E41: [Bonus Episode - Latent.space] Building the AI x UX Scenius with Linus Lee of Notion AI

[Bonus Episode] Latent.space hosts, Alessio and Swyx, sit down with Linus Lee of Notion AI to discuss Linus’ experience starting the AI/UX community, prompt engineering at Notion, and designing AI interfaces and agents. The latent.space podcast aims to be the first place where AI engineers hear about the latest AI news and technology trends. We’ve had several guests in common – including Shreya Rajpal of Guardrails, Jonathan & Abhi from the recently acquired for $1.3B MosaicML, and also Riley Goodside from Scale AI – and also explored similar themes, including the question of which AI companies have durable business moats, and the flurry of experimentation going into AI agents right now. We recommend checking them out! Keep up to date with Latent.space here: Website: https://www.latent.space/ Spotify: https://open.spotify.com/show/2p7zZVwVF6Yk0Zsb4QmT7t?si=95fb033e3c014975 Apple Podcasts: https://podcasts.apple.com/us/podcast/latent-space-the-ai-engineer-podcast-codegen-agents/id1674008350 Youtube: https://www.youtube.com/@LatentSpace-podcast We're hiring across the board at Turpentine and for Erik's personal team on other projects he's incubating. He's hiring a Chief of Staff, EA, Head of Special Projects, Investment Associate, and more. For a list of JDs, check out: eriktorenberg.com. RECOMMENDED PODCAST: The HR industry is at a crossroads. What will it take to construct the next generation of incredible businesses – and where can people leaders have the most business impact? Hosts Nolan Church and Kelli Dragovich have been through it all, the highs and the lows – IPOs, layoffs, executive turnover, board meetings, culture changes, and more. With a lineup of industry vets and experts, Nolan and Kelli break down the nitty-gritty details, trade offs, and dynamics of constructing high performing companies. Through unfiltered conversations that can only happen between seasoned practitioners, Kelli and Nolan dive deep into the kind of leadership-level strategy that often happens behind closed doors. Check out the first episode with the architect of Netflix’s culture deck Patty McCord. https://link.chtbl.com/hrheretics TIMESTAMPS (05:52) Starting the AI / UX community (12:23) Most knowledge work is not text generation (18:43) Finding the right constraints and interface for AI (21:27) Linus' journey to working at Notion (25:51) The importance of notations and interfaces (28:29) Setting interface defaults and standards (34:58) The challenges of designing AI agents (42:00) Notion deep dive: “Blocks”, AI, and more (53:22) Prompt engineering at Notion (01:04:22) Lightning Round SPONSOR: Shopify is the global commerce platform that helps you sell at every stage of your business. Shopify powers 10% of ALL eCommerce in the US. And Shopify's the global force behind Allbirds, Rothy's, and Brooklinen, and 1,000,000s of other entrepreneurs across 175 countries. From their all-in-one e-commerce platform, to their in-person POS system – wherever and whatever you're selling, Shopify's got you covered. With free Shopify Magic, sell more with less effort by whipping up captivating content that converts – from blog posts to product descriptions using AI. Sign up for $1/month trial period: https://shopify.com/cognitive This show is produced by Turpentine: a network of podcasts, newsletters, and more, covering technology, business, and culture — all from the perspective of industry insiders and experts. We’re launching new shows every week, and we’re looking for industry-leading sponsors — if you think that might be you and your company, email us at [email protected].

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Nathan Labenz?

  Hello and welcome back to the Cognitive Revolution. Today, we have a special bonus episode to share from our friends at the Latent Space Podcast. Hosted by Alessio Finelli, Partner and CTO in Residence at Decibel Partners, and Swix, Writer, Speaker, Developer Advocate, and Creator of the popular Small Developer Project, the Latent Space Podcast aims to be the first place where AI engineers hear about the latest AI news and technology trends. Since launching in February, right around the same time that we started the Cognitive Revolution, they've followed a remarkably similar and at times outright overlapping path in their simultaneous quest to understand everything going on in AI. We've had several guests in common, including Shreya Rajpal of Guardrails, Jonathan and

• 0:43Nathan Labenz?

  Avi from the recently acquired For $1.3 billion Mosaic ML, and also Riley Goodside from Scale AI, the world's first prompt engineer. We've explored some similar themes, including the question of which AI companies have durable business modes, and also the flurry of experimentation going into AI agents right now. My belief, as you probably know, is that the AI capabilities overhang, that is, the gap between what is possible with current AI models and what has actually been made to work in practice today, is huge and arguably still growing, and that much more high-quality AI education is badly needed to help close the gap, so that we can better understand how AI systems work and begin to realize the benefits of AI deployment before even

• 1:27Nathan Labenz?

  more powerful systems are created. With that in mind, I'm happy to introduce the Latent Space Podcast with a guest that I've also tried to book but haven't been able to so far, Linus Lee of Notion AI. Linus is a prolific developer and writer, whose work has been very influential for me and many others in the space. As he's prototyped and publicly shared a variety of writing assistants and other creative user interfaces and experiences, all designed to help integrate large language models into practical workflows. As you'll hear, he's one of the deepest and most sophisticated thinkers about the future of knowledge and creative work. Beyond this, you can subscribe to the Latent Space Podcast feed for weekly updates, including a very lively recent discussion with George Hots of The Tiny Corporation, which, among

• 2:10Nathan Labenz?

  other things, contained some juicy rumors about the nature of GPT-4. I hope for now you enjoy this introduction to the Latent Space Podcast, and we'll see you back next time for another edition of the Cognitive Revolution.

• 2:33Speaker 01

  Hey everyone, welcome to the Latent Space Podcast. This is Alessio, Partner and CTO in Residence at Decibel Partners. I'm joined by my co-hosts, Squicks, Writer and Editor of Latent Space.

• 2:43Speaker 02

  And today, we're not in our regular studio. We're actually at the Notion New York headquarters. Thanks to Linus, welcome.

• 2:50Speaker 03

  Thank you, thanks for having me. Excited to be here.

• 2:52Speaker 02

  Thanks for having us here in your beautiful office. It is actually very startling how gorgeous the Notion offices are. And it's basically the same aesthetic.

• 3:01Speaker 03

  It's a very consistent aesthetic. It's been, it's the same aesthetic in San Francisco and the other offices. And it's been for many, many years.

• 3:08Speaker 02

  Yeah, you take a lot of craft in everything that you guys do.

• 3:12Speaker 03

  Yeah, I think it, we can, we'll, I'm sure, talk about this more later. But there is a consistent kind of focus on taste that I think flows down from Ivan and the founders and into the product.

• 3:21Speaker 02

  So I'll introduce you a little bit, but also there's just, you're a very hard person to introduce because you do a lot, a bunch of things. You got your BA in Computer Science at Berkeley. Even while you were at Berkeley, you were involved in a bunch of interesting things at Replit, Catalyst X, Hack Club and Dorm Room Fund, which is also, I always, always love seeing people come out of Dorm Room Fund because they tend to be very entrepreneurial. You're a product engineer at Idea Flow, Residence at Beta Works. You took a year off to do independent research and then you've finally found your home at Notion. What's one thing that people should know about you that's not on your typical LinkedIn profile?

• 3:58Speaker 03

  Ooh.

• 3:59Speaker 02

  Just on the personal side?

• 4:01Speaker 03

  Wow, putting you on the spot. I think, I think, I mean, just because I have so much work kind of out there. I feel like professionally at least, anything that you would probably want to know about me, you can probably dig up. But I'm a big city person, but I don't come from the city. And so I went to school, I grew up in Indiana, in the middle of nowhere, near Purdue University, a little suburb. I only came out to the Bay for school and then I moved to New York afterwards. And which is where I'm currently, I'm in, I'm in Notion New York. But, you know, I still carry within me a kind of love and affection for like small town, Indiana, small town, flyover country.

• 4:33Speaker 02

  Okay, we do have a bit of indulgence in this. I'm from a small country and I think Alessio, you also kind of identified it with this a little bit. What's something that people should know about Purdue? Purdue chickens.

• 4:45Speaker 03

  Purdue, yeah. Purdue has one of the largest international student populations in the country, which, I don't know, I don't know exactly why, but because it's a state school and it focuses a lot on STEM topics, Purdue's well-known for engineering, and so we tend to have a lot of folks from abroad, which is particularly rare for a university in, I don't know, that's kind of like predominantly white American and kind of midwestern state. That makes Purdue and the surrounding sort of area kind of a younger, more diverse international island within the, I guess, broader world that is Indiana.

• 5:21Speaker 02

  Fair enough. We can always dive in into sort of flyover country or, you know, small-town insights later. But you and I, all three of us actually recently connected at AIUX SF, which is the first AIUX meetup. Essentially, which just came out, came out of like a Twitter conversation. You and I have been involved in HCI Twitter, is kind of how I think about it, for a little bit. And when I saw that you were in town, Jeffrey Litt was in town, Maggie Appleton in town, on the same date, I was like, we have to have a meetup. And that's how this thing was born. Well, what did it look like from your end?

• 5:52Speaker 03

  From my end, it looked like you did all of the work and I-

• 5:56Speaker 02

  Well, you got us the notion.

• 5:58Speaker 03

  Yeah, yeah, it was also in the Notion office, it was in the San Francisco one. And then thereafter, there was a New York one that I sadly couldn't make. But yeah, from my end, it was, and I'm sure you were too, but I was really surprised by both the mixture of people that we ended up getting and the number of people that we ended up getting. There was just a lot of attention on, obviously, there's a lot of attention on the technology itself of GPT and language models and so on, but I was surprised by the interest specifically on trying to come up with interfaces that were outside of the box and the people that were interested in that topic. And so we ended up having a packed house and lots of interesting demos. I've heard multiple people comment on the event afterwards that they were positively surprised by the mixture

• 6:41Speaker 03

  of both the ML, AI-focused people at the event, as well as the sort of interface HCI-focused people.

• 6:47Speaker 02

  Yeah, I kind of see you as one of the leading, I guess, AIUX people. So I, I hope that we're maybe like starting a new discipline, maybe.

• 6:56Speaker 03

  Yeah, I mean, there is this kind of growing contingency of people interested in exploring the intersection of those things, so I'm excited for where that's going to go.

• 7:04Speaker 02

  I don't know if it's worth going through like favorite demos. It was a little while ago, so I don't know if.

• 7:11Speaker 01

  There was, I forget who made it, but there was this new document writing tool where you could apply brushes to different paragraphs.

• 7:19Speaker 03

  Amelia's.

• 7:20Speaker 01

  Yeah, where you said a tone and like both in terms of like writer inspiration and then a tone that you wanted. And then you could like drag and drop different tones into paragraphs and like have the model rewrite them. It was the first time that it's like, you know, it's not just autocomplete. You know, there's more to it and it's not ask it in a prompt. It's like this funny like drag and emoji over it.

• 7:42Speaker 03

  Right.

• 7:43Speaker 02

  I actually thought that you had done some kind of demo where you could select text and then augment it in different moods. But maybe it wasn't you, maybe it was just someone else.

• 7:51Speaker 03

  I had done something similar with slightly different building blocks. I think Amelia's demo was there was sort of a preset palette of brushes and you applied them to text. I had built something related last year. I prototyped a way to give people sliders for different semantic attributes of text. And so you could start with a sentence and you could, you had a slider for length and a slider for like how philosophical that text is and a slider for how positive or negative the sentiment in the text is. And you could adjust any of them in the language model reproduce the text. Yeah, similar, but continuous control versus like distinct brushes, I think is an interesting distinction there.

• 8:26Speaker 02

  I should add it for listeners. If you missed the meetup, which most people will have not seen it, we actually did a separate post with timestamps of each video. So you can kind of look at that.

• 8:36Speaker 01

  Sorry, Linus, this is unrelated, but I think you built like over 100 side projects or something like that. I think there's a lot of people.

• 8:43Speaker 02

  I know it's 100.

• 8:45Speaker 01

  I think it's a lot of them.

• 8:46Speaker 03

  A lot of them are kind of small.

• 8:49Speaker 01

  Yeah, well, I mean, it still counts. I think there's a lot of people that are excited about the technology and want to like hack on things. It's like, do you have any tips on like how to box? Like what you want to build, you know, it's like, how do you decide what goes into it? Because all of these things, you could build so many more things on top of it. Like where do you decide when you're done?

• 9:06Speaker 03

  So my projects actually tend to be, I think especially when you. People approach project building with a goal of learning. I think a common mistake is to be over ambitious and sort of not scope things very tightly. And so a classic kind of failure mode is like, you say, I'm really interested in learning how to use the G P T four A P I. And I'm also interested in vector databases. I'm also interested in next J S. And then you devise a project that's going to take like many weeks and you glue all these things together. And it could be a really cool idea, but then especially if you have, you know, sort of a day job and other things that life throws you away. It's hard to actually get to a point where you can ship something.

• 9:46Speaker 03

  And so one of the things I got really good at was saying, one, knowing exactly how quickly I could work, at least on the parts of the technologies that I knew well. And then only adding one new unknown thing to learn per project. So it may be that like for this project, I'm going to learn how the embedding A P I works. Or for this project, I'm going to learn like how to do vector stuff with PyTorch or something. And then I would scope things so that it fit in like one chunk of time, like Friday night to Sunday night or something like that. And then I would scope the project so that I could ship something as much work as I could fit into like a two day period. So that at the end of that weekend, I could ship something. And then afterwards, if I want to add something, I have time to do it and a chance to do that. But it's already shipped, so there's already momentum and people are using it or I'm using it.

• 10:30Speaker 03

  And so there's a reason to continue building. And so only adding one new unknown per project, I think is a good trick.

• 10:36Speaker 02

  First came across you, I think because of Monocle, which is your personal search engine. And I got very excited about it because I always wanted a personal search engine until I found that it was in a language that I've never seen before.

• 10:48Speaker 03

  There's a towel tower of like little tools and technologies that I built for myself. Yeah. Oh, one of the other tricks to being really productive when you're building side projects is just to like use a consistent set of tools that you know really, really well. And so like for me, that's a go and my language and a couple of other libraries that I've written that like I know all the way down to the bottom of the stack. And then I barely have to look anything up because I've just debugged all every possible issue that could come up. And so I could get from start to finish without getting stuck in like a weird bug that I've never seen before. But yeah, it's a weird stack.

• 11:20Speaker 02

  It also means that you probably are not aiming for, let's say, open source glory or whatever, right? Like because you're not publishing in the JavaScript ecosystem.

• 11:28Speaker 03

  Right, right. I mean, I've written some libraries before, but a lot of my projects tend to be like. The way that I approach it is less about building something that other people are going to use en masse.

• 11:38Speaker 02

  And make yourself happy.

• 11:39Speaker 03

  Yeah, more about like, here's the thing that I built, if you wanna, and often I learned something in the process of building that thing. So like with Monocle, I wrote a custom sort of full text search index. And I thought a lot of the parts of what I built was interesting. And so I just wanted other people to look at it and see how it works and understand it. But the goal isn't necessarily for you to be able to replicate it and run it on your own.

• 11:59Speaker 02

  Well, we can kind of dive into your other AI UX thoughts. You've been, as you've been diving in, you tend to share a lot on Twitter. And I just kind of took out some of your greatest hits.

• 12:08Speaker 03

  Yeah.

• 12:09Speaker 02

  This is relevant to the demo that you picked out, Alessio, and what we're talking about, which is most knowledge work is not a text generation task. That's funny, because a lot of what Notion AI is, is text generation right now. Maybe you want to elaborate a little bit.

• 12:22Speaker 03

  Yeah. Yeah, I think the first time you look at something like GPT, the shape of the thing you see is like, oh, it's a thing that it takes some input text and generates some output text. And so the easiest thing to build on top of that is like a content generation tool. But I think there's a couple of other categories of things that you could build that are sort of progressively more useful and more interesting. And so besides content generation, which requires a minimum amount of wrapping around ChatGPT, the sort of second tier up from that is things around knowledge, I think. So if you have, I mean, this is the hot thing with all these vector DB things going around. But if you have a lot of existing context around some knowledge about your company or about a field or, you know, all of the internet,

• 13:05Speaker 03

  you can use a language model as a way to search and understand things in it and combine and synthesize them. And that synthesis, I think, is useful. And at that point, that I think the value that that unlocks, I think, is much greater than the value of content generation. Because most knowledge work, the artifact that you produce isn't actually about like writing more words. Most knowledge work, the goal is to understand something, synthesize new things, or like propose actions or other kinds of knowledge to knowledge task. And then the third category, I think, is automation. Which I think is sort of the thing that people are looking at most actively today, at least from my vantage point in the ecosystem, things like the React prompting technique, and just in general, letting models propose actions or write code to accomplish tasks. That's also moving far beyond generating text to doing something more interesting.

• 13:50Speaker 03

  But yeah, so much of what you, the value of what humans sit down and do at work isn't actually in the words that they write. It's like all the thinking that goes on before you write those words. And so, how can you get language models to contribute to those parts of work?

• 14:06Speaker 01

  I think when you first tweeted about this, I don't know if you already accepted the job, but you tweeted about this and then the next one was like, this is a Notion AI subtweet. Yeah. I didn't realize that. Yeah, I think like, the best thing that I see is like when people complain and then they're like, okay, I'm gonna go and like, help make the thing better. Right. So what are like, some of the things that you've been thinking about? I know you talked a lot about some of the flexibility versus like intuitiveness of the product. Like, the language is really flexible, right? Because you can say anything. And it's funny, the models never ignore you, they always respond with something. So no matter what you write, something is gonna come back.

• 14:45Speaker 03

  Yeah.

• 14:46Speaker 01

  But sometimes you don't know how big the space of action is, how many things you can do. So as a product built there, like, how do you think about the trade-offs that you're willing to take for like, your users? Where like, okay, I'm not gonna let you be as flexible, but I'm gonna create this guardrails for you. Like, what's the process to like, think about the guardrails and like, how you wanna funnel them to the right action?

• 15:09Speaker 03

  Yeah, I think what this trade-off you mentioned around flexibility versus intuitiveness, I think gets at one of the core design challenges for building products on top of language models. A lot of good interface design comes from tastefully adding the right constraints in place to guide the user towards actions that you wanna take, or just like, as you add more guardrails, the obvious actions become more obvious. And one common way to make an interface more intuitive is to narrow the space of choices that the users have to make and the number of choices that they have to make. And that intuitiveness, that source of intuitiveness from adding constraints, is kind of directly at odds with the reason that language models are so powerful and interesting, which is that they're so flexible and so general. And you can ask them to do literally anything and they will always give you something.

• 15:54Speaker 03

  But most of the time, the answer isn't that high quality. And so there's kind of a distribution of like, there are clumps of things in the action space of what a language model can do that the model's good at. And there's like parts of the space where it's bad at. And so one sort of high level framework that I have for thinking about designing with language models is there are actions that the language model's good at and actions that it's bad at. How do you add the right constraints carefully to guide the user and the system towards the things that the language model's good at? And then at the same time, how do you use those constraints to set the user expectations for what it's going to be good at and bad at? There, one way to do this is just literally to add those constraints and to set expectations. So a common example I use all the time is if you have some AI system to answer questions from a knowledge base.

• 16:39Speaker 03

  There are a couple of different ways to surface that in a kind of a hypothetical product. One is you could have a thing that looks like a chat window in a messaging app. And then you could tell the user, hey, this is for looking things up from a database. You can ask a question and it'll look things up and give you an answer. But if something looks like a chat, and this is a lesson that's been learned over and over for anyone building chat interfaces since like 2014, 15. If you have anything that looks like a chat interface or a messaging app, people are going to put some like weird stuff in there that just don't look like the thing that you want the model to take in because the expectation is, hey, I can use this like a messaging app and people will send in like, hi, hello, you know, weird questions, weird comments. Whereas if you take that same, literally the same input box and put it in like a thing that looks like a search bar with like a search button, people are going to treat it more like a search window.

• 17:22Speaker 03

  And at that point, inputs look a lot more like keywords or a list of keywords or maybe questions. But that simple act of like contextualizing that input in different parts of an interface, set the, reset the user's expectations, which constrain the space of things that the model has to handle. And that you're kind of adding constraints because you're really restricting your input to mostly things that look like keyword search. But because of that constraint, you can have the model fit the expectations better. You can tune the model to perform better in those settings. And it's also less confusing and perhaps more intuitive because the user isn't stuck with this blank page syndrome problem of, okay, here's an input, what do I actually do with it? When we initially launched Notion AI, one of my common takeaways personally from talking to a lot of my friends who had tried it.

• 18:07Speaker 03

  Obviously, there were a lot of people who were getting lots of value out of using it to automate writing emails or writing marketing copy. There were a ton of people who are using it to like write Instagram ads and then sort of paste it into the Instagram tool. But some of my friends who had tried it and did not use it as much, a frequently cited reason was, I tried it, it was cool. It was cool for the things that sort of Notion AI was marketed for. But for my particular use case, I had a hard time figuring out exactly the way it was useful in my workflow. And I think that gets back at the problem of it's such a general tool that just presented with a blank prompt box. It's hard to know exactly the way it could be useful to your particular use case.

• 18:44Speaker 01

  What do you think is the relationship between novelty and like flexibility? I feel like we're in kind of like a prompting honeymoon phase where like the tools are new and then like everybody just wants to do whatever they want to do, you know? And like, so it's good to give these interfaces because people can explore. But if I go forward in like three years, ideally, I'm not prompting anything, you know? Like the UX has been built for most products to like already have the intuitive, kind of like a happy path built into it. Like, do you think there's merit in a way? If you think about ChatGPT, like if it was limited, like the reason why it got so viral is like people were doing things that didn't think a computer could do, you know? It's like write poems and like, you know, solve riddles and like all these different things.

• 19:28Speaker 01

  How do you think about that, especially Notion where like it's a new, Notion AI is kind of like a new product in an existing thing? How much of it for you is like letting that happen, you know? And like seeing how people use it and then at some point be like, okay, we know what people want to do, like the flexibility is not, it was cool before, but now we just want you to do the right things with the right UX?

• 19:50Speaker 03

  I think there is value in always having the most general input as an escape hatch for people who want to take advantage of that power. At this point, Notion AI has a couple of different manifestations in the product. There's the writer, there's a thing we called an AI block, which is a thing that you can always sort of re-update as a part of document. It's like a live, a little portal inside the document that an AI can write. We also have a relatively new thing called AI autofill, which lets an AI fill an entire column in an Notion database. In all of these things, speaking of adding constraints, we have a lot of suggested prompts that we've worked on and we've curated and we think work pretty well for things like summarization and writing drafts to blog posts and things. But we always leave a fully custom prompt for a few reasons. One is if you are actually a power user and you know how language models work,

• 20:35Speaker 03

  you can go in and write your custom prompt. And if you're a power user, you want access to the power. Another is for us to be able to discover new use cases. And so one of the lovely things about working on product like Notion is that there's such an enthusiastic and lively kind of community of ambassadors and people that are excited about trying different things and coming up with all these templates and new use cases. And having a fully custom action or prompt, whenever we launch something new in AI, lets those people really experiment and help us discover new ways to take advantage of AI. I think it's good in that way. There's also a sort of complement to that, which is if we wanted to use feedback data or learn from those things and help improve the way that we are prompting the model or the models that we're building,

• 21:20Speaker 03

  having access to that like fully diverse, fully general range of use cases helps us make sure that our models can handle the full generality of what people want to do.

• 21:29Speaker 02

  I feel like we've segued a lot into already the Notion conversation. And maybe I just wanted to bridge that a little bit with your personal journey into Notion before we go into Notion proper. You spent a year kind of on a sabbatical, kind of on your own self-guided research journey and then deciding to join Notion. I think a lot of engineers out there thinking about doing this, maybe don't have the internal compass that you have or don't have the guts to basically make no money for a year. Maybe just maybe share with people like how you decided to basically go on your own independent journey and what got you to join Notion in the end.

• 22:04Speaker 03

  Yeah, what happened?

• 22:07Speaker 03

  Yeah, so for a little bit of context for people who don't know me, I was working mostly at sort of seed stage startups as a web engineer. I actually didn't really do much AI at all for prior to my year off. And then I took all of 2022 off with less of a focus on it. Ended up sort of in retrospect, becoming like a lot of pivots to AI year, which was like beautifully well timed. But in the beginning of the year, there was kind of a one key motivation and then one key kind of question that I had. The motivation was that I think I was at a Sort of a privileged and fortunate enough place where I felt like I had some money saved up that I had saved up explicitly to be able to take some time off and investigate my own kind of questions

• 22:52Speaker 03

  because I was already working on lots of side projects and I wanted to spend more time on it. I think I also at that point felt like I had enough security in the like companies and folks that I knew that if I really needed a job on a short notice, I could go and I could find some work to do. So I wouldn't be completely on the streets. And so that security, I think, gave me the confidence to say, okay, let's try this kind of experiment. Maybe it'll only be for six months, maybe it'll be for a year. I had enough money saved up to last like a year and change. And so I had planned for a year off. And I had one sort of big question that I wanted to explore. Having that single question, I think, actually was really helpful for focusing the effort. Instead of just being like, I'm going to side project for a year, which I think would have been less productive.

• 23:37Speaker 03

  And that big question was, how do we evolve text interfaces forward? So so much of knowledge work is consuming walls of text and then producing more walls of text. And text is so ubiquitous, not just in software, but just in general in the world. There were like signage and menus and, you know, books and it's ubiquitous, but it's not very ergonomic. There's a lot of things about text interfaces that could be better. And so I wanted to explore how we could make that better. A key part of that ended up being, as I discovered, taking advantage of this new new technologies that let computers make sense of text information. And so that's how I ended up sort of sliding into AI. But the motivation in the beginning was less focused on like learning a new technology and more just on exploring this general question space.

• 24:17Speaker 02

  Yeah, you had the quote, text is the lowest denominator, not the end game.

• 24:20Speaker 03

  Right, right. I mean, I think if you look at any specific domain or discipline, whether it's like medicine or mathematics or software engineering, in any specific discipline where there is a narrower set of abstractions for people to work with, there are like custom notations. One of the first things that I wrote in this like exploration year was this piece called Notational Intelligence, where I talk about this idea that so much of, as a total sidebar, there's a whole other like fascinating conversation that I would love to have at some point, maybe maybe today, maybe later, about. How to evolve sort of like a budding scene of research into like a fully fledged field. So like, I think AI UX is kind of in this weird stage where like, there's a group of interesting people that are interested in exploring this space of how do you design

• 25:04Speaker 03

  for this new fangled technology? And how do you take that and go and build sort of like best practices and powerful methods and tools?

• 25:11Speaker 02

  We should talk about this.

• 25:12Speaker 03

  We should talk about that at some point.

• 25:12Speaker 02

  Okay.

• 25:13Speaker 03

  But in a lot of established fields, there are notations that people use that really help them work at a slightly higher level than just raw words. So like, notation for describing chemicals and notations for different areas of mathematics that let people work with higher level concepts more easily.

• 25:28Speaker 02

  Logic, linguistics.

• 25:30Speaker 03

  Yeah. And like, I think it's fair to say that some large part of human intelligence, especially in these more technical domains, comes from our ability to work with notations instead of work with just the raw ideas in our heads. And so, and text is a kind of notation. It's the most general kind of notation, but it's also because of its generality, not super high leverage. If you want to go into these specific domains, and so I wanted to try to improve on that frontier.

• 25:52Speaker 02

  Yeah. You said in our show notes, "One of my goals over the next few years is to ensure that we end up with interface metaphors and technical conventions that set us up for the best possible timeline for creativity and inventions ahead." So, part of that is constraints, but I feel like that is one part of the equation, right? What's the other part that is more engenders creativity?

• 26:10Speaker 03

  Tell me, tell me a little bit about that. What were you thinking there?

• 26:14Speaker 02

  Because I feel like, you know, we talked a little bit about how you do want to constrain, for example, the user interface to guide people towards things that language models are good at. But, and creative solutions do arise out of constraints. But I feel like that alone is not sufficient for people to invent things.

• 26:33Speaker 03

  I mean, there's a lot of directions I think they could go from that. The origin of that, that thing that you're quoting is when I decided to come help work on AI at Notion, a bunch of my friends were actually quite surprised, I think because they had expected that I would have gone and worked.

• 26:51Speaker 02

  It did switch. I was eyeing that for you. I mean.

• 26:55Speaker 03

  Worked on a, at a lab or at my own company or something like that. But, one of the core motivations for me joining an existing company and one that has lots of users already is this exact thing where like, in the aftermath of a new foundational technology emerging, there's Kind of a period of a few years where the winners in the market get to decide what the default interface paradigm for the technology is. So like mini computers, personal computers, the winners of that market got to decide Windows R and like how scrolling works and like what a mouse cursor is and how text is edited. Similar with mobile, the concept of like a home screen and apps and things like that, the winners of the market got to decide.

• 27:37Speaker 03

  And that has profound, like, I think it's difficult to understate the importance of in those few critical years, the winning companies in the market choosing the right abstractions and the right metaphors. And AI to me seemed like it's at that pivotal moment where it's a technology that lots of companies are adopting. There is this well-recognized need for interface best practices. And Notion seemed like a company that had this interesting balance of it could still move quickly enough and ship and prototype quickly enough to try interesting interface ideas, but it also had enough presence in the ecosystem that if we came up with the

• 28:18Speaker 03

  right solution or one that we felt was right, we could push it out and learn from real users and iterate and hopefully be a part of that story of like setting the defaults and setting what the dominant patterns are.

• 28:29Speaker 02

  Yeah, it's a special opportunity. One of my favorite stories or facts is it was like a team of 10 people that designed the original iPhone. And so all the UX that was created there is essentially what we do use as smartphones today. Including predictive text, because people were finding that people were kind of missing the right letters. So they just enhanced the hit area for certain letters based on what you're typing.

• 28:51Speaker 03

  I mean, even even just the idea of like, we should use QWERTY keyboards on tiny smartphone screens. Like, that's a weird idea, right?

• 28:59Speaker 02

  Yeah. QWERTY is another one. So I have RSI. So this actually affects me. QWERTY was specifically chosen to maximize travel distance, right? Like, it's actually not ergonomic by design because you wanted the keyboard, the key typewriters to not stick. But we don't have that anymore. We're still sticking to QWERTY. I'm still sticking to QWERTY. I could, I could switch to the other ones. I forget.

• 29:20Speaker 03

  Oh, Comac, yeah.

• 29:21Speaker 02

  Anytime, but I don't, just because of inertia.

• 29:24Speaker 03

  I have another thing like this. So going even farther back, people don't really think enough about where this concept of buttons come from. Right, so the concept of a push button as a thing where you press it and it activates some binary switch. I mean, buttons have existed for, like, mechanical buttons have existed for a long time, but really, like, this modern concept of a button that activates a binary switch really gets, like, popularized by the popular advent of electricity. Before the electricity, if you had a button that did something, you would have to construct a mechanical system where if you pressed down on a thing, it affects some other lever system that affects, as is like the final action. And this modern idea of a button that is just a binary switch got popularized in electricity. And at that point, a button has to work in that way that it does in like an alarm clock,

• 30:08Speaker 03

  because when you press down on it, there's like a spring that makes sure that the button comes back up and that it completes the circuit. And so that's the way the button works. And then when we started writing graphical interfaces, we just took that idea of a thing that could be depressed to activate a switch. All the modern buttons that we have today in software interfaces are like simulating electronic push buttons where you like press down to complete a circuit, except there's actually no circuit being completed. It's just like a square on a screen. It's all virtualized. Right, and then you control the simulation of a button by clicking a physical button on a mouse. Except if you're on a trackpad, it's not even a physical button anymore. It's like a simulated button hardware that controls a simulated button in software. And it's also just this cascade of like conceptual backwards compatibility that gets us here.

• 30:53Speaker 03

  I think buttons are interesting.

• 30:55Speaker 01

  Where are you on the skeuomorphic design love-hate spectrum? There's people that have like high nostalgia for like the original, you know, the YouTube icon on the iPhone with like the knobs on the TV.

• 31:05Speaker 03

  I think a big part of that is, at least the aesthetic part of it, is fashion. Like, fashion taken very literally, like, in the same way that like the early, like, Y2K 90s aesthetic comes and goes. I think skeuomorphism, as expressed in like the early iPhone or like Windows XP, comes and goes. There's another aspect to this, which is the part of skeuomorphism that helps people understand Intuit software, which has less to do with

• 31:32Speaker 03

  Schemorphism, making things easier to understand per se and more about like, like a slightly more general version of schemorphism is like, there should be a consistent mental model behind an interface that is easy to grok. And then, once the user has the mental model, even if it's not the full model of exactly how that system works, there should be a simplified model that the user can easily understand and then sort of like adopt and use. One of my favorite examples of this is how volume controls that are designed well often work. Like on an iPhone, when you make your iPhone volume twice as loud, the sound that comes out isn't actually like at a physical level twice as loud, it's on a log scale. When you push the volume slider up on an iPhone, the speaker uses like four times more energy. But humans perceive it as twice as loud.

• 32:17Speaker 03

  And so the mental model that we're working with is, okay, if I make this, this volume control slider have two times more value, it's gonna sound two times louder, even though actually the underlying physics is like on a log scale. But what actually happens physically is not actually what matters. What matters is how humans perceive it and the model that I have in my head. And there, I think there are a lot of other instances where the schemorphism isn't actually the thing. The thing is just that there should be a consistent mental model. And often, the easy consistent mental model to reach for is the models that already exist in reality, but not always.

• 32:46Speaker 01

  I think the other big topic, maybe before we dive into the notion, is agents. I think that's one of the toughest interfaces to crack, mostly because, you know, the tax box, everybody understands that. The agent is kind of like, it's like human-like feeling, you know? Where it's like, okay, I'm kind of delegating something to a human, right? I think like trying to give you maybe the example of like a calendar, like a Savvy Cal is like an agent, because like it's scheduling on your behalf for something.

• 33:14Speaker 03

  That's actually a really interesting example, because it's a kind of a, it's a pretty deterministic, like there's no real AI to it. Yeah, but it is, it is agent in the sense that you're like delegating it and automate something.

• 33:24Speaker 02

  Yeah, it does work without me. It's great.

• 33:26Speaker 01

  So that one, we figured it out. Like, we know what the scheduling interface is like.

• 33:30Speaker 02

  Well, that's the state of the art now.

• 33:33Speaker 01

  Right.

• 33:33Speaker 02

  But, you know, for example, the person I'm corresponding with still has to pick a time from my calendar, which some people dislike. Sam Lessin famously says it's a sign of disrespect. I disagree with him, but, you know, it's a point of view. There could be some intermediate AI agents that would send emails back and forth like a human person to give the other person who feels slighted that sense of respect or personalized touch that they want. So there's always ways to push it.

• 34:02Speaker 01

  Yeah, I think for me, you know, other stuff that I think about, so we were doing prep for another episode, then I had an agent and asked it to do like a, you know, background prep on like the background of the person. And it just couldn't quite get the format that I wanted it to be, you know? But I kept to have, the only way to prompt it is like, give it text, give it text example, give it text example. What do you think like the interface between human and agents in the future will be? Like, do you still think agents are like this open-ended thing that are like objective-driven? Where you say, hey, this is what I want to achieve versus I only trust this agent to do X. And like, this is how X is done. I'm curious, because that kind of seems like a lot of mental overhead, you know,

• 34:46Speaker 01

  to remember each agent for each task versus like, if you have an executive assistant, like they'll do a random set of tasks and you can trust them because they're a human. But I feel like with agents, we're not, not quite there.

• 34:58Speaker 03

  Agents are hard. The, the design space is just so vast. I've, since all of the like early agent stuff came out around AutoGPT, I've tried to develop some kind of a thesis around it. And I think it's, it's just difficult because there's so many variables. One framework that I usually apply to sort of like existing chat-based prompting kind of things that I think also applies just as well to agents is this duality between what you might call like trust and control. So you, just now you brought up this example of you had an agent try to write some, write up some prep document for an episode and it couldn't quite get the format right.

• 35:38Speaker 03

  And one way you could describe that is you could say, oh, the, the agent didn't exactly do what I meant and what I had in my head, so I can't trust it to do the right job. But a different way to describe it is I have a hard time. Controlling exactly the output of the model. And I have a hard time communicating exactly what's in my head to the model. And they're kind of two sides of the same coin. I think if you if you can somehow provide a way to with less effort, communicate and control and constrain the model output a little bit more and constrain the behavior a little bit more. I think that would alleviate the pressure for the model to be this like fully trusted thing, because there's no need for trust anymore. There's just kind of guardrails that ensure that the model does the right thing. So developing ways and interfaces for these agents to be a little more constrained in

• 36:22Speaker 03

  its output, or maybe for the human to control its output a little bit more or behavior a little bit more, I think is a productive path. Another sort of more recent revelation that I had while working on this AI autofill thing inside Notion is the importance of zones of influence for AI agents, especially in collaborative settings. So having worked on lots of interfaces for independent work on my year off, one of the surprising lessons that I learned early on when I joined Notion was that if you build, a collaboration permeates everything, which is great for Notion, because collaborating with an AI, you reuse a lot of the same metaphors for collaborating with humans. So one nice thing about this autofill thing that also kind of applies to AI blocks, which

• 37:04Speaker 03

  is another thing that we have, is that you don't alleviate this problem of having to ask questions like, oh, is this document written by an AI or is this written by a human, like this need for auditability, because the part that's written by the AI is just in like the autofilled cell or in the AI block. And you can you can tell that's written by the AI and things outside of it, you can kind of reasonably assume that it was written by you. I think anytime you have sort of an unbounded action space for for models like agents, it's especially important to be able to answer those questions easily and to have some sense of security that in the same way that you want to know whether your coworker or collaborator has access to a document or has modified a document, you want to know whether an AI has permissions to access something. And if it's modified something or made some edit, you want to know that it did it.

• 37:50Speaker 03

  And so as a complement to constraining the model's action space proactively, I think it's also important to communicate and have the user have an easy understanding of like what exactly did the model do here? And I think that helps build trust as well.

• 38:02Speaker 02

  Yeah, I think for AutoGPT and those kinds of agents in particular, anything that is destructive, you need to prompt for, I guess, or like check with, check in with the user.

• 38:12Speaker 03

  Prompt.

• 38:13Speaker 02

  I know, I can't say, I know, it's overloaded now, I can't say that.

• 38:16Speaker 03

  You have to confirm with the user.

• 38:17Speaker 02

  You confirm to the user, yeah, exactly.

• 38:18Speaker 03

  Yeah.

• 38:19Speaker 02

  Yeah.

• 38:19Speaker 03

  That's tough too, though, because you don't wanna stop. Yeah, one of the, one of the benefits of automating these things, that you can sort of like, in theory, you can scale them out arbitrarily. I can have like a hundred different agents working for me. But if that means I'm just like spending my entire day in a daily use of notifications, that's not ideal either.

• 38:35Speaker 02

  Yeah, so then, then it could be like a reversible destructive thing with some kind of timeouts, a time limit. So you could reverse it within some window. I don't know. I, yeah, I've been thinking about this a little bit because I've been working on a small developer agent.

• 38:49Speaker 03

  Right, right. Or maybe you could like batch a group of changes and can sort of like summarize them with another AI and approve them in bulk or something.

• 38:56Speaker 02

  Which is surprisingly similar to the collaboration problem.

• 38:58Speaker 03

  Yeah, yeah, yeah, exactly. Yeah.

• 39:02Speaker 03

  I'm telling you, the collaboration, a lot of the problems with collaborating with humans also apply to collaborating with AI. There's a potential pitfall to that as well, which is that there are a lot of things that, some of the core advantages of AI end up missing out on if you just fully anthropomorphize them into like human-like collaborators. But, but.

• 39:19Speaker 02

  Do you have a strong opinion on that? Like, do you refer to it as it?

• 39:22Speaker 03

  Oh, yeah. I'm a, I'm an it person, at least for, for never, in, in 2023.

• 39:27Speaker 02

  Yeah. So that, that leads us nicely into introducing what Notion and Notion AI is today. Do you have a pet answer as to what is Notion? I've heard it introduced as a database, a WordPress killer, a knowledge base, a collaboration tool. What is it?

• 39:41Speaker 03

  Yeah, I mean, the official answer is that a Notion is a connected workspace. It has a space for your company docs, meeting notes, a wiki for all of your company notes. You can also use it to orchestrate your workflows if you're managing a project, if you have an engineering team, if you have a sales team. You can put all of those in a single Notion database. And the benefit of Notion is that all of them live in a single space where you can link to your wiki pages from your, I don't know, like onboarding docs, or you can link to a GitHub issue through a task from your documentation on your engineering system. And all of this existing in a single place and it's kind of like. Unified, yeah, like, single workspace, I think, has lots of benefits. That's the official line. There's an asterisk that I usually enjoy diving deeper into,

• 40:26Speaker 03

  which is that the whole reason that this connected workspace is possible is because, underlying all of this, is this really cool abstraction of blocks. In Notion, everything is a block. A paragraph is a block, a bullet point is a block, but also a page is a block. And the way that Notion databases work is that a database is just a collection of pages, which are really blocks. And you can, like, take a paragraph and drag it into a database, and it'll become a page. And you can take a page inside a database and pull it out, and it'll just become a link to that page. And so, this core abstraction of a block that can also be a page, that can also be a row in a database, like an Excel sheet, that fluidity and this, like, shared abstraction across all these different areas inside Notion, I think, is what really makes Notion powerful. This Lego theme, this Lego building block theme permeates

• 41:11Speaker 03

  a lot of different parts of Notion. Some fans of Notion might know that when you join Notion, you get a little Lego minifigure. It has Lego building blocks for workflows. And then, every year you're at Notion, you get a new block that says, like, you've been here for a year, you've been here for two years. And then, Simon, our co-founder and CTO, has a whole crate of Lego blocks on his desk. And he just likes to mess with, because, you know, he's been around for a long time. But this Lego building block thing, this, like, shared, sort of, all-encompassing, single abstraction that you can combine to build various different kinds of workflows, I think it's really what makes Notion powerful. And one of the sort of background questions that I have for Notion AI is, like, what is that kind of building block for AI?

• 41:53Speaker 02

  Well, we can dive into that. So what is Notion AI? So I kind of view it as like a startup within the startup. Could you describe the Notion AI team? Is this, like, how seriously is Notion taking the AI wave?

• 42:06Speaker 03

  The most seriously. The way that Notion AI came about, as I understand it, because I joined a bit later... I think it was around October last year, all of Notion team had a little offsite. And as a part of that, Ivan and Simon kind of went into a little kind of hack weekend. And the thing that they ended up hacking on inside Notion was the very, very early prototype of Notion AI. They saw this GPT-3 thing. The early, early motivation for starting Notion, building Notion in the first place for them, was sort of grounded in this like utopian end user programming vision, where software is so powerful, but there are only so many people in the world that can write programs. But everyone can benefit from having a little workspace or a little program

• 42:51Speaker 03

  or a little like workflow tool that's programmed to just fit their use case. And so how can we build a tool that lets people customize their software tools that they use every day for their use case? And I think to them seemed like such a critical part of facilitating that, like bridging the gap between people who can code and people who need software. And so they saw that, they tried to build an initial prototype that ended up becoming the first version of Notion AI. They had a prototype in, I think, like late October, early November, before ChatGPT came out and sort of evolved it over the few months. But what ended up launching was sort of in line with the initial vision, I think, of what they ended up building. And then, once they had it, I think they, they wanted to keep pushing it. And so at this point, AI is a really key part of Notion's strategy and what we see Notion becoming going forward.

• 43:36Speaker 03

  In the same way that like blocks and databases are a core part of Notion that helps enable workflow automation and all these important parts of running a team or collaborating with people or running your life. We think that AI is going to become an equally critical part of what Notion is. And it won't be Notion is a cool connected workspace app and it also has AI. It'll be that like what Notion is, is databases, it has pages, it has space for docs, and it also has this sort of comprehensive suite of AI tools that permeate everything. And one of the challenges of the AI team, which is, as you said, kind of a startup within a startup right now, is to figure out exactly what that. All permeating kind of abstraction means, which is a fascinating and difficult open problem.

• 44:20Speaker 01

  How do you think about what people expect of Notion versus what you want to build in Notion? A lot of like this AI technology kind of changes. You know, we talked about the relationship between text and human and like how human collaborates. Do you put any constraints on yourself? And it's like, OK, people expect Notion to work this way with these blocks. So maybe I have this crazy idea and like I cannot really pursue it because it's there. I think it's a classic like innovator's dilemma kind of thing. And I think a lot of founders out there that are in a similar position where it's like, you know, series C, series D companies. Like you're not quite yet the super established one. You're still moving forward, but you have an existing kind of following and something that Notion stands for.

• 45:04Speaker 01

  How do you kind of wrangle with that?

• 45:06Speaker 03

  Yeah, that is in some ways a challenge and that Notion already is a kind of a thing. And so we can't just scrap everything and start over. But I think it's also, there's a blessing side of it too in that because there are so many people using Notion in so many different ways, we understand all of the things that people want to use Notion for very well. And then so we already have a really well defined space of problems that we want to help people solve. And that helps us. We have it with the existing Notion product and we also have it by sort of rolling out these AI things early and then watching, learning from the community, what people want to do with them. And so based on those learnings, I think we, it actually sort of helps us constrain the space of things we think we need to build. Because otherwise, the design space is just so large. But whatever we can do with AI and knowledge work.

• 45:50Speaker 03

  And so watching what people have been using Notion for and what they want to use Notion for, I think helps us constrain that space a little bit and make the problem of building AI things inside Notion a little more tractable.

• 46:00Speaker 01

  I think also just observing what they naturally use things for. And it sounds like you do a bunch of user interviews where you hear people running into issues and or describe them as. The way that I describe myself actually is I feel like the problem is with me that I'm not creative enough to come up with use cases to use Notion AI or any other AI.

• 46:20Speaker 03

  Which isn't necessarily on you, right?

• 46:22Speaker 01

  Exactly.

• 46:22Speaker 03

  Again, like goes way back to the early, the thing we touched on early in the conversation around like, if you have too much generality, there's not enough, there are not enough guardrails to obviously point to use cases.

• 46:32Speaker 01

  Blank piece of paper. I don't know what to do with this.

• 46:36Speaker 02

  So I think a lot of people judge Notion AI based on what they originally saw, which is write me a blog post or do a summary or do action items. Which fun fact, for latent space, my very, very first Hacker News hit was reverse engineering Notion AI. I actually don't know if I got it exactly right. I think I got the easy ones right, and then apparently I got the the action items one really wrong. So there's some art into into doing that. But also, you've you've since launched a bunch of other products. And maybe you've already hinted at AI AutoFill. Maybe we can just talk a little bit about what does the scope or suite of Notion's AI products have been so far? And what you're launching this week?

• 47:15Speaker 03

  Yeah, so we have, I think, three main facets of Notion AI. And Notion at the moment, we have sort of the first thing that ever launched with Notion AI, which I think that helps you write. It's going back to earlier in the conversation. It's kind of a writing, kind of a content generation tool. If you have a document and you want to generate a summary, it helps you generate a summary. Pull out action items, you can draft a blog post, you can help it improve your it's help to improve your writings. It can help fix grammar and spelling mistakes. But under the hood, it's a fairly lightweight, a thick layer of prompts. But otherwise, it's a pretty straightforward use case of language models, right? And so there's that, a tool that helps you write documents. There is a thing called an AI block, which is a slightly more constrained version of that,

• 47:59Speaker 03

  where one common way that we use it inside Notion is we take all of our meeting notes inside Notion. And frequently, when you have a meeting and you want other people to be able to go back to it and reference it, it's nice to have a summary of that meeting. So all of our meeting notes templates, at least on the AI team, have an AI block at the top that automatically summarizes the contents of that page. And so whenever we're done with a meeting, we just press a button and it'll re summarize that, including things like, you know, what are the core action items for every every person on the in the meeting? And so that block, as I said before, is nice because it's like a constrained space for the AI to work in. And we don't have to like prompt it every single time. And then the newest member of this sort of AI collection of features is AI AutoFill, which brings Notion AI to databases. So if you have a whole database of like user interviews and you want to pull out

• 48:42Speaker 03

  what are the companies, where their like core pain points, what are their core features, maybe what other competitor products they use, you can just make columns. And in the same way that you write Excel formulas, you can write a little AI formula, basically, that where the AI will look at the content of the page and pull out each of these key pieces of information. The new, slightly new thing that AutoFill introduces is this idea of a more automated kind of background AI thing. So with Writer, the sort of like AI in your document product and the AI block, you have to always ask it to update, you have to always ask it to rewrite. But if you have a column in a database, in a Notion database, or a property in a Notion database, it would be nice if you, whenever someone went back and like changed the contents

• 49:26Speaker 03

  of the meeting note, or something updated about the page, or maybe it's like a list of tasks that you have to do and the status of the task changes, you might want the summary of that task or detail of the task to update. And so anytime that are, you can set up an auto-filled Notion property so that anytime something on that database row or page changes, the AI will go back and sort of auto-update the like auto-filled value. And that I think is a really interesting part that we might continue leading into of like, even though there is AI now tied to this particular page, it's sort of doing its own thing in the background to help automate and alleviate some of that pain of automating these things. But yeah, Writer blocks and AutoFill are the three sort of cornerstones we have today.

• 50:05Speaker 01

  You know, there used to be this glorious time where like, uh, Rome Research was like the hottest knowledge company out there and then Notion built backlinks and.

• 50:14Speaker 03

  How?

• 50:15Speaker 01

  Yeah.

• 50:17Speaker 03

  I don't know if we are to blame for that.

• 50:19Speaker 01

  No, no, but how, how do backlinks play into some of this? You know, I think most AI use cases today are kind of like a single page, right? Kind of like this document, um, I'm helping with this. Do you see some of these tools expanding to do changes across things? So we just had Edamar from Codium on the podcast and he talked about how agents can tie in. And specs for features, test for features and the code for the feature. So like the three entities are tied together. Like, do you see some backlinks help AI navigate through knowledge basis of companies? Where like you might have the document the product uses, but you also have the document that like marketing uses to then announce it and like as you make changes, like the AI can work through different pieces of it?

• 51:04Speaker 03

  Definitely. If I may get a little theoretical from that.

• 51:07Speaker 01

  Yeah, yeah.

• 51:07Speaker 03

  One of my favorite ideas from my last year of hacking around building text augmentations with AI for documents is this realization that, you know, when you look at code in a code editor, what it is at a very lowest level is it's just text files. A code file is a text file and there are like maybe functions inside of it and it's a list of functions, but it's a text file. But the way that you understand it is not as a file, like a Word document. It's a kind of a graph, like you have a function, you have call sites to that function. There are places where you call that function. There's a place where that function is tested. Many different definitions for that function. Maybe there's a type definition that's tied to that function. So it's a kind of a graph. And if you want to understand that function, there is advantages to be able to traverse that whole graph and fully contextualize where that function is used.

• 51:52Speaker 03

  Same with types and same with variables. And so even though its code is represented as text files, it's actually a kind of a graph. And a lot of the of what all that like key interfaces, interface innovations behind IDEs is helping surface that graph structure in the context of a text file. So like things like go to definition or VS Code's little window view when you like look at references. And interesting idea that I explored last year was what if you bring that to text documents? So text documents are a little more unstructured, so there's a less, there's a more fuzzy kind of graph idea. But if you're reading a textbook, if there's a new term, there's actually other places where the term is mentioned. There are probably a few places where that's defined. Maybe there are some figures that reference that term. If you have an idea, there are other parts of the document where.

• 52:36Speaker 03

  The document might disagree with that idea or cite that idea. So there's still kind of a graph structure. It's a little more fuzzy, but there's a graph structure that ties together like a body of knowledge. And it would be cool if you had some kind of a text editor or some kind of knowledge tool that let you explore that whole graph. Or maybe if an AI could explore the whole graph. And so back to your point, I think, taking advantage of not not just the backlinks, backlinks is a part of it, but the fact that all of these inside Notion, all of these pages exist in a single workspace. It's a shared context. It's a connected workspace. And you can take any idea and look up anywhere to fully contextualize what like, what a part of your like engineering system design mean? Or what we know about our pitching their customer to a company? Or like, if I wrote down a book, what are other places where that book has been mentioned?

• 53:20Speaker 03

  Like, all these graph following things, I think, are really important for contextualizing knowledge.

• 53:25Speaker 02

  Part of your job at Notion is prompt engineering. You are maybe one of the more advanced prompt engineers that I know, like, out there. And you've always you've always commented on the the state of prompt ops tooling. What is your process today? What do you wish for?

• 53:42Speaker 03

  There's a lot here. I mean, the prompts that are inside Notion right now, they're not complex in the sense that like agent prompts are complex, but they're they're complex in the sense that there is even a problem as simple as like summarize a page. A page could contain anything from no information, if it's a fresh document, to like a fully fledged news article. Maybe it's like a meeting note. Maybe it's like a bug filed by somebody at a company. Like, the the range of possible documents is is huge. And then you have to distill all of it down to like always generate a summary. And so describing that task to AI comprehensively is pretty hard. There are a few things that I think I ended up leaning on and as a team, we end up leaning on for the prompt engineering part of it. I think one of the early transitions that we made was that the the initial prototype

• 54:26Speaker 03

  for Notion AI was built on instruction following the sort of classic instruction following models, text W G zero zero three and so on. And then at some point, we all switched to chat based models like Claude and the new Chat GPT Turbo and these models. And so that was an interesting transition. It actually kind of made few shot prompting a little bit easier, I think, in that you could give. The few-shot examples as sort of previous turns in a conversation. And then you could ask the real question as the next follow-up turn. I've come to appreciate few-shot prompting a lot more because it's difficult to fully comprehensively explain a particular task in words. But it's pretty easy to demonstrate like four or five different edge cases that you want the model to handle. And it's a lot of times if there's an edge case that you want a model to handle.

• 55:10Speaker 03

  I think few-shot prompting is just the easiest, most reliable tool to reach for. One challenge that in prompt engineering that Notion has to contend with often is we want to support all the different languages that Notion supports. And so all of our prompts have to be multilingual compatible. Which is kind of tricky because our prompts are written, our instructions are written in English. And so if you just have a vanilla, a naive approach, then the model tends to output in English, even when the document that you want to translate or summarize is in French. And so one way you could try to attack that problem is to tell the model answer in the language of the user's query. But it's actually a lot more effective to just give it examples of not just English documents, but like maybe summarizing an English document, maybe summarize like a ticket filed in French, summarize an empty document where the document is supposed to be in Korean.

• 55:54Speaker 03

  And so a lot of our few-shot prompt included prompts in Notion AI tend to be very multilingual and that helps sort of support our non-English speaking users. The other big part of prompt engineering is evaluation. The prompts that you exfiltrated out of Notion AI many weeks ago, surprisingly pretty spot on, at least for the prompts that we had then, especially things like summary. But they're also edited because we've evolved them a lot more and we have a lot more examples. And some of our prompts are just really, really long, they're like thousands of tokens long. And so every time we go back and add an example or modify the instruction, we want to make sure that we don't regress any of the previous use cases that we've supported.

• 56:35Speaker 03

  And so we put a lot of effort and we're increasingly building out internal tooling infrastructure for things like sort of what you might call unit test and regression tests for prompts with like handwritten test cases, as well as tests that are driven more by feedback from, from Notion users that have chosen to share their feedback with us.

• 56:54Speaker 02

  You just have like a hand-rolled testing framework or use Jest or whatever. And nothing custom out there. You, you, you basically said you've looked at so many prompt ops tools and you're sold on none of them.

• 57:06Speaker 03

  So that tweet was from a while ago. I think there are a couple of interesting tools these days, but I think at the moment, Notion uses pretty hand-rolled tools. Nothing too heavy, but it's basically a for loop over a list of test cases. We do do quite a bit of using language models to evaluate language models. So our unit test descriptions are kind of funny because the test is literally just an input document and a query, and then we expect the model to say something. And then our qualification for whether that task passes or not is just ask the language model again, whether it's like, it's like, looks like a reasonable summary, or whether it's in the right language.

• 57:42Speaker 02

  Do you have the same model? Do you have like Anthropic criticized, you know, OpenAI or OpenAI criticized Anthropic?

• 57:47Speaker 03

  That's a good question.

• 57:48Speaker 02

  Do you worry about like models being biased towards its own self?

• 57:52Speaker 03

  Oh, no, that's not a worry that we have. I actually don't know exactly if we use different models. If you have a fixed budget for running these tests, I think it would make sense to use more expensive models for evaluation rather than generation. But yeah, I don't remember exactly what we do there.

• 58:07Speaker 02

  And then one more follow up on, you mentioned some of your prompts are thousands of tokens.

• 58:11Speaker 03

  Yeah.

• 58:11Speaker 02

  That takes away from my budget as a user.

• 58:14Speaker 03

  Yeah.

• 58:15Speaker 02

  Isn't that a trade off that's a concern? So there's a limited context window.

• 58:19Speaker 03

  Yes.

• 58:20Speaker 02

  Some of that is taken by you as the app designer, product designer, deciding what system prompt to provide. And then the remainder is what I as a user can give you to actually summarize as my content.

• 58:32Speaker 03

  Yeah, in theory, I think in practice, there are a couple of trends that make that an issue. I think for, so for things like generating summaries, a summary is only going to be so many tokens long. If our prompts are generating you 3,000 token summaries, like we're not, the prompt is not doing its job anyway.

• 58:48Speaker 02

  Yeah, but the source doc is.

• 58:50Speaker 03

  The source doc could be longer. So like if you wanted to translate a 5,000 token document, you do have to truncate it, and there is a limitation. It's not something that we are super focused on at the moment for a couple of reasons. I think there are techniques that if we need to help us compress those prompts, things like parameter efficient fine tuning. And also the context lengths are, it seems like the dominant trend is that context lengths are getting cheaper and longer constantly. Anthropic recently announced their 100,000 token context model recently. And so I think in the longer term, that's going to be sort of be taken care of anyway by the models becoming more accommodating of longer contexts, and it's more of a temporary limitation.

• 59:26Speaker 02

  Cool. Shall we talk about the professionalizing of a scene?

• 59:30Speaker 03

  Yeah, I think one of the things that is a helpful bit of context when thinking about HCI and AI in particular is like historically HCI and AI have been sort of competing disciplines, competing very specifically in the sense that they often fought for the same sources of funding and the same kinds of people and attention throughout the history of computer science. They used to, HCI and AI both used to come from the same or like very aligned similar parallel motivations of we have computers, how do we make computers work better with humans? And one way to do it was to make the machine smarter. Another way to do it was to design better interfaces. And through the AI booms and busts, when the AI boom was happening, HCI would get less funding. And when AIs had winters, HCI would get a lot

• 1:00:14Speaker 03

  more attention because it was sort of the alternative solution. And now that we have this sort of renewed attention on how to build better interfaces for AI, I think it's interesting that it's kind of a scene now. There are podcasts like this where I get to talk about interfaces and AI, but it's definitely not a fully fleshed field. My favorite definition of sort of what distinguishes the two apart came, comes from Andy Metusak, where he, I'm going to butcher the quote, but he said something to the effect of a field has at their disposal powerful set of like established tools and methods and standards and a shared set of like core questions they want to answer. And so if you look at like machine learning, which is obviously a really dominant established field, if you want to answer, if you want

• 1:00:57Speaker 03

  to like evaluate a model, if you want to answer, if you want to solve a particular task or build a model that solves a particular task, there are powerful methods that we have, like gradient descent and specific benchmarks for building solutions and then reevaluating how to do the solutions. Or if you have an even more expensive problem, there are surely attempts that have been made before and then attempts that people are making now for how to attack that problem and frameworks to think about these things. In AI and UX, I think, we're like very early in the evolution of that space and that community. And there's a lot of people excited, a lot of people building, but we have yet to come up with a set of like best practices and tools and methods and frameworks for thinking about these things. And those will surely arise. And as they do, I think we'll, we'll see

• 1:01:38Speaker 03

  the evolution of the field. In like prompt engineering and using language models in products at large. I think that community is a little farther along. It's still very fast moving because it's really young. But there are like established prompting techniques like Reacts and distillation of large instruction following models. And these these techniques, I think, are the beginnings of like best practices and powerful tools at the disposal of this language model using field.

• 1:02:06Speaker 02

  Yeah, and mostly it's just following Riley Goodside. It's how I learned about prompting techniques.

• 1:02:10Speaker 03

  Right, right. Yeah, pioneers.

• 1:02:14Speaker 02

  But yeah, I am actually interested in this. I've we've recently kind of rebranded the podcast or the newsletter somewhat in towards being for this term AI engineer, which I kind of view as somewhere between machine learning researcher and software engineer, some kind of in between mix. And I think creating the media, creating meetups, creating a de facto conference for it, creating job titles. And then I think that core set of questions that everyone wants to get better at, I think that is essentially how this starts.

• 1:02:46Speaker 03

  Yeah, yeah.

• 1:02:47Speaker 02

  Pretty excited of.

• 1:02:48Speaker 03

  Creating a space for the people that are interested to come together, I think is a really, really key important part of it. I'm always, whenever I come back to it, I'm always amazed by how like, if you look at the sort of golden era of like theoretical physics in the early 20th century or the golden era of early personal computing, there are maybe like two dozen people that have contributed all of the significant ideas to that field and they all kind of know each other. I always found that really fascinating. And I think the causal relationship actually goes the other way. It's not that all these people happen to know each other. It's that because there was that core set of people that always that were very close to each other and shared ideas often. And they were co-located, that that field is able to blossom. And so I think creating that space is really critical.

• 1:03:31Speaker 02

  Yeah, there's a very famous photo of the Soviet Conference in 1927, where Albert Einstein, Niels Bohr, Marie Curie, all these like top physics names are all in one.

• 1:03:41Speaker 03

  How many Nobel laureates are in that photo, right?

• 1:03:43Speaker 02

  Yeah, and when I tweeted it out once, people were like, I didn't know these all lived together and they all knew each other. And they, you know, they must have exchanged so many ideas.

• 1:03:51Speaker 03

  I mean, similar with artists and writers that help a new kind of like period blossom.

• 1:03:57Speaker 02

  Now, is it going to be San Francisco and New York, though?

• 1:04:01Speaker 03

  That's a spicy question. I don't know. We'll see.

• 1:04:05Speaker 02

  Well, we're glad to at least be a part of your world, whether it is on either coast. But it's also virtual, right? Like, we have a Discord, like, it's happening online as well, even if you're in a small town like Indiana.

• 1:04:20Speaker 02

  Cool, lightning round?

• 1:04:21Speaker 01

  Awesome. Yeah, let's do it. We only got three questions for you. One is acceleration, one exploration, then I found I'll take away. So the first one we always like to ask is like, what is something that happened in AI that Utah would take much longer than it does?

• 1:04:36Speaker 03

  Prices coming down. Prices coming down and/or being able to get a lot more bang for your buck. So things like GPT 3.5 Turbo being, I don't know, exactly the figure like 10 times, 20 times cheaper.

• 1:04:48Speaker 02

  Then GPT, then DaVinci 03.

• 1:04:50Speaker 03

  Then DaVinci 03 per token, or the super long context Claude, or MPT Storywriter, these like long context models that take, theoretically would take a lot of compute to run, but they're sort of accessible to us now. I think they're surprising because I would have thought that in, before these things came out, that cost per token and scaling context length, and these were like sort of core constraints that you would have to design your AI systems around. And it ends up being like, if you just wait a few months, like OpenAI will figure out how to make these models 10 times cheaper. Or Anthropic will figure out how to make the models like be able to take a million tokens. And that, the speed at which that's happened has been surprising and a little bit frightening because it invalidates a lot of the assumptions that I was operating with and I have to recalibrate.

• 1:05:34Speaker 02

  Yeah, there's this famous law called Wirth's Law, also known as Gates's Law, that basically says, it's like, software engineers will take up whatever hardware engineers give them. And I feel like there's a parallel law right now where language model improvements, AI UX people are going to take up all the improvements that language model people will give them. So, you know, they're trying to, while the language model people are improving the costs by a single order of magnitude, you with your Notion AI AutoFill are increasing by orders of magnitude the amount of consumption that's.

• 1:06:02Speaker 03

  Yeah, exactly. So before, before the show started, we were just talking about how when I was prototyping AI AutoFill. Just to make sure that things sort of like scaled up, okay, I ended up running autofill on a database with like six thousand pages and just summaries and the issue is like very long pages and I ended up running through something like two or three million tokens in a matter of like twenty minutes.

• 1:06:21Speaker 02

  Yeah.

• 1:06:21Speaker 03

  Which is not too expensive luckily because

• 1:06:23Speaker 02

  It's going to be fine.

• 1:06:24Speaker 03

  cheaper, but but it is like five or six dollars, which the concept of like running a test on my computer and it's spending the price of like a nice coffee is kind of weird thing still that I'm getting used to.

• 1:06:37Speaker 02

  And Notion, like Notion AI currently is ten dollars a month, something like that. So you like, there's ways to make Notion lose money.

• 1:06:43Speaker 03

  Negative, you've just got negative growth margins on that test. Not sanctioned by Notion, but I mean, obviously

• 1:06:51Speaker 02

  You'll figure it out.

• 1:06:52Speaker 03

  you should use it to, you know, improve your life and support your workflows in whatever way is it's useful.

• 1:06:56Speaker 02

  Okay, second question is about exploration. What do you think is the most interesting unsolved question in AI?

• 1:07:03Speaker 03

  Predictability, reliability. Well, in AI broadly, I think is much harder, but with language models specifically, I think how to build dependable systems is really important. If you ask Notion AI or if you ask ChatGPT or Claude, like maybe a bullet list of X, Y, Z, sometimes it'll make those bullets with like the Unicode center dot, sometimes it'll make them with a dash, sometimes it'll like add a title, sometimes it'll like bold random things and all of the things are fine, but it's a little jarring if every time the answers are a little stochastic. I think this is much more of a concern for when you're automating tasks or having the model make decisions by itself. Predictability, dependability, so much of the software that runs the world is sort of behind the scenes decision making programs that run inside enterprises and automate systems and make decisions for people and audibility, dependability is just so critical

• 1:07:48Speaker 03

  to all of them. One avenue of work that I'm really intrigued by is in these decision making systems, not having the model sort of internally as a black box make decisions, but having the model synthesize code that makes decisions. So you might ask the model for things like summarization, like natural language tasks you have to ask the model, but if you want it to, I don't know, let's say you have a document and you want to filter out all the dates. Instead of asking the model, hey, can you grab all the dates? You can ask the model to write a regular expression that captures a particular set of date formats that you really care about. And at that point, the output of the model is a program and the nice thing about a program is you can kind of check it. There's lots of nice things. One is it's much cheaper to run afterwards. Another is you can verify it and the program becomes a kind of a, what in design we call a boundary

• 1:08:31Speaker 03

  object, where it's a shared thing that exists both in the sphere of the human and the sphere of the computer and you can iterate on it to fix bugs and you can co-evolve this object that is now like a representation of this decision that you want the model to, the computer to make. But it's audible and dependable and reliable. And so I'm pretty bullish on code generation and other sort of like program synthesis and program verification techniques, but using the model to write the initial program and help the people maintain the software.

• 1:08:59Speaker 02

  Yeah, I'm so excited by that. Just in terms of reliability, I'll call out our previous guest.

• 1:09:04Speaker 01

  Roche Ball.

• 1:09:05Speaker 02

  Yeah. And she's working on Garbriels AI. There's also LMQL and then Microsoft recently put out Guidance, which is their custom language thing. Have you explored any of those?

• 1:09:13Speaker 03

  I've taken a look up on all of them. I've spoken to Shaya. I think this general space of like more, speaking of adding constraints to general systems, adding constraints, adding program verification, all of these things I think are super fascinating. I also personally like it a lot because I, before I was spending a lot of my time in AI, I spent a bunch of time looking at like programming languages and compilers and interpreters. And there is just so much amazing work that has gone into how do you build automated ways to reason about a program, like compilers and type checkers and so on. And it would be a real shame if the whole field of program synthesis and verification just became like AskGPT-4. But actually it's not, like they work together. You write the program, you synthesize the

• 1:09:56Speaker 03

  program with GPT-4 from human constraint, human descriptions. And then now we have this whole set of powerful techniques that we can use to more formally understand and prove things about programs. And I think the synergy of them, I'm excited to see.

• 1:10:09Speaker 01

  Awesome. This was great, Linus. Our last question is always, what's one message you want everyone to remember today about the space, exciting challenges?

• 1:10:19Speaker 03

  We were at the beginning, maybe this is really cliche, but

• 1:10:22Speaker 01

  It's okay.

• 1:10:23Speaker 03

  One thing that I always used to say about when I was working on text interfaces last year was that I would be really disappointed if in a thousand years humans are still using the same kind of like writing tools and writing systems that we are today. Like, it would be pretty surprising if we're still sort of like writing documents in the same way that we are today in a thousand years, right? Like, and the language and the writing system hasn't evolved at all. If humans want, plan to be around for many thousands of years into the future, writing has really only been around for like 2, 3,000 years and it's like sort of modern form. And we should, I think, care a lot more about building flexible, powerful tools than about backwards compatibility if we plan to be around for many more times the number of years that we've been around.

• 1:11:08Speaker 03

  And so I think whether we look at something as simple as language models or as expensive as like humans interacting with text documents, I think it's worth reminding yourself often that the things that we have today are sometimes that way for a reason, but often just because, unartefact of like the way that we've gotten here. And text can look very different, language models can look very different. I personally think in a couple of years, we're going to see something better than transformers. So all of these things are going to change and it's, I think it's important to have your eyes sort of looking over the horizon at like what's coming far into the future.

• 1:11:47Speaker 01

  Nice way to end it.

• 1:11:48Speaker 03

  Yeah.

• 1:11:49Speaker 01

  Well, thank you, Linus, for coming on. This was great.

• 1:11:51Speaker 03

  Thank you. This was lovely. Thanks for having me.