E9: The AI Assistant Revolution with Flo Crivello of Lindy.AI

People have long been imagining AI assistants. Flo Crivello (founder of Teamflow) is turning that dream into a reality with the ambitious project Lindy. Flo sits down with Nathan on Lindy's announcement day to talk about this unique moment in AI, Lindy's capabilities, single-use apps, and how he sees the global impact that AI-powered employees can have on the future of work. 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. Anyone who mentions "The Cognitive Revolution" when signing up for the Lindy beta will get priority access. (0:00) Preview (1:05) Sponsor: Omneky (4:50) Flo demos Lindy (9:21) Security and user experience (11:55) How this AI moment compares to the Industrial Revolution and single use apps (15:00) The dream AI assistant (19:54) Competition and advantages of startups (24:28) Cost vs Quality (31:19) How Lindy uses user data (32:37) Context and data injection (28:11) Comfort with ambiguity (43:10) Use of guardrails and trust concerns (47:26) AI safety and edge cases (49:16) Impact on labor markets (50:27) Economic resilience (52:09) Future of work (55:49) Rapid fire questions SPONSORS: SHOPIFY | OMNEKY 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 *Thank you Omneky for sponsoring The Cognitive Revolution. Omneky is an omnichannel creative generation platform that lets you launch hundreds of thousands of ad iterations that actually work, customized across all platforms, with a click of a button. Omneky combines generative AI and real-time advertising data. Mention "Cog Rev" for 10% off. Join 1000's of subscribers of our Substack: https://cognitiverevolution.substack Websites: lindy.ai *mention "The Cognitive Revolution" when you sign up for the waitlist to get priority access* thecognitiverevolution.ai Twitter: @CogRev_Podcast @labenz (Nathan) @Altimor (Flo) 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 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.

• 0:00Flo Crivello?

  And he, Napoleon, had the highest material of all. He had aluminum. And I'm drinking La Croix right now. And aluminum has become so cheap, we literally throw it away. I'm literally drinking an aluminum and then I'm going to throw this away. So this would have been unthinkable. So I think it's going to be the same thing with code where it's like an app today costs of the order of 10 to 100,000 dollars to build. We're rapidly approaching a world where an app costs of the order of one or 10 cents to build. And when that happens, you basically start having disposable apps. You basically start building an app just for this one session that you have right now. And that's basically what Linde is doing.

• 0:38Nathan Labenz?

  Hello and welcome to The Cognitive Revolution, where we interview visionary researchers, entrepreneurs and builders working on the frontier of artificial intelligence. Each week, we'll explore their revolutionary ideas and together we'll build a picture of how AI technology will transform work, life and society in the coming years. I'm Nathan Labenz, joined by my co-host, Erik Torenberg.

• 1:01Speaker 02

  Omnikie uses generative AI to enable you to launch hundreds of thousands of ad iterations that actually work, customized across all platforms with a click of a button. I believe in Omnikie so much that I invested in it and I recommend you use it too. Use Cogrev to get a 10% discount.

• 1:19Nathan Labenz?

  Our guest today is Flo Crivello, founder and CEO of Linde, an AI assistant that aims to put your life on autopilot, online at lindy.ai. One of the smart but already increasingly mainstream takes in AI right now is that the next big thing will be actions. That is, taking the AI paradigm beyond generating inner text and allowing models to use tools to take actions and ultimately get things done. There are a number of ways to go about doing this, but they all involve using a language model to bridge the gaps between a user's request, typically specified in natural language, and an action space, which could combine APIs, web browsers,

• 2:03Nathan Labenz?

  other software applications or interfaces, and eventually even robotic controls, all while taking into account the user's history and preferences, which unfortunately for developers, tend to be spread out across email accounts, Slack conversations, calendar history and much, much more.

• 2:23Nathan Labenz?

  Now, people have imagined AI assistants seemingly forever, at least from the Jetsons to her. But recently, they've become a central focus. ChatGPT broke through to the mainstream in part because of its accessible chat assistant-style interface. And in the context of AI safety, assistants have been described as a laboratory for alignment. To state the obvious, economically, an AI assistant that really works will be incredibly valuable technology. Indeed, Flo talks about Linde as a virtual employee. And with such a big prize to be won, he'll surely face competition from some of the biggest companies in the world.

• 3:05Nathan Labenz?

  Alexa and Google Assistant are already in millions of homes, and Siri is rumored to be due for an update in June. Still, Linde is no hackathon project. It's a serious effort with a dedicated team, backed by real resources and critical connections. Flo has raised $50 million, and his team was building on GPT-4 for a month prior to launch. They've got a level of ambition that sometimes creates life-changing products and generational companies. So while Linde is being announced today and currently still in very limited beta, in keeping with our goal of helping listeners understand and, where possible, experience the near-term AI future, we did get Flo to agree to prioritize anyone who mentions

• 3:50Nathan Labenz?

  the cognitive revolution when signing up for their waitlist. One note: Flo does give me a screen share to give a demo of the product during this interview. And while that part of the conversation doesn't go on for very long, and I think you should be able to follow it from the audio alone, we do have video available on YouTube. And you can also watch some flashy demo animations on their site at Linde AI as well. Now, I hope you enjoy this preview of the AI assistant future with Flo Crivello. Flo Crivello, welcome to the cognitive revolution.

• 4:25Flo Crivello?

  Thanks for having me.

• 4:27Nathan Labenz?

  Really excited for this conversation. You are launching something new, and I think it is something that is going to, you know, you and others who are exploring this space, I think, are going to really change the way that people use computers by harnessing the power of AI and making it useful for practical, everyday tasks. So tell us about what you're building, what you're launching, and let's get into it.

• 4:53Flo Crivello?

  Yeah, so we're building LinkedIn.ai. So it's an AI assistant that can basically do stuff for you. Like, you can think of it as ChatGPT that can also use your applications and access your data in order to make your work. So it can own your calendar, own your email, book travels, send contracts, scrape the web, help you prospect, help you recruit, and all of that kind of stuff. So we've basically trained a large language model to be able to use tools and to train itself on how to use tools based only on the documentation of these tools. So we literally just feed it the human documentation of an API, like the Stripe API or the Slack API, and then it's

• 5:35Flo Crivello?

  just knowing how to use the API. And then we can give it a goal, and it figures out how to use which APIs when in order to fulfill its goal. So for example, if I go like, "Help me find half an hour with Nathan tomorrow," it's going to first hit up the Google Calendar API to find my calendar and my availabilities, and then it's going to compile an email and hit the Gmail API to send you these availabilities by email. And then, once you reply, it's going to hit the Gcal API again in order to put that time on both our calendars. So that's the idea.

• 6:08Nathan Labenz?

  We're actually going to do the screen share and this will be available in the video version of the podcast. So if you are listening, you can still listen along and hopefully it will be clear enough. But if you really want to see the thing in action, then flip over to the video version and you'll actually be able to see a preview of this product, lindy.ai.

• 6:28Flo Crivello?

  Here I have a sort of like big text field in the center. It's funny, actually, the first good name for the product was Google. It's like Google except it does stuff for you. And I'm going to type "find me ten software engineers in San Francisco." So I'm actually already using Linde for hiring quite a bit. It's really, it's making it 50x easier for me. So it loads a little bit and then it asks me, "Hey, you're going to use some prospecting credits?" And then now it just showed me a list of software engineers. And if I click, there's name, current role, LinkedIn. And if I click, I see actual software engineers. So this guy works at Stripe, this person works at Atlassian, and so forth. And there's a little bit of debug information here.

• 7:14Flo Crivello?

  None of this UI is hot coded. Basically, the AI has UI components that it just mixes and matches in order to do its job at any given time. So now I can follow up with these candidates and I can be like, "Draft them a recruiting email to join lindy.ai, a startup building an AI assistant." Or I could be like, "Draft them a recruiting email based on this JD." And here I could paste a Google Docs link or a Notion link and it would read the JD and draft the recruiting email based on that. Or I could go like, "Personalize the recruiting email for these people." And boom, it sends the email and I can just customize it here. Again, this UI is basically built by the AI

• 7:57Flo Crivello?

  at runtime and I can send the email. I'm just going to cancel here because I don't actually want to send this email. You can also ask Linde to help you with recurring tasks. So for example, here, and this one is a prototype, I can be like, "Before my meetings, send me the Zoom and attend these LinkedIn's and summary of our last interactions." And it's going to create this automation, which is that five minutes before every meeting, it's going to grab the Zoom link, grab my emails with these people, grab my meeting recordings with these people, and grab the LinkedIn of these people, and then send me a summary of all of that on Slack. Back to real demo, again, can grab the meeting recordings because it's currently already

• 8:39Flo Crivello?

  joining my meetings. So it joins my meetings. I can see my list of meeting recordings here, and I can click on any meeting. I see the like text notes for me. It summarizes the meetings, and then I can chat with it about my meetings. Like, what was the meeting about? What did Flo think? What was the takeaway of the meeting? Or I can go like, hey, can you update my sales force based on this meeting? Or can you create tasks on linear based on this meeting? And so on and so forth. So, you know, it's basically like an AI employee. And so just like an AI, just like an employee, you can talk to it however you'd like. You can send it emails, you can talk to it on Slack, or you can invite it to your meetings and collaborate with it however you'd like.

• 9:21Nathan Labenz?

  Boy, there's a lot here that I think is fascinating, and I'm excited to unpack. For one thing, I see that you had a monologue that went longer than the recommended max of 230. That sounds like the sort of AI coaching that I also need as somebody who has more than once gone beyond the maximum monologue recommended length. So I think that's funny. Right off the bat, we're seeing these kind of coaching type interactions, you know, coming back to us from our AI assistants. Like, in some ways, you know, they already know better than we do. But let's, let's kind of just, you know, unpack a lot of what we just saw, because you showed a lot

• 10:03Nathan Labenz?

  very quickly. First thing that I want to understand is just, how does this thing handle my accounts? You know, because I've experimented with some similar paradigms, even going back to the GPT-4 red teaming days. You know, one of the first things I thought was, holy moly, like, this thing is so smart, it might be able to sort of use itself as a tool. You know, it's so good at coding, maybe it can even like break down problems and sort of delegate the solving of those problems to itself recursively. And who knows what it might, you know, be able to accomplish then? What I found was one of the biggest challenges practically in getting that to work was

• 10:42Nathan Labenz?

  just, everything is protected by authentication and you know, it's hard to, especially if you're just kind of, you know, one person like I was at the time, exploring something new. It's hard to get enough stuff connected and to figure out all that like, security logistics to actually get something to work. So obviously in your demo there, you had stuff set up. But what is a new user experience and how are you thinking about that, both from like a user experience standpoint and ultimately, obviously, security is going to be critical to this application as well?

• 11:14Flo Crivello?

  So it's a little bit like onboarding an executive assistant. Like when you start working with an executive assistant, you hand over to them the credentials of whatever tools they are going to need to get their job done. So here, it's actually more secure than that because you don't give the AI your password, you just OS with Google or Linear or Asana and so on and so forth. So, and we make you OS with them on a just-in-time basis. So the first time you ask it, "Hey, help me find time with this," and it may be like, "Look, I need, I need your calendar and your email to do that." And then you connect with Google and that, that takes two clicks. Yes, it's funny what you say about recursively calling itself, because that is something that it does. So the way

• 11:55Flo Crivello?

  it works is that it basically writes code. And I'd love to get back to that because I think it's insanely interesting, the fact that it writes a one-time piece of code for this. But when you ask it to do something, it writes a piece of code to fulfill the tasks that you ask it. And in that code, it's got functions to call itself back. So for example, in the example that I just gave of summarizing your last interactions with someone, it's going to write a piece of code that's like, hit up the Zoom API to drag, to grab the recordings, hit up the Gmail API to grab your emails of this person, and then call myself back to summarize these pieces. Again, like to go back to the code thing, I always compare the moment we're in to, I think like, that's why like the name of the podcast,

• 12:40Flo Crivello?

  The Cognitive Revolution, is very apt. It's striking the parallels between this and the industrial revolution. Like the industrial revolution made some goods so cheap that we could now have single-use goods. Like we have like big pens, like disposable big pens and like solo red caps that you use only once and throw away. Like that was unthinkable before the year seventeen hundred. Like a cap was probably like the equivalent in two days dollars of, I don't know, like five hundred dollars or something. And now you have like caps that are like five cents, right? There is this crazy story of Napoleon at some point had a banquet with a bunch of world leaders and the lower leaders had like silver cutlery and the higher leaders had gold cutlery,

• 13:21Flo Crivello?

  the higher leaders had platinum cutlery and he, Napoleon, had the highest material of all. He had aluminum. And I'm drinking La Croix right now and aluminum has become so cheap, we literally throw it away. I'm literally drinking in aluminum and then I'm gonna throw this away. So this would have been unthinkable. So I think it's gonna be the same thing with code where it's like an app today costs of the order of ten to one hundred thousand dollars to build, while rapidly approaching a world where an app costs of the order of one or ten cents to build. And when that happens, you basically start having disposable apps. You basically start building an app just for this one session that you have right now. And that's basically what Linde is doing. You give it a use case, it's going to build a one-time app, it's going to write the code for that app

• 14:06Flo Crivello?

  for that one session. And then when you're done with it, it's just going to throw the app away.

• 14:09Nathan Labenz?

  That is fascinating. I've done that a couple of times already. You know, we're just to to date the recording of this podcast, we are at GPT four plus seven. As I've started, you know, marking everything from the date of GPT four release. And it has been kind of amazing to just, you know, a couple of times when I had a particular need, whether for a little data analysis or like make me a little chart, you know, out of this data, you know, GPT four can do that straight out of the box for some basic things. It obviously does not have all of the authentication and sophistication, you know, let alone like the sort of recursive, you know,

• 14:51Nathan Labenz?

  or access to a runtime. So there's a lot here that that you are adding on to the kind of core functionality. But even just with the base, I see some of that potential. I'd love to hear like use cases that you see for these kind of, you know, one-off, single-use applications, maybe how that kind of compares to the example you showed, where it was like setting up an automation, because that's not a single use, right? That's like an ongoing thing. And then I also really want to kind of get into the sort of consumer versus business paradigm here a little bit. And, you know, just thinking about that, I kind of wonder, like, is this something you think individuals will use kind of on an ad hoc basis and do whatever they need to do?

• 15:33Nathan Labenz?

  Or do you see this getting into business and kind of becoming part of process as well? Lot there, so take your time.

• 15:40Flo Crivello?

  The first ones that Lindy does very well is anything that you would give to an executive assistant. So the few very big ones are going to be emailing, calendaring, contract sending, prospecting, recruiting. So emailing, for example, when you wake up in the morning with Lindy, you open your inbox and Lindy triages your emails for you. She tells you, hey, those emails are like really important for you to look at right now. And she pre-drafts replies to the emails based on not only your voice, but your voice for that particular recipient. So she learns how you speak to each person.

• 16:22Flo Crivello?

  You probably don't speak the same way to your wife as you do to investors, hopefully. And so she's going to draft the email. And so you open your inbox in the morning, the emails are pre-drafted, and you can review them, you can edit them if needed, and then you can send them away. So that's one example. Calendaring, she can manage your entire calendar. You can be like, again, find me time with Nathan. She can handle conflicts automatically. She handles arbitrary preferences. So I can be like, hey, I don't want any meeting on Friday, so I don't want any meeting before 11:00 a.m. That's my focus time. And she's going to respect the preferences. So those are some of the very big use cases that it handles quite well right now. It's basically replacing an executive assistant for a lot of people.

• 17:03Flo Crivello?

  Regarding the consumer and professional thing, look, I think we're basically seeing a new type of computing. I think of this as the next operating system. And it's funny, it's actually an inverse operating system because normally an operating system leaves on top of your hardware underneath your applications and data. This leaves on top of your applications and data. And so you just start using this instead of using a lot of your applications and data. And it it just patches together all of your applications and data to do the work that you want to do. So at the end of the day, I think this is just how people, regardless of whether they're consumers or professionals, I think this is just how people are going to use their computers moving forward.

• 17:47Flo Crivello?

  I think that the computing experience of the future is not you doing work on a computer. It is you having a conversation with your computer. And then the computer works with you to do the work or even does the work for you.

• 18:00Nathan Labenz?

  Yeah, that certainly is the dream and it seems like it's becoming a possibility, you know, rather than just a dream extremely quickly. Could you tell us a little bit about how you decided that like this was the moment to pursue this dream? I mean, this in some sense, this like goes back to, you know, the Jetsons, right? Or like, you know, kind of post-war science fiction or even even, you know, to some degree, like wartime, pre-war science fiction. And it's always been, you know, kind of like one day, maybe we could achieve this. It seems like now, again, like everybody's kind of got the sense that this is coming into focus.

• 18:42Nathan Labenz?

  What gave you the confidence to set up to build a product? And how are you thinking about this moment that we're in where, you know, presumably you're going to see a pretty healthy amount of competition from other startups, but even, you know, more so probably like, you know, there's rumors of a new Siri, Google Assistant, you know, is going to get a lot smarter, we have to imagine. So how are you seeing your place in the landscape and the kind of opportunity to carve out a niche for your? Yourselves with this business.

• 19:10Flo Crivello?

  So how I decided now was the time. I mean, I feel like, and perhaps the audience knows that as well. Like, it's very obvious, right? When you're close to it, it's so obvious that like there is something unprecedented going on right now in AI. I have been following AI. I think I started to follow it up close more than 10 years ago. Like the image that moment sucked me in like for a lot of people, like CNNs and RNNs started to work. And I, at the time, I was a software engineer and I followed some AI courses on Coursera. And it was super exciting, but I couldn't come up with killer use cases back then. And I kept following along and I actually got really excited and I almost started a startup in this field when GPT-2 came about.

• 19:54Flo Crivello?

  And I wanted to start something in enterprise sort of GPT-2. I played around a little bit with it. I did some experiments and still it didn't feel quite ready just yet. And then, you know, GPT-3 came about, ChatGPT came about very recently. And, and, you know, when you start playing with these products, you get very good results very quickly. And then you're like, okay, now is the time when actually this is, this is starting to become very valuable. Regarding the competition, I'll say a few things. One, I don't think too much about the competition. I, you know, I have so many minutes in the day. So I try to spend them thinking about my customers and not too much about my competitors. I think it's Jeff Bezos who said, you know, your competitors aren't giving you any money anyway. So what, what do you think about them?

• 20:38Flo Crivello?

  I also think, I mean, it's going to be such a huge market that, you know, it's, it's just going to be the mother of all markets. I think there's going to be many winners in this space. I also think the Google assistants and series generally one handicap that these incumbents have is speed. I think it's Reid Hoffman who said, we are driving at night on uncertain terrain in the fog and no one can see very far along. And, and no one is exactly strong in that kind of environment. But I think that huge companies are even less strong. I think startups thrive in this kind of chaotic environment. So I think startups are going to have an advantage here structurally to play in this kind of environment.

• 21:19Nathan Labenz?

  How do you think that plays out in practice? Like, are there, obviously the speed factor and just kind of, you know, the willingness to ship without a million, you know, sign-offs and all that. Everybody's familiar, I think, probably with those dynamics. Do you see like specific? Features or use cases or kind of paradigms that you think you or startups in general are better able to embrace as compared to say, you know, a Siri or a Google Assistant?

• 21:49Flo Crivello?

  Yes, I think so. To go back to your question earlier about customer or professional, although I see this eventually as a universal new kind of computing paradigm. Right now, we are focused on the sort of, again, executive assistant use case for busy professionals. You have too much to do, you have too little time, too many meetings on your calendar will help you put order into this chaos. That is something that Google Assistant or Siri or Alexa can't do. They are consumer products. And these companies are so huge that they have to cater to a universal audience from day one, right? It's the same reason why, look, Apple has FaceTime and they've actually been fleshing it out. Now you can send FaceTime links in your calendar invites and all of that stuff.

• 22:34Flo Crivello?

  But Zoom is still a formidable company because they are focused on the B two B professional use case, which Apple structurally cannot go after. Yeah, I think that is one example of an instance where startups will have an advantage. It's not one specific use case, it's both the ability to focus on any use case versus having to cast such a wide net. I think that's one. Yes, to your point, like the bureaucracy and the legal rounds of review inside companies are another thing. I also think generally, yes, startups can be more nimble, they can change direction, they can make more experiments and they can take more risks. You can just allow yourself to release a product that is slightly more rough, ray on the edges, slightly less ready.

• 23:16Flo Crivello?

  And you don't risk what happened to Google where it's like they released Bard or they announced it and there were issues and they lost billions of dollars of market cap overnight. That's not a risk for you as a startup.

• 23:29Nathan Labenz?

  So let's talk a little bit about how it all works. I mean, you've given us a little bit of insight into that. But you know, I'd love to kind of understand. One of the big themes that we've, we've heard in, in talking to different entrepreneurs and builders is this kind of constant advancement of the model. And there's then also the question of how do we think about using like the best available model, which right now is pretty clearly GPT-4. I need to spend a little more time with Claude V1.2 as well, but I think it's safe to say GPT-4 is ahead for the moment. So it can do more and more stuff. Then you may also, you know, think, geez, that could be expensive. Or there may be some things that it can't do that we need to train our own models to do.

• 24:12Nathan Labenz?

  So we may have sort of a mix of models. So I'd love to hear kind of your thoughts on the, the mix of models first. And then I also want to talk about like memory and client profiles, but we'll, we'll come back to that. So tell me first, how are you thinking about what models to use? Is it a mix? Are you training your own?

• 24:28Flo Crivello?

  The criteria we use here to make the decisions or the quality of the responses, that's our single only criteria. And I really insist on the single only criteria. We don't care about cost, for example. So right now, that's, that's the guidance I've given as a team. It's like, ah, like GPT-4 is so expensive. And I'm like, I don't, I don't want to have this conversation. We're not talking about costs in this room. Right now, let's, let's, because most lies on your side. I think the, the cost of these models has been divided by 20 over the last year. Like this is not a problem. So right now, Lindy is costing us a pretty penny to serve to customers. It's, it's dozens of dollars per month per, to your customer. And that's just okay with me. I don't, I don't really care about that just yet.

• 25:10Flo Crivello?

  With that said, so right now, yes, GPT-4 is head and shoulders about everyone else. And we've tried pretty much everything out there. But it is often surprising to people to what extent you can actually get better results than GPT, at least GPT 3.5 on at least one narrow use case. If you build your own model based on that one use case. To your point, I think the landscape is changing super rapidly right now. And so what we've had to build is we've had to build the infrastructure to be model agnostic and be able to swap out the model super quickly and retrain new models very quickly once a new open source option out there starts working.

• 25:52Flo Crivello?

  Right now, we mostly run on GPT-4, but we also are building and fine-tuning our own models, and they are getting better very rapidly on our benchmarks. And so eventually, I think we're going to have a mix of GPT-4 and our own model, and again, purely for quality reasons. So we have a ton of data that we have collected through a lot of means, and we're using that data to fine-tune our own model, and that actually makes it almost as good, and I think soon better than GPT-4, again, for our one specific use case.

• 26:23Nathan Labenz?

  I think what you said there about not caring about cost right now is definitely seems to be a trend and is definitely one that I subscribe to as well. It is crazy to think that just six months ago, I believe it was August of last year, was when the cost of like the mainline models dropped from six cents a thousand to two cents a thousand tokens. Then they dropped that further down to 0.2 cents per thousand tokens, and then reintroduced, you know, a higher price point with GPT-4. That difference is a, right now, it's a basically a 20x, you know, the the structure being a little bit

• 27:02Nathan Labenz?

  different, where GPT-4 has a different price for input tokens versus output. But it is interesting to hear that basically, yeah, just pay up for the 20x difference. All we care about is quality, and, you know, we kind of have enough faith in price drops continuing slash, you know, our own ability to engineer stuff that, you know, it will resolve itself over time. I do think that makes a lot of sense, but it does it does take some conviction about a view, point of view of where the world is going.

• 27:35Flo Crivello?

  And conviction and money.

• 27:38Nathan Labenz?

  Yeah.

• 27:39Flo Crivello?

  But it's not a tough decision to make, really. It's, I mean, it's, when you look at that curve, it's so smoothly going down. And so it's never bad against Morsler, and here it's even more than Morsler. It's an order of magnitude even faster than Morsler. So the question really, and that's a little bit of a mouthful, but does cheap get better faster than good gets cheaper? Our, our bet, and so far, history has always proven that out, is that good gets cheap very fast, and cheap doesn't necessarily become good.

• 28:11Nathan Labenz?

  Yeah, that's interesting. So how much easier has GPT-4 made your life in building this product relative to 3.5?

• 28:22Flo Crivello?

  A lot easier. It's pretty shocking, the improvement. I think the bigger context window alone is a huge deal. Like, you used to have to hack your way around the context window and summarize and embed and fine-tune and do a bunch of crap to not have to deal with the context window. And it's, the context window is still a constraint, but much less of one. It's really not something you have to worry about nearly as much. So that alone is a game changer. You also need to do a lot less prompt engineering for GPT-4. GPT-3 and GPT-3.5, you had to coerce them into giving you the output that you expected.

• 29:05Flo Crivello?

  And so lots of prompt engineering, it's like, you are like a smart assistant, you don't want to do X, Y, Z. GPT-4 works out of the box a lot more often. It works full shot, you just ask it to do something with zero prompt engineering around it, and it just works a lot more often. So it's made a pretty dramatic difference. I would say the only downside so far of GPT-4 has been the speed of it, or lack thereof. It's pretty slow, the generation is quite slow. And so you may have noticed during the demo just now, it takes a few seconds for it to answer. Oh, well, it's not the end of the world, because the way we think about it is like, even if the model takes 30 seconds to answer, which it doesn't,

• 29:46Flo Crivello?

  even if it took 30 seconds, that would still be way faster than a human executive assistant. So I think that in a lot of ways, the product is superhuman as an executive assistant out of the box. It's available 24/7, and it answers in like 10 seconds or less.

• 30:03Nathan Labenz?

  Yeah, so one of the things, I think in many respects, that is gonna undeniably be true. You know, for one thing, there just aren't that many executive assistants out there that know how to code. Although they can increasingly figure that out with GPT-4 as well, perhaps. But one thing that a human

• 30:21Nathan Labenz?

  Teammate is going to do that, I think you're also working toward, but it's not quite clear to me exactly how you'll accomplish it yet, is really getting to know you over time, you know, and knowing your tendencies, preferences, you know, you mentioned kind of relationships and talking to different people in your, your life in a somewhat different way, a different voice. What I see out there in the world right now is kind of the default paradigm would be like, we embed all your stuff and then at runtime, when you ask for something, we kind of translate that to a database query against the embeddings and then we pull out stuff and rank and then we take whatever kind

• 31:06Nathan Labenz?

  of pops up to the top of that ranking and stuff that into the context window for the language model to kind of use that information to inform what it's going to do. And I think that makes a lot of sense, certainly for things where I have like just tons of content sitting around, you know, or for companies who have like internal knowledge management systems already that they might want to layer a chat onto that makes a ton of sense. Are you doing that? Or it sounds like you might be doing something a little bit different because you said like the model writes code to search. It sounded like you were talking about like just searching my Gmail directly, for example. So how do you think about kind of ingesting all this content, embedding it, trying to have a semantic

• 31:51Nathan Labenz?

  search versus using like the many search APIs that exist? Or maybe it's a combination of both?

• 31:57Flo Crivello?

  So we do a mix of three things. There's basically, there's two kinds of things that the model needs. It needs your preferences and it needs your data. And we do three things to get those. So for the preferences, it's funny, actually, this is an example of a time when GPT-4's context window really helped because at first we were like, Oh, let's do what you just described. Like, let's encode your preferences in a vector database. Let's pull them when you give something. It's like, hey, like, help me find time with Nathan. And then vector sift against your preferences. This said, don't find time on Fridays. And so I'm going to inject that into the context window. And then we took a step back and we were like, wait a minute, you guys. Like, 32,000 tokens context window? Like, how many preferences do people have? Like, you're not going to have like 30 pages of preferences to give to your assistant?

• 32:42Flo Crivello?

  Like, so it's like, let's just stack all of that up in the context window and we'll deal with all of that later. And once you really have too many, and that works really well. So that covers the preferences. Then the two other things we do all for your data. So, for example, if I go like, write an email to Bob about this report that I just wrote in Google Doc or whatnot. There's two ways that Lindy is going to pull this data. One is using what we called a context injector that leaves upstream of the code generation model. So the context injector takes a prompt from the user and tries to figure out what questions it needs to answer, what context does it need to fulfill this prompt?

• 33:27Flo Crivello?

  And where can it find this context? And then it pulls this context from these sources and injects them into the prompt. So it sort of hydrates the prompt. So for example, if I go like, send an email to John. Suppose I literally just said that, send an email to John. The context injector is going to go like, who's John? Like, so that's the first question, like, who's John? Right? How am I going to find that out? It's like, okay, like, I have all of these sources of information. I have his contact, I have his calendar, I have his email. And I'm going to use a mix of all of these to find out who John is most likely. And so if I always work with the same John, day in, day out, and email him, he's gonna be like, okay, this is John. So that's the context injector. And then, even without the context injector,

• 34:10Flo Crivello?

  the last thing we do is that the code generation model, yes, is also able to write code to pull this context at runtime. So if I say again, write an email to John following this Google Doc, the Google Doc thing is probably going to be handled at the code generation phase. And the code is going to be like, hey, I'm going to hit up the Google Doc API, pull this Google Doc, summarize it, and then send an email to John about it. So that's the last part of how we inject this context.

• 34:36Nathan Labenz?

  And then, do I imagine correctly also that you're kind of building your own? Profile of each user that kind of lives in its own place, like in your database that, you know, that didn't exist otherwise?

• 34:51Flo Crivello?

  Yes, so we connect to all the sources of data that you'll feel comfortable connecting us to. So your email, your meetings, your documents, all of that stuff. And then, yeah, I mean, it basically knows you better than anyone and it can use all of the data to personalize everything. I wanted to highlight that privacy obviously is one. So we've actually laid out seven constitutional principles and privacy is one of them. It's actually funny, the etymology of the word secretary is secret. So this is a person that can hold your secrets. So we take that super seriously. You know, Lindy never leaks information and we as a company, or not in the business of selling this information, we will never do ads, for example. What we do is we, you know, we charge you money for access to Lindy.

• 35:36Flo Crivello?

  And so that that aligns the incentives quite nicely and we take the privacy super seriously.

• 35:41Nathan Labenz?

  Yeah, that'll definitely be an important selling feature to to continue to emphasize, I'm sure. The constitutional principles I also think are really fascinating. I mean, it's kind of a, on some level, it's like a company core values, you know, type thing. And that is fairly familiar, but I couldn't help but also call to mind Anthropic's recent constitutional AI publication where they basically spell out, you know, a small, I think it was just like two pages worth of guidance for here's what we want our AI to be. Here's how we want it to treat people, you know, we want it to be helpful, honest, harmless.

• 36:23Nathan Labenz?

  And then they've devised this sort of self correcting mechanism whereby they're able to use the model to critique its own performance according to its principles and suggest improvements that, you know, ways it might have been better and more in line with its with its principles. Is that kind of a paradigm that you're developing internally as well? Like, how deep does this constitutional concept run?

• 36:49Flo Crivello?

  Yeah, 100%. It runs a super deep. So we wrote down these principles, we put them on our homepage as a way for everybody to be able to review them and for us to hold ourselves to them. And then we use them at every level of the company. We use them when we make high-level strategic decisions, is that aligned with our principles? We use them when we make lower-level product decisions, like, hey, what do the principles say here? And then, yeah, we use them when we train the AI. So we fine-tune the AI on data, we RL-IF it, we RL-AI-F it, and, you know, we use these principles to do each of these things.

• 37:26Nathan Labenz?

  One thing that I thought was really interesting was principle number five, comfort with ambiguity, which has historically been a tough one for any computer system. Obviously, with the language models, we're getting a lot of kind of progress, you know, coming our way for free, so to speak. But I thought you had some really interesting comments when we were speaking about this, about how you want the tool to be able to do things for you, you really want to get stuff done, but you also have to be mindful of context and understand, you know, how confident are we that we're about to do the right thing? And, you know, what would be the cost of making a mistake? So tell me about that part of the paradigm.

• 38:11Flo Crivello?

  Definitely. So the image we always use internally about this comfort with ambiguity is we say that Lydette takes a message to Garcia. And taking a message to Garcia, that's an essay, it's like a very famous essay, it's about some war between the US and Cuba, and there was an American general who wanted to get a message to the head of the Cuban intelligence who was hiding somewhere in the mountains. Garcia takes a letter and he goes to a soldier and he hands the letter to him and he's like, take this message to Garcia. And the soldier asks, who's Garcia? And the general takes the letter back, goes to the next soldier, takes this message to Garcia. And the soldier asks, where is Garcia?

• 38:53Flo Crivello?

  And then the general takes the message back, goes to the field soldier and goes like, take this message to Garcia. And the soldier goes, done. Right? And so that's what we're aiming for. So again, the other concrete example that I gave just earlier applies here like, send an email to John. You don't want your assistant to ask you who John is if you're meeting with John day in, day out. Now, one of the other funding principles is reliability. Like, Lindy doesn't screw up. And so there is a little bit of a tension between reliability and compared with ambiguity. If you take too much initiative, you may screw up from time to time. And so, given a choice between both of these principles, reliability always

• 39:35Flo Crivello?

  wins. We never sacrifice reliability. And the top heuristic that we use to decide, hey, can you feel comfortable to do this or not? There's two factors that we use. We use perplexity and we use the stakes of the decision that Lindy is making. So the perplexity is, hey, how certain are you that this is the right thing to do? And then the stakes is, assuming you screw up, how bad is it? So for example, like, I will take the initiative to send an email and fire John. Like, whoa, whoa, whoa. Assuming this is the wrong thing, how bad is it? It's pretty bad. If you're firing someone, you're not supposed to fire them. It's pretty bad. So we use, we use both of these factors. And Lindy uses both of these factors to decide whether to move

• 40:20Flo Crivello?

  forward or whether to ask for confirmation. One, one other heuristic we use generally is, is the action that you're taking read only or is it read right? Like, read only actions, like, I'm going to go ahead and search his email. Or for example, if I go like, um, um, hey, find me a way, I'm moving apartments right now and I have cats that I'm moving. And so I would ask Lindy, find me a way to move these cats. And like, you know, I'm gonna take a plane, like, what are the requirements? Lindy could also go like, I'm also gonna research online options for cat moving companies. Right? So like, there's no risk involved in that. Read, write actions, oh, I'm going to send an email on your behalf. I'm going to make a purchase. I'm going to do all of that stuff. For read, write, basically Lindy always asks for confirmation.

• 41:04Nathan Labenz?

  Interesting. Okay, so I was gonna ask, first of all, about just the confidence when it comes to moving forward on a particular action. Is that something, I hadn't, I haven't actually had a chance to dig into this yet, but I had heard somebody say that with GPT-4, OpenAI was no longer returning the log props for, like, the top, you know, historically, they've returned the log props for, like, the top five most likely tokens at least, so you could kind of see what the, you know, the leading options were via the API. That's how I would assume you would do something like this, assuming you're using GPT-4, but I had heard that that function maybe wasn't there anymore, so do I have that wrong, or is there another strategy that you're able to use?

• 41:45Flo Crivello?

  No, that, that's right. That is another downside of using GPT-4 for us. My hypothesis is that OpenAI is removing this in order to remove the ability of competitors to distill their models, because you can use this perplexity to actually kind of build your own GPT-4 out of GPT-4, and I don't think OpenAI is too happy about that.

• 42:06Flo Crivello?

  It sucks, though. Well, our kind of use case, it really sucks that we don't have access to this information anymore. And that's another, that's another reason why we are looking into building our own model. Yeah.

• 42:15Nathan Labenz?

  So for the moment, what, what else do you have available to you in the absence of the, the log probabilities?

• 42:22Flo Crivello?

  We, we've trained another perplexity model. Just takes a prompt as input and the context and takes the action that you're about to perform as output and tries to figure out, like, hey, like, does this really follow that with high certainty? And we're training that model right now.

• 42:37Nathan Labenz?

  Gotcha. Okay, cool. That's interesting. And then on the action side, the main approach is just to present the action to the user for confirmation, which is, it makes a ton of sense, obviously. Do you think there's potential there also for, like, a sort of ensemble of models where, like, it sounds like you, similar to the, the perplexity side? I could imagine you having kind of a, you know, good judgment model that comes in and says, you know, is this a good idea or should I, you know, hit the brakes before, you know, the kind of more base action model does its thing?

• 43:11Flo Crivello?

  Yes and no. We do compose models together, but or rather agents and workflows together. But that complexity and that guardrails is, yeah, it's just something that's present every step of the way. It's just like we constantly apply guardrails and we have many of them. And again, not screwing up and reliability is one of the priorities. We do though, again, we do compose models and workflows, which I think in a really cool way. So for example, you can go like, hey, um, when I ask you to send a contract, I want you to do that via DocuSign because there's like a bunch of alternatives out there. So go to DocuSign, generate the contract, I have templates, and then send them by email.

• 43:52Flo Crivello?

  That's what I mean by send a contract. So now I can be like, hey, send an NDA to Bob, right? And then I can go like, hey, when I ask you to onboard a vendor, what I want you to do first is send them an NDA, right? And so basically is that onboard a vendor workflow now uses another sub workflow, which is send a contract. And you can end up building an entire constellation of workflows like that.

• 44:18Nathan Labenz?

  Very interesting. So how, kind of two-part question then next. How far do you trust this today? Like, would you say, go book me a flight and just be confident you're going to get, you know, a flight that's going to satisfy your need? You know, what kind of percent hit rate do you think you would have on, on something like that? And what about actions that kind of can't be taken via API? I actually don't know if you could like, you know, you can search for flights via an API, but can you buy a flight via an API? And, you know, whether you can or can't, what happens when you need to like execute a transaction online that you just can't do with, you know, kind of generated code?

• 45:00Flo Crivello?

  I trust it quite a bit because I think we are doing a pretty good job with the guardrails. And it's asking me for confirmation before doing anything and it's doing a pretty good job at taking my preferences into account. So yes, I do trust it to like book flights. Hey, Miami to SFO is going to send me options. I'm going to be like option two and then it's just going to take care of everything else for me. That's amazing. The API thing, first of all, you would be shocked to see how much you can do via APIs. You can really do most things. From time to time, we will run into something that you cannot do. And we are building, that's not available quite yet, but we're working on that. We are building also a web browsing agent. So we are going to have that fallback where it's like, hey, if you need to use the web and not an API, you will be able to do that.

• 45:46Flo Crivello?

  And we're making fast progress on that as well.

• 45:49Nathan Labenz?

  You showed, you know, a little bit of the product and the user experience earlier. How do you think that evolves? Does it kind of stay like a single text box that you can kind of interact with anywhere? Or what is the, are we going full her here? Or what do you kind of envision, you know, my procedural daily use of this tool being over time?

• 46:14Flo Crivello?

  Yeah, I think it's going to be, I think it's going to have a lot of frontends and a lot of incarnations. And we actually view it as one of our key reasons to exist as a company, to build this frontends. We basically want to be to large language models as the PC was to the CPU. Right? It's like, look, like the PhDs did it. Like they, they gave us this amazing thing that works really well. It is not an end product. It's a miracle really that large language models are so usable today with just a text field. But they can go so much further than that. If you give them, if you package them up, you give them access to the right tools, the right applications, the right data, and you give them the right frontends. So yeah, we don't think of ourselves as building a text box. We think of ourselves as building that holistic product with all of these frontends.

• 46:58Flo Crivello?

  So yes, you will be able to send it emails, talk to it on Slack, invite it to your meetings, send it voice memos. Eventually, we do want you to be able to give it a phone call and it will actually respond to you. So to answer your question, it will, yeah, it will do all of these things for you.

• 47:12Nathan Labenz?

  You don't have to answer this question, but just calling back one of our earlier episodes that we did with the founder and CEO of Replica. What's your stance on erotic role play with Lindy?

• 47:24Flo Crivello?

  That's not a cool job to be done.

• 47:28Nathan Labenz?

  We'll have to, we'll maybe see what, what jailbreaks your early users might be able to pull out of it. You know, in all seriousness, you do lean on the safety work that OpenAI has done under the hood. And it'll be interesting to see too, how people choose to use something like an OpenAI, which, you know, for all the jailbreaks and whatnot that we have seen online, they've done a lot, you know, to bring it under control and have made a ton of progress versus, you know, we're just this last week. Also, we had the sort of llama release from Facebook and then the alpaca fine tuning of that coming out of Stanford. And then there was this brief moment where it was like, oh, it's just like text adventure three. And then

• 48:12Nathan Labenz?

  it was like, oh, no, it's not, you know, we're taking it down based on, you know, feedback from the community of like problematic uses. I don't know if there's a question there, but yeah, I wonder about the degree to which training your own models opens up like a whole can of worms in terms of safety and edge cases and, you know, just blind spots that you maybe don't have to worry about as much if you use, you know, such a well established provider. Any thoughts on that?

• 48:40Flo Crivello?

  It does open that kind of windows. And to me, that's like a whole other broader conversation around AI safety, which is like, look, I'm very happy that OpenAI is taking safety seriously enough, though even them, they can't really stop jailbreaks from happening and they happen all the time. And even if OpenAI was doing a perfect job, which they are not, you can't really stop other people from building their own models. And so whether we want it or not, we are going to have these models out there and we are going to have them do basically anything any human wants. So I do think that is something we're going to have to grapple with as a society.

• 49:17Nathan Labenz?

  Yeah, no doubt. It's, it's coming at us quick. Other things, obviously, that we're going to be grappling with in the wake of GPT four, OpenAI, along with a professor from Penn and another researcher, Dr. Bradley from Open Research, published a study of the anticipated labor market impacts of large language model technology. You can kind of read their charts and estimates in a few different ways, but to me, one big bottom line is that. It seems like they're kind of setting a lower bound of like 25% of all work. Yeah, it's kind of what they would seem to suggest is like the minimum amount of work that

• 49:59Nathan Labenz?

  language models could ultimately take on, especially as they're equipped with all the surrounding tools which you are, uh, which you are building. So I'd love to your, maybe your thoughts first of all on like, how, how do you assess that report? Does that like 25% number seem like low, high? And what do you think it's going to look like for us to adjust to this world where we have virtual employees? As awesome as that sounds, a lot of people are pretty worried about what that's going to mean for society.

• 50:27Flo Crivello?

  I think there's two time horizons, right? There's like the next five years or so, and then there's like AGI and ASI, right? Uh, AGI and ASI, all bets are off. Like, I can't comment on that. Like, no one knows what's going to happen. Certainly, I think it's going to be quite disruptive. In the meantime, I'm not too worried about job losses because that's just something people have been worried about forever. And I think it's just a failure of imagination for humans to realize like, human needs and wants are infinite. And as you free up humans, because now some tasks are automated, you actually free them up to do other things that only them can do. And so, look, I think something like 90%

• 51:09Flo Crivello?

  of the active population in the early 1900s were farmers. Today, it's something like 5%. So it's, it's a huge transformation. And there is very low unemployment, at least in the US. So not a huge concern. I think it's Mark Andreessen who's very fond of saying, you know, every, every quarter, I believe, uh, there's some official numbers that come out with the unemployment numbers. And so it's always like a net number. It's like this quarter, X thousand jobs were lost, lost or created. But it's always the net number that makes the headline. But the gross numbers are huge. It's always like millions of jobs created and millions of jobs lost. And then there's like a tiny difference that's like the net result. And so I think that the bottom line here for me is that economies

• 51:53Flo Crivello?

  are a lot more resilient and elastic and dynamic than people realize. And they can reconfigure themselves extremely quickly. So I'm not worried about that. I actually am very excited about our growing the GDP by 20% uh, and making people 20, 30, 50% more efficient.

• 52:09Nathan Labenz?

  So one thing that's not there, I wonder if you have a take on this, is the sort of Keynes vision from, you know, 80 years ago or whatever now, where maybe even more, where, you know, he famously projected that by this point in history, we'd all be working a lot less, right? The idea was supposed to be that we'd enjoy our like material comforts and maybe only have to work a couple hours a day. I think it was like 15 hours a week or something, which was kind of the, you know, the, the long range forecast. Obviously, nothing like that has happened. It doesn't sound like you foresee that either. Like you, it sounds like you more see people continuing to work hard for the foreseeable future, but just being more productive because they're able to delegate more stuff to AI. Is that right?

• 52:53Flo Crivello?

  No, I think Keynes wasn't too far off. Like it's, I think it was just too early and too eager in his predictions. But if you look at hours worked per year, per person in the US, it has been going down steadily over the decades. I think we work 30 or 40% fewer hours now than we did 100 years ago. So, no, I think that's also part of the solution. Like at some point, perhaps you reach decreasing returns and perhaps you're like, look, you know, we are actually filling up most of our needs and a lot of people are actually deciding to work less hard in order to have less and just to have more time. I think that is going to be part of the solution certainly.

• 53:30Nathan Labenz?

  Well, I'm hoping for some of that. It definitely seems like more leisure would be a good thing. And, you know, I'm personally very much looking forward to the era of robot servants, both digital and, you know, potentially even domestic robots and who knows what else too. So I know you're super busy. You've got a launch that you're doing and I appreciate you coming on to talk to us about this. Just a couple of kind of closing questions. One is, how do people find you? How do they find the product? Like, how do they sign up? And then, you know, we're, we're going to do a little, as the episode releases, we'll do a little promo online where we can get a couple of people off the waitlist and into early access. But just for the general audience, like,

• 54:16Nathan Labenz?

  Where do they go to sign up? And what do you think the timeline is to really ramp up the user base?

• 54:21Flo Crivello?

  So you can go to lindy.ai to sign up to the way to list right now. It's a very limited private beta. But yeah, any cognitive revolution listener, go there and say in the notes of the form that you're coming from cognitive revolution and we're going to prioritize you. The timeline is we are onboarding new customers every week. And I think we will reach general availability sometime this year.

• 54:45Nathan Labenz?

  Sometime this year, it could be obviously, you know, next month, or it could be nine months from now still. What do you think are like the big things that you need to iron out? Or the sort of, do you have like a shortlist of kind of key problems that you need to solve before you can go really wide with this thing?

• 55:01Flo Crivello?

  Definitely, I think that the reliability of it is one thing that right now it's working well. But to me, it's very similar to self-driving where it's like, you know, there's like one fatality every million miles or whatever. So look, you actually need a lot of data. You need like three million miles or maybe half pay factor of like a hundred here to know whether you're actually doing a good job. So here it's the same thing. We need a lot of data before we're actually comfortable putting that in the hand of a lot more people. We're also just learning so much and so fast about what people want to use this kind of assistant for. And so as we're learning, we're adjusting our plans and we want to make sure that we've built and crafted an amazing medical product before we put it in the hands of the general population.

• 55:41Nathan Labenz?

  What AI tools are making an impact in your life today? Like stuff that anybody can go try, but that you have found yourself going back to?

• 55:49Flo Crivello?

  Apart from Lindy, you know, GPT-4, ChatGPT are pretty huge for me. The reveal blog post that I'm releasing as part of this announcement was written with GPT-4. I recorded my, I'm not a terrible writer, but I just hate it. And it takes me a long time to write one thing. It takes me like four or six hours every time. This time what I did is I just recorded myself. And I rambled for like fifteen or twenty minutes in the most unstructured way imaginable. And then I transcribed this using Whisper and I fed that to GPT four with some lightweight guidance on like the style I'm going for, which is colloquial, minimal, simple.

• 56:32Flo Crivello?

  And and I'm like, write a blog post with these guidelines according to that transcript. And it did it almost first shot and it took me only half an hour to write that blog post instead of six hours. So that's a huge one for me. There's also that, um, I think there's like Y C search or something like that. It's like some people, it's like a Heroku app. It's like some people indexed and embedded all the Y C videos and blog posts. And now you can search them, uh, which is pretty awesome. So you can be like, how should I think about hiring or how should I think about remote? How should I think about co-working spaces? Uh, and you get distilled startup wisdom in like five minutes. So like that one's been pretty awesome. What else do I use these days? I I use Whisper quite a bit.

• 57:15Flo Crivello?

  There is a Whisper radio that's called a Matt Whisper. Um, unaffiliated, I just think it's a really well-built product and I use it quite a bit for voice memos and and recording myself.

• 57:27Nathan Labenz?

  Interesting. You're pretty consistent. I mean, Whisper is a is a is one that hasn't come up a ton, um, but is awesome. But it is interesting. We're in this moment where for all of the insane proliferation of apps and, you know, Ben's Bites is dropping dozens per day on people still. In this show, when I have asked people what they use, it has been pretty consistent that like the main thing is kind of chat G P T. And there aren't, you know, there's been mention of a number of other things, but mostly it's people are like, yeah, I mostly use chat G P T. And then, you know, maybe one or two other things as well. So I wonder what ultimately that means.

• 58:08Nathan Labenz?

  It doesn't seem like it bodes super well for the like first wave of applications anyway. But obviously the second wave, you know, the more the kind of thing that you're building, there's just a lot more to it than most of the stuff that we've seen so far. So I do think that will probably change, but. I don't know, it is interesting. It doesn't seem like there's maybe that much room for like a thousand AIs in most people's lives.

• 58:31Flo Crivello?

  No, I, I, I think to your point, and this is one area where AI applications diverge from a commonly received startup wisdom. It's that I think a lot of these AI applications are a bit too narrow. So again, it's like YC search thing. It's like one tool to search YC videos. If you look at product hands, there's like hundreds of like AI tools coming out every week right now. People don't wanna have to manage hundreds of tools and hundreds of logins and hundreds of links, like they want one big tool. And so I think right now, ChatGPT is the one thing that comes closest to this one big tool. But I think again, tools like Lindy are certainly another attempt to create something that is very broad and general purpose.

• 59:11Nathan Labenz?

  Is there a limit on what this type of thing can do? I mean, you're positioning it for starters as kind of a, you know, AI assistant. But, you know, is it also going to become like an AI accountant and an AI lawyer for you as well? Like, is there any limit on how much this thing can do over time?

• 59:34Flo Crivello?

  You know, I, I think of it as like Amazon or Craigslist, right? And like these things are getting unbundled, right? Like Amazon's vision was, you know, the everything stool, we sell everything. But at the end of the day, we're not saying like this is literally going to kill all software, right? Like, yeah, there's also going to be point solutions for very specific needs, just like Amazon. There's like, there's also Zillow if you wanna buy a house. And there's like platforms if you wanna buy a car and there's platforms if you wanna buy your dress and like, you know, and furniture. There's like some, some verticals have like very specific needs. And I'm gonna use these verticals when I buy a car or house or a dress or like some furniture. But by and large, where everything else, I use Amazon. So that's, that's my mental model here. I, I, Lindy is Amazon and sure, you know, like when you're gonna wanna prospect, you're gonna use a CRM or like ZoomInfo and whatnot.

• 1:00:20Flo Crivello?

  But I think people are gonna have one big text field in which they type whatever stuff they wanna get done.

• 1:00:26Nathan Labenz?

  Okay, here's a hypothetical question. Let's imagine in a world not too long from now, a million people already have the Neuralink implant in their skulls. Now, if you get one, you will be able to think and communicate directly via your thoughts with Linde and, you know, the computer in general. You'll have essentially thought to text. Would you be interested in getting one in your own head?

• 1:00:57Flo Crivello?

  Uh, not at first. The, the, the, the privacy and the security issues here are problematic. So I think I would wait a very long time before getting something implanted into my skull, skull, uh, probably 10 years or more.

• 1:01:11Nathan Labenz?

  Yeah, you only have one skull, but answers on that have also been surprisingly varied. There are some early adopters among our, uh, previous guests. Okay, last question. You're setting out with a really ambitious vision to build what you hope will be like a big part of our, our future lives, certainly our computing lives. If you could zoom out even farther than that, what are your kind of greatest hopes for and greatest fears for the next handful of years as AI permeates all parts of society?

• 1:01:45Flo Crivello?

  I do hope we can get people rid of menial work. I think we're seeing so many humans right now, humans or AGI, right? We're seeing so many humans do work that they shouldn't be doing, that robots should be doing. Like mind numbing data entry work and stuff like that. And like, you shouldn't have to spend time going back and forth to find time and, you know, like playing with people's schedules. Like, that should be the job of a robot. So I'm, I'm very hopeful for that, for sure.

• 1:02:13Nathan Labenz?

  Any big fears?

• 1:02:15Flo Crivello?

  I think misinformation can become a problem. Mostly, I have my side to over the very long-term existential risks. I do believe there is a low, but non-zero percent chance that things go very wrong with AGI.

• 1:02:32Nathan Labenz?

  Would you venture any, um, you know, remedies, prescriptions, regulations, guidelines? And do you have any sense for how we can minimize that risk?

• 1:02:42Flo Crivello?

  That's very uncharacteristic of me, but...

• 1:02:46Flo Crivello?

  I think we need regulation. We don't trust private sector to self-regulate for basically anything that is touching safety. Airlines or extremely regulated banks or extremely regulated buildings. When you, when you build a new house that's extremely regulated for safety reasons. We need the same thing with AI.

• 1:03:07Nathan Labenz?

  Well, it's going to be an interesting few years, to say the least. And looking forward myself to getting off the waiting list and getting into lindy.ai, the AI assistant, Flo Crivello, founder and CEO. Thank you very much for being part of the cognitive revolution.

• 1:03:23Flo Crivello?

  Thanks a lot, Nathan.

• 1:03:25Speaker 02

  Omniki uses generative AI to enable you to launch hundreds of thousands of ad iterations that actually work, customized across all platforms with a click of a button. I believe in Omniki so much that I invested in it and I recommend you use it too. Use Cogrev to get a 10% discount.