Interview with Ivan Zhao: How We Built Notion to Replace Every Other App
Ivan Zhao, co-founder of Notion, on starting from scratch, obsessing over simplicity, and why the best software feels like it disappears.
Ivan Zhao doesn't like being interrupted. Not mid-sentence, not mid-thought — and certainly not mid-build. When Notion nearly ran out of money in 2015, he and co-founder Simon Last moved to Kyoto for eight months to rebuild the product from scratch, away from Silicon Valley's noise. That stubbornness, it turned out, was the point.
Today, Notion is used by over 35 million people. We sat down with Ivan to talk about what it actually takes to build software that gets out of your way.
Origins
Q
You've described Notion as a response to software fragmentation. What did you actually see that made you think this was a real problem worth solving?
I was a programmer, so I kept trying to build my own tools. Wikis, databases, task lists — I'd stitch them together with code. At some point I realised: why doesn't something like this just exist? Why is everyone switching between six apps to do one job?
The problem wasn't that the apps were bad. Trello is great. Evernote was great. The problem is that your thinking doesn't live in categories. An idea becomes a task becomes a doc becomes a database. The tools made you translate constantly, and every translation loses something.
Q
You rebuilt the whole product from scratch in Kyoto. Most people would call that a disaster. You've called it necessary. Why?
The first version had the right idea but wrong architecture. We were working around limitations in the foundation rather than building on it. You can't fix that with iterations. The only honest thing is to start over.
Kyoto was a deliberate choice. We needed to be somewhere with no meetings, no demos, no pressure to look like we were making progress. Just work. Eight months of that gave us something we couldn't have gotten in San Francisco in two years.
"You can't fix a wrong foundation with iterations. The only honest thing is to start over."— Ivan Zhao, Co-founder & CEO, Notion
Product philosophy
Q
The block-based editor was a big technical bet. What made you confident the middle ground worked?
Confidence isn't quite the right word. It was more like — we couldn't see another way. If everything is a block, and blocks can be any type, and blocks can be nested, then you've built a language. You're not building features, you're building primitives. Features come from combining primitives.
The bet was that users are smarter than software designers think. Give people the right primitives and they'll build things you never imagined. That's what happened. The templates people have built in Notion — we couldn't have designed those top-down.
Q
Notion is famously hard to explain. Do you see that as a marketing problem or a design problem?
Both, but mostly neither. It's a category problem. When something is genuinely new, the most honest description sounds wrong. We used to say "it's like Lego for your work." People nod, then try it, then get it.
The fix was templates. Show someone a real CRM built in Notion, a real habit tracker, a real engineering wiki. The category stops mattering when they see the outcome.
Building the team
Q
You stayed at two people for a long time. What was that actually about?
Trust, honestly. Adding people to a small company changes the culture before you have one. Simon and I shared enough context that we could disagree fast and move fast. Adding a third person when you're still figuring out what you're building slows you down in ways that aren't obvious.
Two people who are fully bought in ship faster than five people who have competing intuitions. When we eventually hired, we hired people who had strong taste, not just skills. Skills can be learned. Taste is much harder to teach.
Q
Notion has a reputation for a very specific aesthetic — calm, considered, not "SaaS-looking." Was that intentional from the start?
Yes. I studied fine arts before computer science. The intersection of those two things is what I've always been interested in — software that feels like it was designed by someone who cared. Most software looks like it was designed by a committee that was afraid of making a choice.
There's a Japanese concept, ma — negative space, the meaning in what's absent. We think about that a lot. The goal isn't to add things. It's to remove everything that shouldn't be there until what's left is inevitable.
"The goal isn't to add things. It's to remove everything that shouldn't be there until what's left is inevitable."— Ivan Zhao
What's next
Q
AI is changing how people interact with productivity software. How does Notion think about it — is AI a feature, or does it change the product fundamentally?
It changes the product fundamentally. Not because of any one capability, but because the interaction model shifts. Right now, you're directing the software. With AI, the software can understand intent and act.
For Notion, this means the same primitives get more powerful. A block can now generate, summarise, connect. The underlying model turns out to be a very good foundation for AI, because AI works best when the data is structured and typed. We didn't plan for that. It was lucky architecture.
Q
Last question: what do you tell founders who are building something that doesn't fit a category yet?
Be patient with the confusion. The confusion means you're doing something real. If everyone immediately understands your product, you're probably in a category that already has a winner. The hardest part isn't building it — it's staying convinced that the category will exist, while it doesn't yet.
Find five users who love it. Not who say they like it — who actually use it every day and would be devastated if it disappeared. Build for those five until you find fifty. The explanation will come later.
IZ
Ivan Zhao
Co-founder & CEO, Notion
Ivan Zhao co-founded Notion in 2013 with Simon Last. Before Notion, he studied cognitive science and fine arts at UBC, which shaped his belief that software should be as expressive as a blank canvas. Notion is used by over 35 million people and teams including Figma, Pixar, and The New York Times.