Mark Pincus shares proven strategies for building breakthrough consumer products with AI. Learn the "proven, better, new" framework and founder mode principl...
How to Build Consumer Products People Love in the AI Era
Key Insights
- Consumer products remain underinvestable but overopportune: Even as venture capital avoids the sector, the technological shift to AI creates unprecedented chances to reinvent basic services people use daily
- The "proven, better, new" framework separates innovation from improvement: Proven elements are validated solutions you can copy legally; "better" means clear improvements; "new" is the untested hypothesis requiring rigorous experimentation
- Founder mode requires both instinct and team alignment: True leadership means trusting your instincts while creating organizational culture where team members embrace pivots and remain intellectually honest about what works
- AI compute costs will eventually become negligible: Within 3-4 years, inference costs should drop dramatically, making today's expensive AI applications mainstream consumer products—builders who start now will have enormous advantages
- The consumer AI revolution is 2-3 years away, not immediate: Just as Amazon took six years to become obvious, the true consumer AI wave will peak around 2029 when costs finally align with mass-market adoption
The Paradox: Why Consumer Is Uninvestable But Inevitable
The consumer sector currently sits in a fascinating contradiction. Venture investors are retreating from consumer-focused startups, viewing the space as saturated and risky. Yet Mark Pincus, founder of Zynga and serial entrepreneur across five companies, argues this is precisely when builders should be most aggressive about creating consumer products.
"Even though the consumer sector may not currently be the most attractive for investment, the opportunity to offer people new, invaluable internet experiences has never been greater," Pincus explains. This paradox stems from technology, not market demand. The fundamental breakthrough isn't in consumer behavior—people still want better, simpler, more delightful tools. The breakthrough is in what's now possible to build.
Consider your smartphone's home screen. For most people, it's half empty. The apps that remain are either generic utilities (clock, notes, camera, weather) or relatively recent discoveries (ChatGPT, Claude). Despite decades of app development, we've invented remarkably few "internet treasures"—services so valuable we can't imagine life without them, like Google or now AI assistants. This suggests the playing field is far from exhausted. Instead, we're at an inflection point where AI and intelligent agents make it possible to reinvent basic services previously thought obsolete or generic.
The disconnect between investability and opportunity reveals an important truth: investor sentiment doesn't determine technological possibility. When Peter Steinberger spends $1+ million monthly on AI tokens to build world-class open-source tools, he's conducting live R&D on what becomes possible when cost isn't a constraint. His work demonstrates that the technology already works—it just costs too much for consumer pricing. That's a temporary condition, not a permanent one.
Understanding the "Proven, Better, New" Framework
One of Pincus's most valuable contributions to founder thinking is a simple but rigorous framework for evaluating new product ideas: the "proven, better, new" model. This framework helps distinguish between improvements (which are safer but less groundbreaking) and genuine innovation (which is riskier but potentially transformative).
Proven elements are existing solutions that work. You can legally copy them without reinventing the wheel. Taking proven elements from other products or industries saves enormous time and energy. For instance, when Pincus created Freeloader, the proven element was the screensaver—a software concept that already existed. The company didn't need to invent what a screensaver was; they just needed to deliver one via internet distribution.
Better elements are improvements that existing users would unequivocally embrace. Making something free when it previously cost money is "better." Making a service twice as fast is "better." Making an experience with 90% less friction is "better." The critical distinction: every current user of the existing solution would agree the improvement is valuable. There's no debate. Everyone prefers free to paid, faster to slower, simpler to complex.
New elements are the untested hypotheses—features, business models, or experiences no one has tried. These require careful experimentation because they might be completely wrong. A new distribution channel. A new use case. An entirely new way of interacting with a service. These are the experimental spikes where breakthroughs occur, but most attempts fail.
When Pincus applied this framework to his book, "Life At The Speed Of Play," using Claude to analyze the content:
- The "proven" aspects (using existing AI models to solve document analysis) scored an A-minus: faster and better than most solutions
- The "better" aspects (improvements over existing tools) scored a B-minus: not consistently superior across all dimensions
- The "new" aspects (novel applications of the technology to book analysis) scored a D: requiring significant human insight and testing to validate
This framework prevents two common mistakes. First, it stops founders from treating proven elements as novel innovations—wasting effort on reinvention. Second, it prevents treating truly new ideas as if they're already validated, which leads to building the wrong thing at scale. The best consumer products typically combine all three: proven foundations that customers already trust, visible improvements that matter, and one or two genuinely new ideas that create differentiation.
The new idea doesn't sustain interest indefinitely. It's the spark that attracts initial trial—the "back of the box" appeal that makes you try a new cereal. But you return for the core product. As Pincus notes, if you tell someone "granola that's always listening and always on," they might try it. But they'll only keep using it if the granola itself is exceptional.
The Concept of "True Signal" and "The Fish Are Running"
Every founder faces a critical question: How do you know when you've genuinely hit something great versus when you're just hoping? Pincus calls this moment "true signal" or "heat"—that lightning-in-a-bottle moment when product, market, and execution align perfectly.
"When you have true signal, you know it. And when you don't, you genuinely don't know it's not the right thing," Pincus explains. This clarity manifests across every dimension. Metrics become almost irrelevant because every feedback loop screams "fuck yeah." Users don't need convincing; they're frantically sharing the product. Your team doesn't need motivation; they're energized because the vision is obvious.
Pincus calls this "the fish are running"—borrowed from fishing culture. When fish are running, fishermen don't need managers exhorting them to work harder. They're eagerly throwing nets all night because the opportunity is exhilarating and obvious. Everyone sees it. No data analysis required. No strategy meetings about whether to double down.
He's only felt this phenomenon a couple of times in his career. With Freeloader, his first company, timing and execution aligned perfectly—2 million downloads in the first month of a free app competing against a paid product. At Zynga, he experienced it repeatedly with game launches and feature releases that felt inevitable in their success before launch day arrived.
This is why Pincus emphasizes that truly great product makers are "collecting winnings, not making bets." By the time they launch, they already know something will resonate. They're not waiting to see if users like it. They know. The conviction comes from something deeper than data—a combination of user research, instinct, pattern recognition across markets, and deep product sense.
The counterpoint is equally important. When true signal is absent—when the data is mixed, the metrics debatable, the team uncertain—that's when you need to ask hard questions. Should you pivot? Should you kill it? Should you keep iterating? This is where many founders struggle, especially after convincing investors and building a team around an idea.
The hardest version of founder work is lying in bed on a Sunday night, thinking, "I don't think this product is right." On Monday morning, you face a choice: Tell the team the pivot is necessary, subtly shift direction without being honest, or endure the project while your conviction erodes. This is where founder authenticity becomes a management tool. If your team knows you'll be honest about conviction, they can collaborate on finding true signal rather than defending a dying idea.
Founder Mode: Instinct, Alignment, and Intellectual Honesty
Brian Chesky's concept of "founder mode"—the idea that "leadership is presence, not absence"—deeply resonates with Pincus's approach to building companies. But Pincus extends this beyond board-level governance to a weekly operating principle.
Founder mode means trusting the instincts that got your company here in the first place, then creating organizational culture where the team understands why those instincts matter. It's not about the founder making every decision. It's about the founder setting the context where rapid learning and course-correction are team values rather than signs of indecision.
"People have said working for Mark can feel like third-grade soccer where every Monday he comes in and falls in love with a different idea. And sometimes they're right. But if they're saying that, I'm also not doing a good job of creating context for me to change every Monday," Pincus acknowledges. The founder mode challenge isn't having instincts. It's building a team culture where instincts are respected and pivots are expected.
This requires deliberately creating intellectual honesty as a team value. When the team meets on Monday morning, the question isn't "How do we execute the plan?" It's "What did everyone learn last week?" Team members should feel safe raising hands and saying, "I saw another product doing this better," or "I don't understand why we're doing this," without fear of being seen as disloyal.
This culture is fundamentally different from execution-machine cultures where there's no room for questions. Some companies must operate in execution mode—squeezing the lemon on a successful product. Other companies must operate in exploration mode—trying things that have never been done. Great founders recognize when their company is in each mode and set expectations accordingly.
The context Pincus describes sounds like: "You know what, we're going to that continent. That hasn't changed. That's our mission. But I'm going to talk at different altitudes now. That's 100,000 feet, our destination. Today, we're at 5,000 feet. This tactic isn't heading toward that continent. We need to go that way instead." This frame lets teams feel the stability of long-term vision while understanding that weekly tactics can shift as data arrives.
The AI Moment: Why 2029 Matters More Than 2026
Current discussions about AI's consumer impact often suffer from temporal confusion. People simultaneously claim AI is revolutionary and wonder why AI products haven't exploded into mainstream consumption. Both can be true because the timing doesn't align yet.
Consider the cost structure. Cutting-edge models like Opus 4.5 (released in December 2024) already deliver remarkable capability. But delivering that capability at consumer scale costs hundreds of thousands of dollars monthly. This creates a paradox: the technology to build magical consumer AI products exists right now. It just requires spending like an enterprise to offer it to a single user.
"If the Opus 4.5 moment was just in December, requiring payment of tens to hundreds of thousands of dollars to get truly substantial work done, then the ideal consumer moment is still three orders of magnitude away," Pincus argues. Three orders of magnitude—a 1,000x difference—means the technology needs to become 1,000 times cheaper before it becomes viable consumer pricing.
When will that happen? Pincus estimates around 2029—about three years from now. This timeline has historical precedent. The internet boom showed similar patterns. When Amazon IPO'd in 1997, most people thought the internet was a niche curiosity. Amazon did important work through the late 1990s and early 2000s, building infrastructure and refining operations. But it wasn't until late 2002—a full six years later—that their financial performance shifted dramatically and "everyone" suddenly realized the internet was transformative. At that point, everyone felt late, because you needed to have started in 1996-1997 to be positioned for 2002's success.
The same pattern is emerging with consumer AI. The current phase (2024-2026) is infrastructure and experimentation. The real consumer revolution arrives 2027-2029 when cost curves align with mass-market pricing. Anyone building now—even if investor capital is scarce—has a six-year head start on the people who start building in 2028.
This reframes the uninvestable nature of consumer AI. It's not that consumer AI is impossible. It's that consumer AI at scale still costs too much for venture returns. But that's a cost problem, not a capability problem. The founder who understands this builds differently. Instead of asking, "How do I monetize this within current cost constraints?" they ask, "What would be magical if compute were free? How would I design this if I had unlimited AI budget?"
Building Now With Tomorrow's Economics in Mind
The practical question for founders is: How do you build consumer products designed for 2029 economics while operating in 2026 cost constraints?
One approach is what Pincus calls "squandering tokens"—deliberately spending heavily on AI computation during the R&D phase to understand what becomes possible without cost constraints. Peter Steinberger's $1+ million monthly token spend isn't frivolous. It's frontier exploration. He's discovering what actually works when cost isn't a limiting factor. That knowledge becomes invaluable when costs drop and others try to replicate it at scale.
Another approach is backwards-building from the future. When costs finally drop, what would people actually want? Not what people want within today's constraints, but what would they want if the constraint disappeared? If you could offer unlimited AI-powered weather prediction, what would that look like? If you could offer unlimited AI photography assistance, what would that look like? If you could offer unlimited AI-powered note-taking with perfect recall and understanding, what would that look like?
This is where the "logs of cost curves" approach becomes powerful. Anyone can look at historical cost-reduction curves for compute, storage, and model inference. These curves are highly predictable. From the iPhone's emergence, you could have mapped out exactly when display costs, memory costs, and compute costs would align to make a smartphone possible. Similarly, you can map when AI inference costs will hit various thresholds.
"You could start building now and working backward. That's an incredible head start," Pincus emphasizes. The founder who understands when a cost transition happens can position products to benefit from it perfectly. They start building for free AI while everyone else is still arguing about affordability. Then when the transition hits, their product is ready for scale.
The free model deserves special attention here. Pincus has built it twice: Freeloader competed against Berkeley Systems' $35 Flying Toasters with a free internet-distributed screensaver. Zynga asked whether you could make quality games free when the industry standard was $60 boxed games. Both times, free won decisively.
"You always know what's better: free. It's one of the rules of the internet: anything that can be free will be free," Pincus notes. This principle applies directly to consumer AI. What would happen if you offered basic versions of services with unlimited AI—free? Not as a loss leader, but because compute is genuinely cheap enough to sustain the business model. That's the future state worth building toward now.
The Abyss, Taste Zones, and Staying Power
Building consumer products toward a distant future requires a different kind of staying power than venture-backed sprints. Pincus describes these periods as "the abyss"—the uncertain space between passionate product pursuits.
"The abyss is this place, in between passionately pursuing products, where we find ourselves uncertain if we'll ever find something we're passionate about again, or if we'll ever emerge from it," Pincus explains. This is the emotional reality of building ahead of your market. You're working on something meaningful while the market doesn't yet see it. Skepticism comes from all directions—investors, peers, sometimes your own team.
The antidote isn't forcing passion for a particular product idea. It's expanding what Pincus calls "taste zones"—deliberately developing taste and conviction across different domains. He stays in love with building by consuming and understanding new products, new interfaces, new experiences. He has ChatGPT and Claude on his phone's home screen because he's genuinely using them, learning from them, developing taste for what AI products could become.
This taste development is essential to staying power. "It's hard to maintain staying power if you're not falling in love with things you're building. We need to go out and fall in love with things to get our inspiration," Pincus says. The founder who spends six months not using any new consumer products, just executing on their existing idea, gradually loses conviction. The founder who continuously falls in love with new products and services maintains the instinct for what matters.
Reimagining Services in the AI Age
Perhaps the most exciting implication of Pincus's thinking is about reinventing existing services. Every major consumer internet success—Google, YouTube, Facebook, Uber, Airbnb—reimagined existing categories in breakthrough ways. Google didn't invent search; they reimagined it. Uber didn't invent transportation; they reimagined car service. Airbnb didn't invent short-term rentals; they reimagined hospitality.
What gets reimagined in the AI era? Almost everything.
Consider services taken for granted: note-taking, email, photography, navigation, communication, information search, productivity tools, entertainment, education. Every one of these could be fundamentally reimagined with AI and agents at the core. The camera app could intelligently process images in real-time, understanding context, automatically organizing photos, suggesting compositions. Notes could be understood semantically, automatically linking related thoughts, predicting what you meant to write. Navigation could adapt not just to traffic but to your preferences, learning over time.
These aren't incremental improvements. They're reconceived services with AI as a fundamental layer, not an afterthought.
The challenge is that most of these opportunities have the "cost problem" Pincus describes. You can build the magical version—the one designed as if AI were free. It will cost $1,000+ monthly to offer to a user. That's not consumer; that's enterprise. But the founder who understands the cost curves knows exactly when that changes. They can position their product to hit the market at the precise moment cost curves flip from prohibitive to viable.
"I would argue that if the 'Opus 4.5' moment was just in December, requiring payment of tens to hundreds of thousands of dollars to get truly substantial work done with those technologies, then the ideal consumer moment is still three orders of magnitude away," Pincus emphasizes. That three-year window is the building phase. By the time the market realizes the opportunity, the builders who started now will already own it.
Practical Advice for Founders Building in 2026
For founders deciding whether to start a consumer product company right now, Pincus offers several practical principles:
Embrace founder mode deliberately. Don't apologize for following your instincts. Instead, invest heavily in creating team culture where instincts are valued and pivots are expected. Frame directional changes as learning, not failure.
Use the proven-better-new framework ruthlessly. Separate what you're copying from what you're improving to what you're genuinely innovating on. Score each honestly. The truly new elements deserve the most intense experimentation because they're most likely to surprise you.
Build for the future cost curve, not current constraints. Ask what would be magical if AI were free. Design for that future. Then figure out how to sustain current operations. The combination gives you a three-year advantage when costs finally drop.
Develop taste voraciously. Use new products. Understand what's working. Fall in love with things you're building. Let taste guide instinct. The founder who stops tasting new things gradually loses conviction.
Remember the fish-are-running test. If you don't feel it, don't have it yet. Iterate faster. Every day you're building something you don't believe in is a day you could be building something you do believe in.
Don't get demoralized by investor sentiment. Consumer is uninvestable right now because costs don't work. That's temporary. The opportunity is massive precisely because capital is skeptical. Build anyway. The timing might feel early, but six years from now it will feel inevitable.
Conclusion
The paradox of consumer products in 2026 is that they're simultaneously uninvestable and inevitable. Investor capital has retreated from consumer-focused startups, viewing the landscape as saturated or risky. Yet the technology to build genuinely new consumer experiences—powered by AI and intelligent agents—has never been more capable. The gap is cost, not capability. Within three years, that gap will vanish.
For founders willing to embrace this timing, the upside is extraordinary. The consumer services and experiences people will love in 2029 are being built right now by founders who understand cost curves, trust their instincts, and build toward future economics rather than current constraints. The opportunity to create internet treasures—services so valuable people can't imagine life without them—hasn't been greater. Start now, and by the time the market catches up, you'll already own it.
Original source: Zynga Founder: Consumer Is Not Investible Right Now - Thats Why You Should Build It
powered by osmu.app