In 2025, when everyone is saying AI will change everything, OpenCode's Dax Raad makes the provocative claim that "AI changes nothing." He emphasizes that the core processes of marketing, onboarding, and retention — essential for building successful products — are areas that AI can never solve alone, still requiring deep human thought and sensibility. Here's a summary of his insights on the three fundamental funnel stages of product success that remain unchanged even in the AI era.

1. Marketing: AI Can't Create "Cool"

The first stage you face when building a product is marketing. Dax points out that many developers either feel aversion to marketing or harbor the illusion that "if the product is good, it'll sell itself." But reality is harsh. People are busy with their own lives and have zero interest in your product. The essence of marketing is stopping those indifferent people in their tracks and making them pay attention.

Generic blog posts or predictable advertisements won't cut it. You need to create something so shocking, funny, or profound that people want to share it with friends and colleagues, saying "Hey, have you seen this? It's insane!" This is what Dax calls the domain of "Cool."

"People don't give a s*** about you. You have to make them stop everything they're doing and look at you. That's a really high bar."

"I've never seen a good idea come from AI. Even when I use it as a brainstorming partner, the results are always too corny. AI can't produce 'Cool.'"

Ultimately, creative marketing ideas that move people's hearts must come from human sensibility, not AI. Even if you fail 9 out of 10 times, you need that one "cool" attempt that sticks in people's minds.

2. The Aha Moment: Cut Ruthlessly

Once marketing has captured people's attention, the next step is getting them to the "Aha Moment" — where they realize the product's value — as quickly as possible. Dax says developers must ruthlessly kill their "darling" features in this process.

You must remove all friction on the path to that moment when users first encounter the product and feel "Ah, this is it! I love it!" Unnecessary sign-up requirements and extraneous feature explanations only drive users away.

"You have to be really brutal. You might have to kill your children. Pick the one most important thing and deprioritize the rest. Look at every step it takes for a user to reach that core moment, and cut them all."

Dax cites ChatGPT as the most successful product example.

"ChatGPT is one of the most successful products of our lifetime. There's just one input box sitting there, and when you type anything in, a human-like answer comes back. Even the most clueless person can use it and instantly think 'Wow, this is amazing' — an incredible aha moment."

Designing a product so users can instantly grasp its core value isn't about algorithm optimization — it's work that requires deep human taste and intuition.

3. Retention: Build Primitives, Not Features

Once you've acquired users, you need to turn them into lifelong customers. What matters here is simultaneously satisfying casual users and power users. People typically think you have to choose between "simplicity" and "power (versatility)" when building a product, but Dax says this is a false dichotomy.

The way to catch both is to build "Primitives" first. Instead of building specific features from the start, you first develop powerful, fundamental building blocks, then assemble them to provide a simple experience for regular users. This way, when power users later want more complex and deeper functionality, you can give them access to the underlying primitives directly, enabling infinite extensibility.

"There's no trade-off between simplicity and power. It's just more work upfront. Instead of building features directly, build the primitives that can be assembled into those features."

"AI can't hallucinate the right mental model for your problem domain. Designing and figuring out which primitives are needed requires enormous intellectual labor."

Designing a product's architecture and creating a deep model that serves both beginners and experts is a domain that requires the developer's profound understanding — something AI cannot do for you.

Conclusion: Painful, but That's What Makes It Valuable

Dax closes his talk with a candid confession. Even he finds it incredibly difficult to uphold these principles every day. The theory is simple, but execution involves bone-grinding effort. Yet he says that's exactly where the hope lies.

"AI lets us do things we couldn't do before, but it doesn't eliminate the daily pain of building something great. My life is still just as hard as it was before. But that's where all the fun and purpose comes from."

In conclusion, the formula for winning products hasn't changed in the AI era. You still need the "cool sensibility" that captures people's hearts, the "bold decisiveness" to focus on what matters, and deep "design ability." AI is merely a tool — in the end, only humans who think fiercely and put in the sweat can build products that win people's hearts.