Notion's CEO wants you to demand better from your tools

26:36Because we're also inspired really much by the craftspeople in Kyoto, how they dedicate their time to build something, not just for money, not just for fame. That is so crazy. But I was actually also in Kyoto last week. I was on vacation and went there for the first time and had an incredible time. Kyoto is amazing. You got a sense. It's a little bit slower pace. People care about the thing, right? People truly care. That's the thing. It's the main thing. It's not the business. It's not the outer surrounding. Oh, absolutely. I mean, like you go to these temples that are, you know, a thousand plus years old and the care and the craftsmanship that they put into them is truly inspiring, right?

27:21It's deeply beautiful. It's, you know, very connected to their spirituality, their religion, the culture, the history. So I could understand why, you know, a founder would go there and take a lot of inspiration. It's a bit slower too, so you can focus on the thing in the virtual space, computers. Yeah, it's not like San Francisco with our go-go nightlife, our Waymos, the party scene, all of that. Or New York, even more of that. One more of these, this isn't strictly a decoder question, but it feels like spiritually decoder question that I wanted to ask you. What is the best productivity tool that you use that is not made by Notion?

28:02I like those chatbot products chatgpt and tropic cloth yeah it's quite amazing especially if talking about feature the conversation mode right I love those you like the voice mode voice mode yeah the voice mode like helping you learn a lot of different things I can just when I'm making coffee waiting for the water to boil just talk with this thing for a little bit for two minutes right perfect what are you asking about what do you like to learn about oof all kinds was most recently. In Japan, I was reading a book about Marshall McLuhan, which is this Canadian media theorist, theologist. Many of his concepts are hard to interpret.

28:47So it's better just to work through with a language model that help you guide. It's the best tutor. Education should be very different, hopefully be very different a few years from now. I think so. This isn't quite a tutoring use case, but I just have to say, you know, I was in Kyoto last week and we were in the neighborhood and kind of had some time to kill. And so I just opened Google Maps and it, you know, sort of opens to where we are. And I took a screenshot and I just sent it to ChatGPT. And I said, tell us a bit about this neighborhood. And it gave me like a history of the neighborhood.

29:18It told me the cool restaurants, cool coffee shops, like a museum, places that we could walk to. I mean, like it truly was as good as I can imagine getting from any guide. And it was as simple as uploading a screenshot. The whole thing took 15 seconds. Like it was, it was wild to me. Especially like one use case I have, like when, if I go to a famous architectural building and say, Hey, I'm at this place, tell me about it. Oh, I'm looking at this part corner of this building. Tell me more about why is that's the case, right? Because if it's famous enough, it's probably part of the corpus of training.

29:50So they, The language model knows about it. You can just truly a guide tour and you don't need another person. You're just talking with your machine. We need to take another quick break. We'll be right back.

30:04Support for the show comes from Rippling. If you're a business owner, here's the truth. SaaS promised to make work easier. But now the average company is buried by hundreds of apps that silo your teams, slow you down, and simply don't work together. That's not SaaS. That's sad. Software as a disservice. That's why you need Rippling. Rippling is the unified platform for global HR, payroll, IT, and finance. They've helped millions replace their mess of cobbled together tools with one system designed to give leaders clarity, speed, and control. By uniting your employees, teams, and departments in one system, Rippling removes the bottlenecks, busywork, and silos your software created.

30:48Automated, perfectly in sync, and seriously simple to use, Rippling gives your company one source of truth for your people, their data, and everything they touch. With Rippling, you can run your entire HR, IT, and finance operations as one. Or pick and choose the products that best fill the gaps in your software stack. And right now, you can get six months free when you go to rippling.com slash decoder. Learn more at rippling.com slash decoder. That's rippling.com slash decoder for six months free. Terms and conditions apply.

31:32We're back with Notion CEO Ivan Zhao. Before the break, we were discussing a major decision Ivan made to reboot Notion a decade ago, one that helped him keep the company alive and eventually grow it to profitability. But now I wanted to ask him about Notion and AI and how he sees his productivity platform evolving alongside the new technology. Well, that seems like a good segue into Notion and AI. We've talked about what Notion is, how it's changed. Notion now builds itself as the AI workspace that works for you. So what does an AI workspace mean to you? What do you want it to be for us. If you think about our strategy during the SaaS era, it's bundling, consolidating a bunch of different tools for knowledge work into one place.

32:18And what's changing in the past couple years is now with all those soft knowledge work Lego in Notion, you can not only provide the tool, but you can assemble them as your AI teammates. They can do the work for you. And we're fortunate to have those knowledge work Legos in one place, so you can piece together in a very interesting way. One end can take notes for you. The other end can help you manage triage projects. Writing documents, those are basic things. But more and more with more Legos and smarter models, you're essentially hiring Notion as your AI teammates. That's the future that we've been building or building more towards.

32:58Yeah. I remember one time I was meeting with you and you just launched some of these AI tools, and you were showing me that Notion AI was taking notes about various meetings. And so you were able to dip into meetings at the company that you did not personally attend, and just kind of quickly catch up on what your co workers were talking about. And I thought, that's super interesting, right? That's like some that's the kind of feature that I can imagine a lot of CEOs wanting. But before this point, they haven't had that level of visibility into their own company. We just launched three separate products a couple months ago, Notion AI for work, including the next version of this enterprise search product you talk about.

33:39So along with that was we launched AI meeting notes product. So all the meetings, you can record and transcribe that. So essentially, your company has a collective brain of what's going on, and you have all the AI knowledge workers on top of Notion to help you transcribe the meeting notes, answer whatever question you may have. It's quite interesting what you can do with the technology now. Of all the AI features that you've added so far, which ones do you personally find the most useful? I use AI meeting notes. Almost every meeting except this one I record. And I use that for meetings so I can share notes with other people.

34:17I use it for myself as a starting position to dump my thoughts. and I can remember later where I can use this to ask AI to turn the transcription into a writing. English is my second language, so I'm not the fastest writer, but AI can do a better writing than I can if I just dump out what I'm on top of my mind through the transcription AI meeting feature. There's a lot of talk right now about AI and whether it might replace workers or entire workflows or functions within an organization. You've talked today about AI being able to serve as a kind of teammate. Do you think that AI and Notion will get to a point where executives will hire fewer people because Notion will do it for them?

35:02Or are you more focused on just helping people do their existing jobs? We're actually putting out a campaign about this in the coming weeks or months. We want to push out a more amplifying, positive notion about what they can do for you. So imagine the billboard we're putting out. It's you in the center. Then with a tool like Notion or other AI tools, you can have AI teammates, right? Imagine, hey, you and me start a company. We're two co-founders. And OpenNotion, SignUpNotion, and all of a sudden we're supplemented by AI or AI teammates, some taking notes for us, some triage, some doing research while we're sleeping.

35:41So all of a sudden we're a team of a company of 10 people. So then the startup can run much faster. That's the notion. That's the vision we want to push more towards the world. It's more of an amplifying force rather than a zero-sum force. And what time frame do you think that arrives on? Does that feel like it's almost within reach? Or do you think we're going to need to see several more research breakthroughs before that sort of thing becomes possible? From someone who's building with this every day, I think a capability, it's pretty much there for, there are different spectrum complexity of knowledge work, right?

36:22The model is quite smart. I would say what's lacking are the plumbings, the toolings that's unlocked the capability of the model. That's essentially what Notion is doing with the Lego blocks, be the plumbing and tooling. So that's one constraint. The other constraint is just how people use it, how people plug into their workforce. Like bureaucracy, sometimes it's a good thing, sometimes it's a bad thing. In this case, I think it's actually a good thing because us to slow things down a little bit, give people the time to adapt to learn with this new tool. I think it's good. So the capability, it's more or less there.

36:57And if not, every three months, you've got a new one. So the trends keep coming, right? At the same time, I think the biggest flaw in the AI models that we have today is that they're not reliable. They don't answer the same question the same way 100 % of the time. And so if I'm relying on it for mission critical stuff, if it's one of the 10 quote unquote people at my company and I tell it to go grab some facts and figures and it just kind of hallucinates the wrong one, that's like really bad. Like, you know, if that were a real worker, I would, you know, I don't know, put them on a performance improvement plan or something.

37:36So how do you think about reliability as a challenge to what kinds of services you want to offer people? It's definitely an issue and getting, I would say getting better in generally. I would say the best closest mental model is like a treating language model, just like a human, just like an intern. Human make mistakes. Your trust level, when you tell another human something, there's nothing guaranteed this human cannot mess it up or tell another person even though you don't want to. So fundamentally building trust this way. People's expectation for software is higher because software for the longest time, it's always doing.

38:13if there's no bugs, they always do exactly what you told, right? AI is a new type of software. Our expectation hasn't been set how to deal with this yet. I think as more people getting used to it and as we learn, change our habit around it and company change their workflow around it, I think we'll find an equilibrium that's like amplify the better part of this technology and deal with the short profile of it. Yeah, I think the podcaster, Dworkesh Patel, said something like, an AI today is better than an intern on day one, but worse than an intern on day five. Because like, on day one, they have all knowledge of human history, and they can sort of dazzle you with their capabilities.

39:01But also, they have trouble learning. And it's hard to sort of show them how to do something once and then have them do it reliably every single time. Whereas a human being could do that. So I'm personally very curious, what is the point when an AI is better than that day five intern? I think all companies, including Notion, are trying to figure out a technique to inject memory and learning into this intern. In the coming quarters, you will see product with this baked in. Okay, now my ears are perking up because it sounds like we're getting a little bit of a preview. Do you want to tell us what you're working on over there?

39:36A couple of months ago, we launched this Notion AI for Work as Enterprise Search. It has AI media notes, has deep research to help you draft documents. The upcoming product, you can actually imagine each one is an AI intern who can do a specialized thing. The upcoming product, you can actually create different flavors, AI intern and teammates that live with you in your workspace. That's as far as I want to share. And Notion essentially can do everything you can do, a human can do. I like the sound of that. let me ask you about this is this is my product request and you know this request because i've had it for a while now but basically when you first started adding ai features in a notion a few years ago i put in this request because every link that has ever been in my newsletter platformer is stored in a notion database in many cases that includes the full article text and what i want is to be able to have a conversation with that particular database right it'd be so useful for research and brainstorming columns at the same time it's you know i don't know hundreds of thousands of millions of words.

40:40It's not the sort of thing that you could easily throw into a context window and just let me have that conversation. So my question, Ivan, is where are we on this dream of mine? To having a conversation with all your thousands of articles in Notion? Yes. It's probably already there because there are techniques invented, correct, that you don't have to fit everything into a context window. You can index everything, embedding everything, and piece information out as you need it, right? There's another technique that's been popular in the past year or so called tool use. It's essentially teach the language model, your agent to know how to use search.

41:22So if you have a question that's not in the contact window initially, the agent can go there to, just like a human, can find more information about it. It will take a little bit more time round trip, but eventually it will give you what you want. So all technique will make the use cases you're describing better. I like the sound of that. And you do have an Ask Notion feature already that I imagine can access some element of what I'm talking about. And a lot of this stuff is just sort of on the web. So there are other ways of accessing it. But I just always think, man, if I could have a lightning-fast way of just chatting with this database the way I would chat with a coworker, that would be super cool.

42:03Oh, it should already be in your Notion workspace. Happy to walk you through the new enterprise search we just launched. It's perfect for that. Okay, great. All right. We'll troubleshoot that offline. OpenAI has come up a couple times today. You work closely with them. Recently, they announced you can use ChatGPT to create presentations and slide decks. All of the big labs are working on these full-stack virtual assistants that they say might someday be able to do anything a remote worker. might be able to do. Do you think that they'll get there in the next, let's say, five years? And what role do you see Notion playing in a world where AI's capabilities are rapidly expanding that way?

42:45One way to think about it on the spectrum of whether it's a more personal B2C flavor or it's like B2B team first flavor. I would say most labs product currently building is more of a personal assistant flavored. It can help you to do work or happy or cheat on your homework. And usually B2C, it tends to be winner takes all or there are a few winners. And I think it makes sense for labs like OpenAI to go really hard towards that direction. Notion, Squarely, is a B2B company. Our product, our business model is for business, other businesses. And inside B2B, there's required making different trade-offs.

43:24There's so many different subcategories and it has to be team first to start. That's why our AI teammates, our Asian AI system, fundamentally live in a team space that you or the rest of your company employee live in. That's the angle we're taking. There will be, in my opinion, there will be many different winners in the B2B because B2B usually is not winner takes off. And you need to make very different trade-offs. That's interesting. I'm still not sure I totally understand. Say a bit more about what does this B2B AI world look like And why is it that that will have many winners, whereas B2C maybe doesn't?

44:01Well, you can think about it in the professional setting, right? So if you think all knowledge work, there is lawyers, there's accountants, and there is programmers, right? There is council of support. And they're all different. It requires you to make a little bit different tradeoff. There are different types of AI agents. You already can see this in the first generation AI agent. They need to be specialized. They need to plug into different contexts. versus the consumer side, you just want to chat with your AI chatbot. It's very universal. That's why in the B2C side, in the previous generation, there's iPhone and Android.

44:35There are two things, largely. In the B2B, it's SaaS. There's thousands of different companies, hundreds of different categories. So they force your product to make, whether it's to be a software or AI product, to make very different trade-offs. You cannot be an airplane and a submarine at the same time. That's why in the professional setting, You see a lawyer's accountant, there's a lawyer agent, there's a financial agent. They behave very differently. They need to behave very differently compared to your personal system that you wake up every day you can chat with. Right now you sell AI tools as an add-on in your business and enterprise plans.

45:11I'm curious if this hurts your margin at all. We hear a lot about how expensive and compute-intensive, resource-intensive AI systems can be to operate. Is it a challenge to integrate those sort of like resource-hungry tools into your existing subscription? We actually recently merged AI into our main plan because more than half our sales now customers want to buy this AI product. So it makes sense to just simplify the whole pricing buckets to just include it into everything. It does make the margin not as good as pure SaaS, pure software, but it's so powerful. People appreciate it. And still, company is cash flow positive, so our CFO loves that, despite it's a different margin profile.

45:57We've already seen some companies move toward usage-based pricing for AI, which as a consumer, I hate. I don't want to make micropayments to ask ChatGPT a question, but it does seem like that is maybe a better business model. How do you think about the trade-offs there? I don't think people figure it out, especially in the business setting, P2B. the first generation is kind of customer support. Customer support, you can map to they call outcome resolution-based pricing. That makes sense. Then there's sort of the second generation, which is saw right now as coding, right? Coding, there is a C-base, but if you use a lot, you have to go usage-based.

46:37That sort of makes sense because your exchange is a piece of work. You get your file, you get your software at the end of the day, right? So people appreciate that and it saves programmers so much time. than actually writing the piece of software themselves. Knowledge work is nebulous. Knowledge work, you can't put a price on a knowledge work. Like it's this chunk of dock, how much is it worth? You can't really put that. And how good is a piece of knowledge work? How good is your product spec? You can't put a dollar sign behind it. So much harder for a general purpose, knowledgeable product like Notion.

47:12That's the thing, the whole industry need to figure it out. What do you wish that AI would make possible in Notion that isn't quite yet possible yet. You can always get cheaper and faster and smarter. But you know the train is coming for that direction, right? So it's a very, it does require you build company in a different way. I think this is like the software industry is realizing this right now. Say more about that. First of all, this is kind of like, I never worked during the dot-com era. That was a little bit before me. People said during that era, the web standard changes all the time. Every couple of months, three months is different, right?

47:50And I never worked with Intel, like Morse Law era. It's like you can just expect the next 18 months from now, the next CPU can drive whatever software you do. AI sort of feels that, but even on steroids, like every three months, the next model, it's almost can do what you couldn't do before. So does it require you really change how you build software and build product and build a company? It's a couple of things. One is because it's constant changing, you can, and the model itself doesn't like too much restrictions. You need to build a harness just around the right places. So almost like if you build too much around the train track, the next train comes, you just make what you just build obsolete, right?

48:35You should build parallel to the train track. That's number one. Number two is like the language model are not deterministic. It's different from classic software engineer. The metaphor I like to use is that classic software engineer is kind of like building train track or building bridges. It's Newtonian physics. Everything's predictable, right? And if you can't imagine it, you can build it. Sometimes it takes three months, sometimes six months, but eventually you can build it, right? With this language model, I think it's squishy. It's organic. It's kind of the analogy I love to use, like brewing beer, right?

49:10You cannot tell the ease, okay, my beer is going to taste this. Please ferment yourself, become like that. You have to channel what's in the model. The best you can do is create an environment, massage the data, massage the context, the hope the best, right? So this requires a much more of an iterative approach. You cannot come from your vision or customer need first. You have to come from what the technology gives you, what the ease, what the beer gives you. So really much allow your team to be more empirical, more experiment, less of this kind of waterfall class way of like spec to code should be more like incremental iterative.

49:49All those add together make force you have designed a different engineer product design work. Does it change the way that you hire? Does it change the way that you structure teams? Like how does that strangeness that you describe translate into a different company? People need to be more okay with ambiguity. People need to love ambiguity. People need to be more experimental. The boundary between role is even greater. I notion we hire a designer who can cope because if you're an engineer and designer in one, you can think a lot more ambiguously, more fluidly. AI time even pushed that even further because the design and product sit side by side with engineers.

50:34oftentimes what you want, you cannot be built. So you have to really try a bunch of different things. That's why you see a lot of product demos like 60%, 70%, but never become a real product. That's because it's good for making a demo, but to get to a production B2B software, you need to be really good and to be reliable. Oftentimes you never get there. I think about this a lot in the context of these voice-based assistants, which today mostly what I use them for is setting a timer or asking what the weather is. like these very deterministic things. And the companies that are building them are trying to integrate these new AI-based backends.

51:10But it's incredibly hard because if the user is still using the product, they're still going to want to set the timer. And if it goes from doing it correctly 100 % of the time to like 93 % of the time, that's like a much worse product, right? But I think as human, we all learn what's the best, what is this type of technology best at, right? It's when you're having a conversation, like how using the voice mode, you want it to be ambiguous. You want it to go to different places. That's a feature, not a bug. I think us as a whole, as an industry who are making the software with AI and us as an audience who use this, we haven't figured out the stance yet.

51:46It will take some time to figure this out. Use the material for the best use. Well, I want to end just by asking what you think Notion looks like a little bit into the future. And I will not ask you about five years from now, because I don't think anybody has five years worth of visibility into anything. But if I could maybe ask for like two years from now, what do you hope NoShit is doing that it's maybe not doing today? I think going back to what we just talked about, the nature of software is changing. It's changing, evolving from just a piece of tools to this organic matter that tool can do some work for you.

52:25And the heart of this company in the SaaS software, classic software areas, allow people to build tools, to allow people to use this Lego to create whatever tools they want. Because the nature of software changing, what we care about allows you to create AI teammates to help you take some of the most repetitive knowledge work you don't like to do. If we can realize that, there's a lot of implication. The next generation of builders is going to run on the company very differently. and I care about solving that problem. All right. Well, Ivan, thanks so much for joining me today. Thank you for having me.

53:04Thank you to Ivan for taking the time to speak with me and thank you for tuning in. I hope you liked it. If you want to let us know what you thought about this show or what else you'd like us to cover, drop us a line. You can email the team at decoder at theverge.com. They really do read every email or you can hit me up directly on Threads or Blue Sky. I'm at Crumbler on threads and I'm Casey Newton on Blue Sky Decoder also has a TikTok and an Instagram you can check those out at DecoderPod they're a lot of fun and if you like Decoder please share it with your friends and subscribe wherever you get your podcasts Decoder is a production of The Verge and is part of the Vox Media Podcast Network Decoder is produced by Kate Cox and Nick Statt this episode was edited by Xander Adams the Decoder theme music is by Breakmaster Cylinder see you next time

53:56refresh your good vibes with tic tac with the right vibes with the right vibes oh man to

54:12now Now try the different TikToks from mint-free to fruity-sweet. TikTok. Refresh your good vibes.