For fifteen years, the startup world had a favorite mantra. Scott Adams put it most directly: "Ideas are worthless. Execution is everything." Derek Sivers turned it into a formula: ideas are a multiplier of execution, but even a brilliant idea multiplied by no execution equals zero¹. The message was clear. Don't fall in love with the idea. Fall in love with getting it built.

Even then, the mantra was incomplete. A brilliant idea well-executed always had more leverage than a mediocre idea perfectly built. But building was hard enough and expensive enough that it became the obvious constraint. Execution ate everything, so people stopped examining what came before it.

Then AI compressed that gap dramatically.

The shift happened fast. Andrej Karpathy coined "vibe coding" in early 2025 to describe what was already becoming obvious: people could describe what they wanted in plain language and get working code back². By that same year, the term was Collins Dictionary's Word of the Year. Y Combinator reported that a quarter of its Winter 2025 startups had codebases that were 95% AI-generated³. Apple's App Store logged 557,000 new submissions, a 24% jump and the biggest wave since 2016⁴.

Building got easier. Not incrementally. Categorically. Things that took teams months can now be done by one person in days. The cost of turning an idea into a working product is approaching zero.

But here's what didn't get easier: making anyone care.

If anything, that got harder. People encounter somewhere between 4,000 and 10,000 ads and commercial messages daily, and consciously notice fewer than a hundred⁵. When building is easy, everyone builds. More products means more noise. That carefully crafted app is now one of thousands launched the same week, competing for attention spans that keep shrinking⁶.

The bottleneck moved. It shifted from "can we build this?" to two harder questions: what should we build, and can we make anyone care? These aren't engineering problems. They're creative ones.

The first question (what to build) is about product creativity. Taste. The ability to look at what exists and notice what's missing, or what's broken, or what could be ten times simpler. This isn't just "having ideas." Everyone has ideas. Paul Graham made a useful distinction: the best startup ideas aren't thought up but noticed, by people who live close enough to a problem that they can't unsee the absence of a solution⁷. That kind of noticing is one example a creative act. It requires attention, judgment, and the willingness to trust your own dissatisfaction.

If you want the defensibility version of that first question, The AI Moat You Can't Buy argues that the edge is a loop, not an asset.

The second question (making people care) is marketing creativity. Not just storytelling, though that's part of it. Positioning, framing, audience-building, finding the angle that makes someone stop scrolling. The ability to create context around a product so people understand not just what it does but why it matters to them. Most builders have historically ignored this entire skillset or outsourced it. That worked when building was the hard part. Now that everyone can build, earning attention is the most undervalued thing a founder can learn.

That attention problem is exactly the first-aha problem in Your Users Need Two Aha Moments (You're Probably Only Building One).

So here's the counterintuitive part. We spent years telling builders that ideas don't matter. Now the truth has flipped: ideas, in a specific sense, are the main thing that matters. Not only shower-thought ideas. Ideas as creative judgment applied to real problems. The ability to see what's broken and imagine what's better.

And there's a layer beneath this. The same AI that made building easier also commoditized mediocre marketing. Anyone can generate copy, create social posts, produce content at scale. So even on the distribution side, the floor has risen. Generic marketing is now as cheap as generic building. What cuts through is creative marketing: the kind that surprises, that reframes, that earns attention instead of demanding it.

The premium is on creativity at both ends of the pipeline. Product creativity and distribution creativity. The builders who thrive will be the ones who can do both: notice an underserved problem and articulate why their solution matters in a way that resonates.

This is encouraging for a specific kind of builder. The generalist who thinks across domains. Who's good at pattern recognition and communication. Who builds things because they genuinely want them to exist. That person just became well-positioned in a way they weren't before. They don't need a team of engineers anymore. They need good taste and the ability to communicate what they see.

Sivers's old framework assigned dollar values: a brilliant idea with no execution is worth $20, while the same idea with brilliant execution is worth $20 million¹. Execution dominated because it was the expensive, scarce part. AI made execution cheap and abundant. What's left is the creative work on both sides: seeing what to build and making people care that you built it.

The scarce resource was never the code. It was the clarity to know what's worth building and the creativity to make people care. AI didn't change what matters. It just made it impossible to pretend otherwise.

Rabbit Hole

  1. Derek Sivers, "Ideas Are Just a Multiplier of Execution" (2005). Sivers argues that an idea alone is worth nothing and only becomes valuable when multiplied by energetic execution. His framework assigns dollar values: a weak idea is worth $1, a brilliant one $20, while execution ranges from $1 (no execution) to $1,000,000 (brilliant execution). The resulting product is the multiplied value. Scott Adams offered the blunter version in his 2013 book "How to Fail at Almost Everything and Still Win Big": "Ideas are worthless. Execution is everything."
  2. Andrej Karpathy coined "vibe coding" on February 3, 2025, in a post on X describing a new approach to building software by describing what you want in natural language rather than writing code manually. Collins Dictionary named it their Word of the Year in November 2025, reflecting its rapid cultural adoption beyond the tech industry.
  3. Y Combinator, Winter 2025 batch. YC managing partner Jared Friedman reported that 25% of startups in their cohort had codebases that were 95% AI-generated. The figure excluded boilerplate like library imports and measured only code typed by humans versus AI, signaling that AI-assisted development had crossed from experiment to mainstream practice.
  4. Appfigures, "The App Store Just Logged Its Biggest Release Year in Nearly a Decade" (December 2025). Apple's App Store saw 557,000 new app submissions in 2025, a 24% increase from 2024 and the largest wave of new releases since 2016. Appfigures attributed the surge partly to LLM-powered tools lowering the technical barriers to app development.
  5. Estimates of daily commercial message exposure vary widely. Marketing research commonly cites 4,000 to 10,000 exposures per person per day across all formats (digital ads, social media, push notifications, traditional media). The American Association of Advertising Agencies has noted that specific figures are difficult to substantiate, but the directional point holds: exposure vastly outstrips conscious attention.
  6. The attention scarcity problem is structural: more content competing for finite human attention means every new product starts with a distribution disadvantage that didn't exist a decade ago. Mobile usage and social media feeds have increased both the volume of messages and the speed at which people scroll past them.
  7. Paul Graham, "How to Get Startup Ideas" (November 2012). Graham argues that the best startup ideas come from noticing problems in your own life that feel intolerable, rather than sitting down to brainstorm. "The very best startup ideas tend to have three things in common: they're something the founders themselves want, that they themselves can build, and that few others realize are worth doing."