather than fundamentally rethinking them:

"I feel like most of the AI products I've been seeing automate existing ways of doing things without taking the opportunity to rethink if it's the right way to do it."

The questioner uses customer support ticket closure as an example—automating ticket closure provides clear ROI, but "closing a ticket is an output and not an outcome." They suggest that during a platform shift, there's a unique opportunity to fundamentally rethink workflows, even though it's "way harder."

Scott responds by sharing a framework he's been developing with his team:

"It's a pyramid as we think about agents and agent experiences and the potential of assistants of all kinds. I think the very base of that pyramid is glorified help—'here's how to do this, here's how to do that.' I think the next layer up on that pyramid is like 'let me do it for you,' so it's the automation."

He uses Photoshop as an example, where an AI could move the cursor and execute commands without the user needing to know the tools.

Scott agrees with the questioner that many AI products stop at these lower levels, but the top of the pyramid—which is what the questioner is describing—is more transformative:

"The top of that pyramid that you're referring to is the proactive—like 'actually don't do that, do this' or 'I know you're thinking of using this color palette, but based on the target demographic of your project, we actually think this color palette might work better.' Having an opinion."

Scott acknowledges this highest level is the most challenging for adoption, particularly in how you present suggestions without being presumptuous:

"As a product builder, I'm thinking, okay, how do you—you can't be presumptuous and say that this is the answer to the human who's using this tool, but what you can say is, you can present it as a tip... and it's like 'sure we'll do this for you, but these are the tips that we would suggest you consider,' and then before you know it, people just start clicking them automatically."

Timestamp: [42:34-45:25]

Another audience question comes from someone building an "orchestration layer for multimodal models" who struggles with limited control over the underlying models, making it difficult to deliver optimal creative experiences. They ask whether orchestration layers will win out over building proprietary models in the long term.

Scott offers an optimistic perspective on how this tension will likely resolve:

"I think that the models are going to become more mature, and they're going to have a lot more APIs that allow you to do lots more on a more granular level with the model. So whatever you're frustrated by not getting access to right now, you're likely to probably get more access to in the future."

He draws a parallel with the evolution of AWS, which started as a broad cloud service but progressively offered more granular capabilities as developers requested them. Scott expects the same pattern with large multimodal language models.

Regarding Adobe's approach, Scott shares their strategy:

"Publicly we've talked about the fact that we're not building our own LLMs, so we're partnering with a lot of different ones. We are building our own media models because we have to, and we believe we can build the most commercially safe, best viable option in the market."

He explains that Adobe is investing in orchestration layers across LLMs for their workflows:

"Whether it's ChatGPT or Anthropic or others—if you use an Acrobat product, there's now this thing called Acrobat Assistant that does things for you. We built an orchestration layer in the product that can leverage various LLMs to do things for you."

Scott concludes by reassuring the questioner: "The granularity is coming."

Timestamp: [45:25-47:18]

An audience member asks how to maintain the directness and transparency of "founder mode" as a company grows and adds management layers.

Scott offers three tactical approaches:

1. Collapse the talent stack wherever possible:

   "When you have an opportunity to have a designer who also has product management skills, or a designer who can also code front end, or collapse, collapse, collapse—there's something that happens, it's like a technical breakthrough with conduits or something. When that happens, like magic happens."

2. Create a culture of direct access to talent:

   "One of the things that I really try to do is when there's an amazing designer who's doing something in the organization, they want to talk directly to the leader. They want to explain why they made a particular decision. Yet in big companies, they report to someone who reports to someone who doesn't even design anymore..."

   Scott acknowledges this approach initially met resistance at Adobe:

   "It took me a couple years with my team when I came back to Adobe to explain to them that this is how I work. Initially people were like 'Whoa, whoa, why did you go to so-and-so? Why didn't you come to me? Why wasn't I even in the email? What's going on here? Am I going to get fired?'"

   Despite the initial discomfort, Scott advocates for direct connections:

   "If you are in an organization that values direct connection between people making decisions, it should be okay. And it's better for talent retention and empowerment when you can do that—go straight to the source of the talent or the information."

3. Foster a culture of directness: Scott shares a conversation with a founder whose company culture is so empathy-driven that even asking questions the wrong way could trigger defensiveness. He emphasized the importance of explicitly discussing and establishing norms around direct communication:

   "You have to have this conversation with your team. You want to have a culture where people can be very direct with one another."

Timestamp: [47:18-49:49]

The final discussion addresses the tension between privacy and personalization in the context of the "luxury software experiences" concept discussed earlier.

Scott frames the issue through a retail example:

"Take like a famous shoewear brand. You expect them to remember what you bought before, and you expect them to know your shoe size and every piece of information you've ever given them. But you also expect them to keep that private. You wouldn't want some random person telling you your shoe size—like that would be creepy."

He highlights an immediate opportunity for companies to better utilize their existing data:

"There's this opportunity for companies just to leverage their own first-party data, which so few companies actually do well, which is pretty amazing to me. So there's a low-hanging fruit opportunity here: just help companies leverage the data they've already got."

Beyond that, Scott envisions a consumer-side solution:

"I do think it's a great opportunity for some brand or company to allow us to have portability over our own preferences and some transparency around and control about how and when they're being shared."

He suggests that as users develop trust in AI systems that manage their data, they may become less concerned about the details:

"I would venture to bet that once we use a product like that and start to trust the AI's decision on what information can be shared for whatever reason, we'll like stop looking and we'll stop caring."

Scott emphasizes that this type of product should be one consumers pay for rather than one where "we are the product," proposing:

"The idea of AI that is surgically sharing information that we want to be shared at every moment in time across our digital experiences such that we can have this hyper-personalized world that we were talking about and feel special."

He concludes with a powerful historical perspective that ties together his vision:

"My litmus test for a great product is oftentimes if it takes us back to the way things once were in a way that we long for, but with scale and efficiency. 200 years ago when we walked around towns, in small towns wherever we lived, we were known, we were recognized. The butcher knew your favorite cut, the shoe cobbler knew your kids' shoe size. This was the world. It was wonderful. We liked being known."

"That was the previous 300,000 or so years of humanity—kind of being known. And then the Industrial Revolution happened and technology, and suddenly we're all anonymous. So I fundamentally believe that we do want to be known, we just want to know how we're known, and we wanted to be on our own terms. So if a technology could crack that, I think it would be huge."

Timestamp: [49:49-52:32]

Newsletter:

• Scott shares that he has a monthly newsletter called "Implications" where he shares ideas and gets feedback
• Available at implications.com
• Described as a great way to connect and engage with Scott

Connect with Scott:

• Scott welcomes connections at Scott.belsky
• Expressed appreciation for the audience and event

Timestamp: [52:32-53:03]

• Early tech networks were valuable not just for connections but for challenging debates—find people you respect who will push your thinking
• Many AI products simply automate existing workflows without rethinking whether they're the optimal approach
• AI assistance exists on a pyramid: basic help (bottom), automation (middle), and proactive guidance with opinions (top)
• AI model infrastructure will likely evolve like AWS did—starting with broad services and progressively offering more granular control
• Adobe is not building its own LLMs but is creating media models and orchestration layers to connect various LLMs to its products
• To maintain "founder mode" in growing companies: collapse talent stacks, enable direct access to talent, and foster a culture of directness
• Companies are underutilizing their existing first-party data to create personalized experiences
• The ideal future of digital personalization mirrors pre-industrial community experiences—being known, but with transparency and control
• Great products often recreate what we've lost from the past but with modern scale and efficiency

Timestamp: [40:07-53:03]

Companies/Products:

• South Park Commons - The community hosting the event, referenced as an example of valuable professional networks
• Instagram - Referenced in anecdote about Kevin Systrom declaring he wouldn't sell (before Facebook acquisition)
• Facebook - Mentioned as the eventual acquirer of Instagram
• Photoshop - Used as an example when discussing AI automation possibilities
• AWS - Used as an analogy for how AI model APIs will likely evolve to offer more granular control
• ChatGPT - Mentioned as one of the LLMs Adobe partners with
• Anthropic - Referenced as another LLM provider Adobe partners with
• Acrobat - Adobe product that now includes "Acrobat Assistant," an AI orchestration layer
• StumbleUpon - Referenced briefly in previous segment's discussion about Garrett Camp

People:

• Tim Ferriss - Mentioned as part of early small tech circles
• Chris Sacca - Mentioned as part of early small tech circles
• Garrett Camp - Referenced again as part of early tech networks
• Gary Vaynerchuk - Mentioned as having shared a photo of early tech gatherings
• Kevin Systrom - Instagram founder mentioned in anecdote about not selling the company

Events:

• South by Southwest - Conference where early tech leaders would gather in Austin, Texas

Concepts:

• AI Pyramid - Scott's framework for AI assistance: help (bottom), automation (middle), proactive guidance (top)
• Orchestration Layer - System that coordinates multiple AI models to deliver cohesive experiences
• First-Party Data - Information companies already possess about their customers
• Collapsing the Talent Stack - Finding team members with multiple skillsets to reduce communication layers
• Implications - Scott's newsletter where he shares ideas (implications.com)

Timestamp: [40:07-53:03]