First Block: Interview with Ivan Zhao and Simon Last, Co-Founders of Notion

Obsession. You got to be a junkie of something, right? I remember tweaking the typography of the website. Someone told me, you got to stop.

Yeah, we would get in arguments because Ivan would be spending literally months on our marketing site and I'd be like, oh, we got to work on the product. And then we get in arguments like, oh, it's not quite ready yet. Ivan is always the one to be thinking it's not quite ready and still needs a few more touches. And I'm pretty obsessive too, although Typically in our dynamic, I was often the one that was like pushing more to be like, okay, it's good enough, let's get it out.

You're listening to First Block, a Notion series where founders and executives from the world's leading companies tell us what it was like to navigate the many firsts of their startup journey and what they learned from that experience. I'm Akshay Kothari, Notion's co-founder and COO. Throughout this season of First Block, We've had amazing guests from some of the fastest growing companies out there. Today we're doing something a little different. I'm excited to sit down with my co-founders Ivan and Simon to give you a peek behind the curtain here at Notion. Welcome to the final episode of the first season of First Block. We've had some amazing entrepreneurs and I'm so excited to wrap this season with Ivan and Simon, my co-founders at Notion. Thank you for making the time today.

Yeah.

Thanks. All right, I'm going to usually start with Ivan, but I'm going to start with Simon today. Simon, tell us your part of the Notion story. I guess how did you meet Ivan? How did you get here?

Yeah, I had just turned 20 at the time and I just finished my second year of college and I came out to California for an internship for the summer. It was my first time on the West Coast. I found on Twitter some people that I was interested in talking to and ended up reaching out to Toby Shockman, who was working with Ivan. And I arranged to meet Toby thinking it was just going to be Toby, but then Ivan shows up and is also there. I didn't know who this guy is. And then, yeah, starts trying to recruit me to join them in this new project that they just started. Yeah, within a couple hours I had been convinced enough to quit the internship and the next day I moved out of my apartment in Palo Alto and took the Caltrain up to the office at the time in the Mission and I lived up there for the rest of the summer and yeah, kind of the rest is history since then. Been working with Ivan ever since.

I think from your perspective, you showed up at the first meeting and started pitching. What gave you the conviction to basically go hire this Simon.

A lot of people are Twitter famous and Twitter famous, they have a portfolio. Simon has a portfolio, great portfolio. He built a lot of games, built a lot of things in this kind of tools for thought space, but he's not Twitter famous and this is like, oh, who's this kid? So clearly talented. Then we met up, then it feels like, wow, we think this very similar, think a lot about tools, think about visual programming languages, alternative way to use your computers. So I think a lot about kids, education system for kids, then feels like this is the right person. We start working together. He's so young at the time, he's under drinking age. And remember the old office next by the TRIDOC in the Mission, And we couldn't get in because Simon's under drinking age, but Trick Dog shared the bathroom with the restaurant next door. So we usually go to the restaurant next door, go through the restroom and go into the bar Trick Dog. So that's how we get Simon in.

They're going to have to close this loophole now.

No, it's fine.

Also, if I remember correctly, because he was still in college, I think you also ended up meeting Simon's Yeah, because dad, right, trying to convince Simon to drop out of school, it doesn't need much convincing.

It's pretty easy. And I think his dad and your sister came over too, right? Dad and sister came over, shook hands with the dads like, I'm not a crazy person. Then yeah.

So you hire Simon as an intern that summer and then talk to me a little bit about that journey to becoming Your co-founder?

I think startup goes through so many ups and downs. Co-founder is not necessarily the person you start a company on the very first day. It's the people who went through the ups and downs together. There's some of the darkest days just like me and Sam were just coding together and people like that and you and Sam going through that has to be part of Can you talk a little bit about those early years?

I guess it also seemed like you guys actually grew the team and then came back down to 2, right? Maybe a little bit about that journey where I guess it was a time when we didn't have product market fit.

Many flavors of the stories on the internet. I think we've been talking, many people ask us about that question. I think people love this kind of hero's journey of return. So the idea is more or less the same. We have this powerful thing called computers, but only people who know how to program can get the full potential out of this medium, this tool. So how can we create better version of programming environment, better version of software, better version productivity tools to get more power out of this for all the billions people who use these computers every day. So that's always the theme. First version of that we tried is kind of like, oh, maybe the better way to get more people out of the power of programming is to create an easier programming environment like Webflow meets Figma. We have Vlad on the show earlier. Tried that for 2+ years, 2, 3 years. Learning of that is actually most people don't think about making software, don't think about programming their own tools. Most people just want off-the-shelf thing that solve their problem really quickly. That's kind of like a big aha for us. We're so idealistic because making tool is so easy for us, so natural, but we thought other people want the same thing as you want, but no, it's like Toolmakers are rare in the world, so it's less about that, but rather more about how do we disguise toolmaking power, programming power into a product that people use every day. So Notion transitioned more from a computing tool, programming language tool into a productivity tool. The ethos still is the same. So how do we— ethos still, how do we hide the computing power into productivity software? That took us 3, 4 years.

Was there a specific moment you remember where you sort of like, aha, we need to actually disguise it as a productivity tool? That's the right wedge into the market.

I think it's probably just gradual. We even tried stuff like the concept, remember that stuff?

Yeah, I do remember a specific moment where we had the explicit thought of like, oh, instead of trying to go directly at programming, let's just hide it in the tools people are already familiar with. And what's a tool that everyone uses every day? Documents. And I think that was an explicit thought. I think another big thing from those early days is I think Ira Glass said when you're early on in your career building expertise, your taste is way higher than your ability to execute. And I think that was very true for us. We just weren't that good at building stuff and it was very frustrating. We knew what we wanted it to look like and we wanted it to be. And I think we just made a lot of technical and product mistakes and I think that's a big reason why we had to reset and start over.

We'd probably try 3, 4 times, 3 or 4 rewrites, right?

Yeah, I think the salient thing is not actually ever giving up and just keep going, keep trying again. And eventually, as long as you're being open and learning from your mistakes, you'll eventually get good enough to do it.

I sort of feel like, Simon, you're among the most positive and pragmatic people I've met, but I'm just trying to think, imagine you in those years, you're 22, 23, things are not working, company's about to lose the funding it has. What was going through your head during that time?

I think Ivan and I both are very kind of level people in terms of our mood and it's very easy for me to, for better or worse, it's very easy for me to just get excited or obsessed about some new thing and just kind of forget all the other problems. And so maybe there were definitely some low points where we'd have discussions and what should we do next? What can we do? But it's so easy for us in our dynamic to talk about, okay, well what if we build this? And the conversation quickly turns into a positive one and the problem we're trying to solve is pretty big and kind of impossible to fully solve. And so that's that's a big benefit because within this massive unsolvable problem, there's always new facets to focus on and new things to get excited about and it's kind of endless.

Yeah, I think obsession's pretty key. I remember when you first joined Notion, you might forgot the Steve Jobs biography came out. It's pretty thick. Then you finished it in one weekend or something and it's like, wow, this kid is obsessed about things. And it's like, why is your eye so red? That's all I was reading the biography the entire weekend. Yeah.

I think a lot of that is, I feel like a lot of similar attributes you carry as well. I remember when we did the values exercise, just coming up with the company values.

Yeah, I did 10 versions or something.

I feel like you probably did 10 versions, but also you read 100 different books of other companies' values. It seems like both of you kind of go deep.

obsession, you got to be a junkie of something. So I remember tweaking the typography of the website. Simon told me, you got to stop.

Yeah, we would get in arguments because Ivan would be spending literally months on our marketing site and I'd be like, oh, we got to work on the product. And then we get in arguments like, oh, it's not quite ready yet. Ivan is always the one to be thinking it's not quite ready and still needs a few more touches. And I'm pretty obsessive too, although typically in our dynamic, I was often the one that was more pragmatic, pushing more to be like, okay, it's good enough, let's get it out.

Sure. I remember in my first few weeks when I joined, I think Ivan was working on a tweet storm for relations. Remember that? We had 10 tweets and I think I spent all of 2 days just working on that tweet storm.

Yeah, this is a good tweet storm. It's about the history of relational database starting from IBM paper to Larry Ellison, how they made Oracle, because the paper from IBM, right?

It's your little Adam Curtis documentary.

I know people should. I still think that's good. Yeah.

So let's talk a little bit about, I guess 2016, 2017, we have this early version of Notion that's starting to hit product market fit. I think a lot of founders go through this, this journey before they get there. I guess as you all started to see a little bit of a pull from market, Can you talk a little bit about how we found our first customers, how we stayed close to them?

I think sort of go back to what Simon was saying. Maybe it's less about the word funding, it's more about the word of executing to something you taste is there. You know you can build a better productivity tool, better programming tool, better text editor. It's just really hard to get there. There's people been building text editor for 3, 4 decades, but you know you can build a better one. So it took us many tries there and you're confident that once you get there, it's a better thing. Some people will realize that and I think that's the thing we'll just keep repeating, repeating notion. It's a good document knowledge, thing backed by a graph data structure is buildable. So once built that, you can see all people actually coming for it. People have discovered the value that we try to inject into the world and launch on things like Product Hunt, which is actually the first place that we pushed the 1.0 out. And from there, talking with people on Product Hunt, talking to people in support tools, realize actually they find true value out of this. And they have similar tastes as our tastes. And so I wouldn't say we stumbled upon it. I just realized it feels like we finally can execute to the level that our taste is there.

The need was already there, but we got to meeting their needs.

Yeah, a better mousetrap. Mousetrap, it's useful. People know what mousetrap is. Can you build a better one?

Not all companies can do this, but we definitely have, which is if you can, it's better to build for yourself at first. You're the person designing and using it as you're building it. And I think to this day, a big percentage of it is we're building for ourselves. We're trying to make a tool that we actually want to use every day and we were the first users. And then after we launched, it's kind of tacking onto that. Okay. Also there's other users now. We have to talk to them in customer support and aggregate their feedback. And so it's kind of building from that core of you actually like it and you want to use it every day and then over time filling more gaps because people use it in ways that you didn't expect or understand.

It's pretty touching actually. Remember we're working on support and—

I was just thinking that actually in my head.

Yeah. And those people that you help solving, they give us feedback, you help troubleshoot their problems. And the moment that we launched back in the day, Product Hunt, you read our comments, those are familiar faces, the people you're having help, they help us, we troubleshoot them in support, they come support you launch the product. It's very much a community feeling of just digital version of that.

I think it might be worth actually talking about that time, at least when I joined, I feel like all 3 of us and the broader team spent so much time in support. And I feel like that was just talking to customers every day. In many cases, I remember some fun stories where there was a bug introduced and a support person would just walk over to Simon and Simon would fix it and then we'd go back to the support person and be like, refresh, try it now. And it sort of fixed it and it's sort of like there was this early customer love and obsession that sort of was felt through the company. And it seems like in some ways by the time I got there, it had already been something that you had designed. Thinking very closely how what support person is doing, what an engineer is doing and the work, the way that gets connected was very strong. Yeah.

On one end it's worth at the end of the day building a tool. Tool has to solve people's problem. We're all human with similar hands, similar feet. So if it doesn't feel good on other people's, hold that a little bit and feel that pain and solve it. And how do you scale this when you have hundreds, not just one or two requests, but hundreds, thousands of that? Can you aggregate into something, some kind of report using system to translate that qualitative feeling? How does this pen or hammer fit in the hand into something that you can stack rank the needs? Because Notion, they're so open-ended. There's so many different requests from so many different type of people. So segmentation, aggregation, tally, create system to do that job rather than you read through 1,000 tickets to do that job.

I just remember you made me turn on my push notifications for Intercom.

Intercom, right?

Yeah. It was just like you'd be buzzing every 3 seconds with a new customer ticket coming in and you'd be like, keep it on because that's the way you're going to build intuition.

Intuition, right? Well, you can turn off the sound.

One thing we did in the early days and I still do this, so useful is once you aggregate all, anyone working on support would actually talk to the person back and forth to try to dig a little bit deeper in what they wanted. It's often a feature request or something that we don't have. And then all those conversations would get tagged and then aggregated. And then if we decided to prioritize one of them, we basically had instant extensive user research where you could just go for some new problem you're trying to solve, look at that tag and read 100 Intercom conversations and within an hour you now pretty much deeply understand the problem you're trying to solve. And it let us just really quickly add new features that really went deep and solved people's problems, which is super useful.

There was a feeling and I remember the ethos that we had around just systems building that I think runs through the company even today in many ways, which is I think let's not throw people at problems. There's ways that we can build better systems that— and I think when I think about these early days, it wasn't like we had a product ops function which was aggregating all this feedback or a research function was talking to all these people who have these problems. We built a system, designed it in a way because we were only 10 people, that sort of got us some of those use cases without having to hire those people.

It's not just Notion the company, but Notion itself, right? What are the building blocks? Can one building blocks like Lego used by so many different things, can you try to have the minimal number of building blocks and minimal number of interaction and communication? In some sense, collaboration is friction, You need to talk, you need to generate heat through friction. It's energy loss. So the best way is don't collaborate, think information just flow to the right places and that's through better system design, finding the right primitive in whatever problem you try to solve. Product try to do that. Ideally company try to do that so you can keep the company lean.

Yeah, I mean some of it's conceptual and some of it's also just born out of necessity of having a very small team.

Yeah.

You just have to do more things with fewer people within engineering. We were only 3 engineers for a very long time or less and we were very obsessed with when we built a new feature or added something, it had to fit in with the rest of the system. We had to build it properly the first time because if you don't build it properly next week, you're going to still be working on it. So you have to be able to build it properly and move on to the next thing. And that means putting a lot of craft into it and designing the system so things interconnect pretty well.

I think in the earlier personas or customers we got were a lot in, I guess what we call EPD, engineering product design. But in many ways Notion is a tool that works for all different functions. In the early days, did you all think about a specific persona or was it more general?

It's more general. I think because we are engineer product design, we build for ourselves. The block model is kind of like engineer designers thinking way of using a text editor. The designer has drag rectangle select and select multiple images that you move at the same time. sort of built for that and that clicks with engineer product design who are system builders, system thinkers. Yeah.

But the aspiration has always been, I've been using this word ubiquity all the time that just our aspiration is really to any person be able to use it in the world. Even though the vibe is a bit technical maybe due to our nature, we tend not to build features that explicitly constrain that. It's like if anyone wants to come and use it, we welcome that.

So I sort of feel like a lot of startups actually face this, which is they want to build a general purpose horizontal tool, but sometimes that actually becomes a crutch for them. It's like if you're trying to build something general purpose for for everyone. And I feel like we've actually faced a little bit of that tension internally too. It's like if we start to think about every persona, every use case, it sort of makes us a little bit, it's hard to think about what do we build next. And I wonder if both of you can talk a little bit about how we've sort of still ended up with a general purpose tool, but internally we've used some constraints to sort of build the right product.

Our approach is typically to, when we try to solve a new problem, our approach is typically to first try to deeply understand the problem, the people that are having this problem. Maybe you think about multiple personas and then in parallel think about the solution space, which involves breaking down the problem into its constituent components and trying to express your solution ideally as some kind of flexible primitives that can solve your problem. And then there's a maze of trade-offs you have to navigate to connect this, the problem and the persona you're trying to solve and your primitive solution. And it's really hard to do that, but yeah, it's just kind of grind away at that process. And as long as you really care about both sides of this, I think you often can arrive at solutions where you can express a primitive that's very broadly useful but also solves a specific problem that—

And ideally solve 2+ personas at the same time.

Yeah, yeah, that's right. And yeah, I think you can only really do this if you're just really obsessed with both sides and not jumping to conclusions too early on either side. It's easy to The PM mindset is to jump to conclusions on the persona or problem side and then the pure engineering mindset is the opposite, pure primitive. It doesn't solve a problem and you have to kind of, yeah, it's a tricky maze to navigate, but it's super worthwhile I think to try hard.

So in that framework, which one are both of you? It feels like maybe Ivan is more on on the PM side and then you're more on the engineering side.

I think we're kind of similar.

I think that's too simple. Yeah, I think we both have parts of ourselves that are expressing both of these things and he's better at keep the system.

He's more pragmatic. I'm better at connecting the dot, but he's good at connecting the dot too. It's trade-off all the way down. So there's a trade-off on the UX side, okay, what do people recognize? What's normal for people? The simplest system tends to be the one that people are most familiar with. It doesn't mean it has the fewest parts, but the one that people are most used to. Then that's the UX trade-off. Then you have UI trade-off, which is go to the pixel land. There's only 20 pixels or 40 pixels for you to wiggle, put your icons or text in. What do you put? Should it be an icon? Should it be a text? Is it too much of something will look ugly and look busy? Then you'd make that kind of slightly smaller trade-off. And on the more technical side, there's trade-off what's buildable, what's not buildable and what's real-time, what doesn't have to be real-time. Then you knock on each of the door, then eventually hopefully find there'll never be a perfect solution, but will be one that satisfies the trade-off you care about more or less the most. And then that's the one we go with.

But as you're speaking through that, it does seem like I think you need a little bit of a time to sort of explore that maze, figure out the right trade-off between conceptual and what will land.

Yeah, I mean, I think what's really important is you need the freedom to not ship until you've found something good. If your only constraint is time and you have to ship something by a certain time, it can be really hard to do that. But as long as you can Maybe you can have a date you want to get it by, but really what you care about is getting it right and you'll push the date if you didn't get it. I think that's really important.

I think another key is remove the boundary of skillset and roles and the team structure because we're just talking about the trade-off design of UX, of users' problem, of technical things. they tend to map into an engineer, a designer, a PM. And go back to the collaboration friction, the communication friction. It's really hard for me to dump my thinking into another designer or PM through writing or talking, but it's a lot easier if you have the multiple skillset in your own head, then you can make that trade-off really, really quickly and in a very creative way that connecting the dot that separate individual person cannot do that. So in some sense we're trying to design our company in similar ways. So designer who can code, engineer who can think about product that help break the boundary, move faster and come up with new solutions.

I definitely feel like that's sort of scaled as we've become a larger company is that we're less like you're a designer or you're an engineer or you're a PM. I feel like so many engineers who are PMs here and so many designers who also code and so many I guess engineers who also have this great design sense and we like it that it's blurry at Notion.

I think blurry is good.

Blurry is good. I want to talk maybe next about international. It feels like we launched in Product Hunt and literally from day one we've been a very global company. global user base, business is also, I think maybe I'm curious in the early days when you saw all these people from Europe and Asia also using it, how did we go about supporting them and then how did we scale that to all the people that we have now around the world?

Yeah, keep your Intercom notification on. I think it used to be 1-minute average reply rate, right? Reply time, median reply time.

I feel like the first 6 months I joined, all I was doing was support and I felt like, oh my God, you guys have hired a very expensive person to do support.

I mean, from a technical perspective, we didn't do much in the early days. Our product wasn't localized. Our servers are all in the West Coast, so it's kind of slow to load it from another place. But Yeah, I think it's just about just trying to make a really good product and not close down the distribution, I guess. We're really leaning in on internet distribution, right? Everyone uses Product Hunt. It doesn't matter what country you're from.

Most startups are pretty English speaking, so the language is less a barrier.

Time zone might be, but still 24-hour support is Also, I think related is we're so blessed to have this incredibly passionate community of builders. I guess when you look back, was there a time that you really felt like, wow, this is so special what we have? The fact that we have this community of people who are just crazy power users pushing the product in ways that even we can't imagine.

here.

Was there an event?

I remember that time we went to Singapore, right? We're trying to go to Singapore to find where's the APAC office, Asia office. I think you tweeted.

You tweeted. I tweeted.

You tweeted on a Saturday afternoon or something.

I think it was Saturday morning. I said, hey, we're here for 36 hours and something like 30 people showed up.

Just 30 people showed up in this rainstorm, but it rains a lot in Singapore anyways. I was like, wow, did not expect that. And generally for me, walking into even a different city and just see people going to a coffee shop and say, wow, this person's using Notion. No idea.

Never gets old.

Never gets old.

Yeah, I'm always really blown away by the crazy setups people make and I'm personally not a huge a productivity person. I kind of code all day and take simple notes and have a to-do list, kind of super basic in a way. But yeah, I've always been so blown away by the stuff people make and maintain and yeah, it's super inspiring.

Yeah.

Maybe switch gears a little bit. Talk a little bit about just the company culture and the values we've built. We're close to 600 people now. Maybe start with you, Ivan. I guess I'm curious, maybe in the early days when we were maybe 30, 40 people, we went down this exercise of writing our own values and it was a very long process, but I feel like it's one of those things that you deeply cared about and went super deep on. Maybe talk a little bit about that exercise, but in general how you've thought about scaling company culture.

Who said this? The execution eats strategy for lunch and culture eats execution for lunch. And then end of the day it's the culture, the norms, the intuitive principle of company. I think maybe you nudged me or someone like, hey, actually write it down.

Might've been Jeff Weiner.

Maybe Jeff Weiner. Jeff did it. Write it down with our 20, 30-ish people. Many great companies, they had it written down somewhere. A lot of our internal, I remember the best one is one of the best one is IKEA's one because they had a company running for maybe 20 years, 10 or 20 years, did not write it down. Then their founder wrote it down into this 10 Commandments of IKEA, I think written in the '70s. And their mission statement is kind of in some sense similar to ours, but in the physical space, the better software, a better life for the many people through furnitures.

Amazing. Actually another one I remember from that IKEA had, I think it was IKEA was revenue is a resource.

Revenue is a resource. Revenue is not the goal, it's the mean to the end. We sort of stole that, Not in Arabic. Yeah, it's a good one.

What about you, Simon? Maybe engineering culture and how we've tried to build that here.

On the values, that's interesting on that process is you want it to be some balance between a thing that you already are and a thing that you aspire to be and it can't be too much of one or the other. So yeah, lots of revs on that both for the For the whole company and engineering. There's definitely a bunch of principles that we had in the early days that I definitely sought to maintain a lot. One of them I like to call attend to the garden. So it's like if you're going and working on fixing some bug and you notice something that's kind of confusing or it seems broken, just take a few extra minutes to go fix that thing. And if everyone's tending the garden, here and there, we have a much better system as a whole. Another one is not having rigid boundaries. Basically anyone can edit any of the code. It's not like some other team's code that you can't edit and you have to go ask them and submit a request and then wait for 2 weeks or something like that.

Yeah, definitely feel like ownership, which is actually I guess first value for us, we're all owners of our mission. is something that we've definitely pushed so that people don't create unnecessary boundaries. And I also really liked your point around, I think the first version of values that we came up was very aspirational. It was something maybe we want to be as opposed to more authentically us. And that's also seemed a little bit like as the company grew, we also have a better understanding of who we are authentically that shaped the final values that we have today.

It's not final though.

It's V2, probably ever evolving. But yeah, it does feel— the latest one did feel like something clicked where it was like, oh yeah, this actually just feels like us a little bit more. I remember there were versions like, oh, that's not quite authentic us.

Didn't stick. It was in a doc but never used.

Yeah. I remember being really frustrated on these revs, just like, why are we spending so much time on this? But it's quite useful, I think.

And maybe one more related thing in terms of company building is recruiting. Maybe Ivan, one thing I've learned from you is just at least the early days I remember unless it was a strong yes, it was always a no. I think maybe one of the sayings now inside the company is when in doubt, say no. I think we cared a lot about talent density. I think you cared a lot about that in the early days and maybe you can talk a little bit about that and the importance of that when you're building a startup, a young company.

It's kind of like when you're buying a shirt, if you have a doubt, unlikely you're going to wear it. So when you have a doubt, put it back on the shelf. Going back to the—

Actually that analogy pushing it I think people just buy a lot of shirts, but I feel like you care about that. You care about each shirt you're wearing almost to an extreme version of that inside the company. You've cared about even today, every employee who joins here, you review them and you go through their references. And so it's a good analogy because I think it applies to your pace.

I have shirts at home I don't wear, so just don't get me wrong. So I do have shirts I don't wear. Maybe I should return them. But I think people join company, the cost is a lot higher. It's like going back to communication. Each person needs to talk to every person around them. They bring their life, they bring their work, bring their opinions. It's very costly if we don't make the right fit. It's not good for Notion, the company, it's not good for them. So it's better just to hire right or partway early. And going back to the primitive concept, can we have the least number of pieces, people working at Notion working on this problem and solve through systems, through a better organization, better culture to solve the company mission? The problem we try to solve, it's connected. So why I'm building a software, try to do that. Building an internal process trying to do that. And the other end, can we find the right people hiring to do that? Same, similar.

I feel like I've been here 5 and a half years and I think a lot of people ask what are some of the best decisions or worst decisions? And in my mind it always ends up being people. It's like finding the right people have really sort of helped us come where we have. Okay, let's talk a little bit about the future. Simon, AI has been such a big part of what we've done in the last year and I feel like you've been the person pushing for AI even before that. You could see it coming. Can you talk a little bit about those early days? This is going to be a few months even before ChatGPT. You saw the wave coming.

Yeah, I think so. In 2019 we got a demo of GPT-3 and it was, yes, super early at the time. This is before they did any of the even instruction tuning, I think. So it was really just raw language model and yeah, we got a demo. We weren't super impressed. I remember we were with an engineer and one of them asked, why are there so many chipmunks in Colorado? And it said, because the government conspiracy.

Probably true.

It just completely made something up and we were like, oh, this seems like not something you could put into a product in a reliable way. So At the time I didn't think that much of it, although I started watching the space and yeah, just watching it super closely. And I think beginning of 2022 we started to see the image models of DALL-E, Midjourney. So I was playing with those a lot. Those are super fun.

Yeah, you got obsessed.

Yeah, those are so fun. I would just be up at night making lots of variations on different things and yeah, around October 2022, we played with an early version of GPT-4 and the language part really clicked in then. I think it was pretty obvious that it was a big step change. It went from not really following instructions and knowing much to being able to follow quite complex instructions and having pretty comprehensive knowledge about everything in the world. It's kind of crazy. I remember we were just on our company retreat to Cancun and we were playing with this on our phones. The chat was like, oh my gosh, this is better than Google. You can just ask for anything and it can follow instructions. Yeah, we immediately got to work trying to figure out what we're going to do with it and started building stuff.

So at this offsite, you were there. We were supposed to hang out with all our employees and the two of you decided to lock yourself in a room and essentially get this prototype out that really inspired the company to really invest in AI. And I guess Ivan, for you, how important is AI as part of our future as the future of work? How do you think about what it's going to do in terms of how people will collaborate and do their work in the future?

It's everything. It's like at the end of the day we're building tools, tools for solving information problems. What language model can do is it understand the information now. Most knowledge work, it's a piece of information come to someone, then they try to reason, think about, then find creative way to pass to another person if it's teamwork. Language model can do some flavors of that. Actually the most mundane repetitive part right now we're building tools to help people, whatever the kind of information knowledge related work. Essentially kind of feel kind of lucky in some way. So like spend five plus years building a text editor, spend three four year four five years building a relational database. Those are fundamental primitives to shifting information around, solving information problems. Then all of a sudden there's a new car engine drop in, that's language model, new type of Lego blocks drop in. It can do things in much more interesting creative ways, have a sense of reasoning baked into this kind of Lego bricks. I think it's completely changing how I think, how we think about product and Notion. I think it deserves changing how the industry thinking about knowledge work, thinking about tools to start with. Software started by solving this problem. Tech industry started for solving those problems. I believe it will change. It should change the entire tech industry and it already is.

Couldn't agree more. At the same time, it does feel a little bit dizzying. I feel like you could be on AI Twitter and just It feels like, I don't know, every day it's changing so much. Simon, you're on Twitter a lot. How do you keep up with what's going on in the world of AI? How do you sort of make sense of it? How do you pull the right threads to put into the product?

Yeah, I try to stay as up to date as I can with just what is the public frontier of knowledge on what is possible. Every time I open Twitter, I pretty much always see something. It's like, oh, or it's something that's like, oh yeah, I should read about that. I should probably know about that thing.

Yeah.

So yeah, and I think it's just continuing to stay really curious and excited about it and trying to avoid getting too burned out by it because it can be really intense. So there'll be, maybe there's periods where I use Twitter a little bit less and just kind of read books instead and then kind of get back into it. But yeah, it's only going to accelerate. So Yeah, Twitter's a really important lifeline.

It's up and down, I feel.

Yeah, I think right now it's a quieter— it's a little bit quieter right now. And a lot of things have gone kind of private, so the labs all stopped publishing.

It's such a new feel where it's almost like you need to play, you need to explore things a bit more. You just don't know what's going to work. You're trying to come up with experiences that you can't be very prescriptive about what you're trying to do. And I think it's definitely something I've seen with our own products is we need to play more to sort of figure out what might actually make it to market.

Yeah, I think that's very true on the technical level too. It definitely, experimenting with language models highly rewards creativity and exploration. No one really knows how they work, so there's just so much low-hanging fruit. of just different techniques and you can very often have some crazy idea. It's like, I wonder if that would work. And then you can just try it that day and be like, oh wow, that actually does work. That's crazy. I experience that once a week probably just like, oh yeah, I wonder if that thing would work. And then yeah, so it really rewards just being very kind of open-minded, free thinking about what's possible and just trying random stuff.

I like that you use the word empirical a lot in the past year. It has to be empirical.

Yeah, right. And also, I mean, the flip side of that is if you have an idea, you don't know if it's going to work until you actually try it. And there's also, you can't really say anything for sure. Does this prompting technique work? Well, I don't know. Did you try it in that situation or not? Did it work? So yeah, I use that word a lot now.

Your favorite word, empirical.

It's my favorite word now.

Empirical evidence.

All right, so maybe to wrap it, we ask these 2 questions to all our guests So I'm going to ask you this. Maybe we'll start with you, Ivan. Describe a day in your life, maybe your customs or rituals that you do. What keeps you going?

Wake up, drink the vitamin water that my mom gave me, make another version for my fiancée. So we both drink it. If my mom was watching, so do that.

She's going to be happy.

She's going to be happy. Yeah. Do that exercise, take a shower, come to work. We usually come around the same time.

I think we all like coming to the office every day.

Yeah, around the same time, usually meetings and usually give me a ride home. Then I'll be back to start doing some real work, thinking, writing, follow up Twitter actually is a good point. And read before I go to sleep. Then the next day on weekends, I have more time to do deep work and reading.

And you really like fancy dinners, right? Like 3-hour dinner.

No, I hate fancy dinner. Burrito, it's better.

I feel like Ivan's always like, why are we here for so long? Can we just have a burrito?

Yeah, truly. You like burrito too.

Yeah, much prefer a quick dinner.

I feel like running is like you do a few times a week.

Couple times a week.

What about you, Simon?

Yeah, I wake up around the same time every day. I either will exercise, I've been doing more strength training or just hop in the shower and just go straight to work. And yeah, mornings are pretty productive for me, so I try to come to the office and just focus if I can. There's always a lot of meetings. I tend to be pretty hardcore about not going to meetings where I don't think I will add value or it's important for me to be there. And yeah, I think I've gotten better at that over the years. So I'll say no to half the meetings and then I'm pretty heads down coding every day these days.

You're lucky, right? He's going to stay up.

Yeah, I know.

We're meeting people.

It's fun.

I like coding.

Yeah.

I mean, also working with the team and going to meetings as well, but only the important ones. And then yeah, come home, usually wind down at the end of the day, just hanging out with my wife, just chit-chatting and I like to go for a long walk in the evening and I always read in bed before I fall asleep. So it's my main reading time is just the time between I lie down in bed and I can't keep my eyes open anymore, but I can actually get through a lot of content. So that's how I read all my books pretty much.

You gave me that hack, which is get the screen brightness down to—

Oh yeah. And I read on my phone. Yeah, I used to use a Kindle, but I went traveling one time without it and It's like, oh, actually the phone is even better, lower latency.

Maybe the last question for you. If you were to write a book about what's gotten you to this point in your career, what would you title it?

It's a very interesting question. I think I would call it Start Again. So yeah, kind of calling back to some ideas we already discussed, but yeah, I think Doing a company like this is a super long marathon and there's always endless new problems. And in order to last through it, you have to have the ability to every day wake up and be like, what is the thing I'm going to get excited about today, this week, this month? And don't get too dragged down by the problems of the past. So start again every morning, every week, every month, every quarter and keep going.

I like And you are so good at wipe the whole part of that code base, throw away the code, then rewrite it.

Clean slate. Yeah.

Clean slate. Most people, oh, I wrote this, I don't want to throw away.

So I'm just going to rip it. Yeah. Some of that is drop your ego and just accept the mistake of the past. That can be hard, but it's a thing to aspire to.

I like that. What about you, Ivan?

I think for me would be perspectives. Truth is multiple folds and connecting the dots, making trade-offs, you need to see different version of the truth. You need to have different perspectives. I really like this quote. I think it's from Alan Kay, one of the computing pioneers. I think it's from Alan Kay. A perspective is worth 80 points of IQ. Or something like that. It's not about your raw intelligence, it's not about your raw CPU, it's about your taste, your perspective, your outlook on things. And by consciously changing that, you can discover new solutions, making something more interesting. And then sometimes that's what creativity is. I think be self-aware of that perspective, consciously change that. It's something I personally, I really enjoy doing.

Amazing. Start Again and Perspective. Great titles. Well, both of you, thank you for making the time. It's an honor to work alongside both of you every day and I think this was just so much fun to get to sit down and chat. I think we should do it again.

Yeah, let's do it again. Sounds good.

First Block is brought to you by Notion for Startups. We at Notion care deeply about startups and founders. And we hope these stories inspire you to keep building. To learn more about how we are supporting startups, please visit notion.com/startups.

Thank you.