1. An Era Where Everyone Does Everything: What Does "Product" Mean?
At the beginning of her talk, Raiza Martin asks the audience to raise their hands based on whether they work in Product, Engineering, or UX (User Experience). Then she says:
"Nowadays it feels like we're all doing everything. That's the crazy thing about AI. That's the crazy thing about this era."
Thanks to AI's advancement, one person can now simultaneously perform multiple roles, and many people are actually doing so. In the past, a PM (Product Manager) who could write SQL queries was considered "technical," but now anyone can easily do that with tools like ChatGPT.
"Now you just need to know what questions to ask, and AI handles the rest."
In this way, the boundaries between roles are blurring, and team composition is changing significantly. Every team now has invisible AI participants. Behind almost everything we work with -- documents, slides -- lurk AIs like ChatGPT, Claude, and Gemini.
"If you have 5 people on a team, you also have 5 ChatGPTs. I don't know what that means yet, but we've got superpowers now."
2. Everything We Use Now Is in Its "Ugliest" State
Raiza explains that a product is an idea from someone's inner world translated into technology. But looking at every product we currently use, it feels like we're in an "awkward and clunky period."
"Everything we use now is going to be the ugliest it ever is. Because all of this was built before AI."
Products built before AI was fully introduced relied on imagination, and now, thinking about the richness of interaction possible, existing products feel too simple and inconvenient.
"Even everyday things like microwaves feel weird now. Is it just me? Am I the only one who wants an AI microwave?"
We're now at the stage of shaping what AI can do and what products can emerge. And she emphasizes that the truly great products haven't been built yet.
3. The Shift in User Experience and the "Era of Rebuilding"
As people use AI-based products, they feel growing dissatisfaction with existing products.
"When you use products like ChatGPT, Cursor, or Claude, you just say what you want and magic happens. Then you have to go back to using dumb products."
A large gap is forming between smart, intuitive products and legacy, inconvenient products. Raiza predicts that an era when everything gets rebuilt is coming.
But she emphasizes that this process is accompanied by chaos.
"Everything is really cool and magical, but at the same time really hard and chaotic. Every layer is being completely rewritten."
In this chaos, opportunities arise, and the role of a product builder is finding kernels of opportunity and popping them like popcorn.
"Our job is to find the kernels of opportunity and pop them like popcorn. That's our only mission."
4. Principles for Building Good AI Products
4.1. Start from Personal Clarity
Raiza says building a real product is a "forceful" experience.
"To build something truly meaningful, you have to force it into the world."
The most important thing is personal clarity. It must start not from team or organizational clarity, but from one person's firm vision, purpose, and taste.
"Knowing clearly what you're building and why -- that's real energy."
During the development of Notebook LM (formerly Tailwind), Raiza was told by many people "Isn't this a stupid idea?" but was able to push forward because of her own conviction.
"I dropped out of college, then went back to school while working at Google. So I really needed it. A tool where I could put multiple documents and slides in one place and interact with them conversationally. I was convinced this was truly valuable."
4.2. Start from the "Job," Not the "Pixels"
When building products, start from the outcome users must achieve, not from appearance (pixels, design).
"Purpose is the one outcome every user must get every time. That's the north star, and that's the filter for useless features."
She warns against "demo disease" -- common in AI products -- where the demo is impressive but the product is actually useless.
"Users don't care if it's AI. They just want to state their intent and have it naturally happen."
4.3. Trust Is Oxygen
A product is a promise between the user and the company. Users try the product trusting that promise, but once trust is lost, they never come back.
"Trust is oxygen. Without it, you have nothing. Most products start at negative credibility."
To build trust, you must honestly reveal the product's limitations (edges) and respond humanly when failures occur.
"Show where the model is dumb, and naturally bridge the gap between user and product. Don't try to paper over it."
4.4. Basics Come First
Before fancy AI features, basic usability (deterministic things) must be perfect.
"A good app is ultimately just a good app. No matter what you put in it, it's still an app."
In Notebook LM's case, the most common first request from users was "summarize," and if that didn't work, users left immediately.
"When a first-time user tries to summarize and it doesn't work, they just leave. They never come back."
5. The Power of "Delight" and "Focus"
5.1. You Can Only Delight After Earning Trust
After earning trust, opportunities arise to delight users. In Notebook LM, users experienced unexpected fun and enjoyment when uploading documents and creating podcasts.
"Delight comes from the playfulness between technical possibilities and user expectations. Only after building trust can you take that extra step."
To delight, users must feel agency, not trickery.
"Delight comes from the feeling that you're in control. The tension of you and the machine creating something together is what matters."
5.2. Don't Fall Into the "Kitchen Sink" Trap
Products that cram in too many features all at once -- "kitchen sink" products -- end up loved by no one.
"What really matters is respecting the user's time, data, and agency. Restraint is the multiplier of innovation."
Raiza honestly admits that with her own product Huxe, she crammed in too many features only to discover that users actually used just one thing.
"I built a kitchen sink product too. I thought it was so cool, but when we actually used it, both I and the users only used one thing."
6. Conclusion: Clarity, Purpose, Trust, Delight, and Restraint
Finally, Raiza concludes that clarity gives the energy to build, purpose enables focus, trust earns belief from users, and delight proves the product's potential. And the "kitchen sink" serves as a checklist for all four.
"Clarity gives energy, purpose gives focus, trust earns belief, delight proves possibility. And the kitchen sink reminds us whether we're keeping those four in check. That's how you build something that isn't ugly."
Key Concepts:
- Changing roles in the AI era
- Personal clarity
- Purpose-centered product design
- Trust and user experience
- Delight and agency
- Focus and the "kitchen sink" boundary
This talk candidly and accessibly delivers principles and insights that anyone building products in the AI era should take to heart. "Everything is ugly, so build something that isn't!"