No More Hustleporn: Ben Tossell on zero-tolerance bar for agent platforms
We pulled out the highlights from Ben Tossell's interview with Logan Kilpatrick. Ben Tossell is the founder of Makerpad, a no-code community that he sold to Zapier. Logan Kilpatrick was previously head of developer relations at OpenAI, and is now leading up devrel efforts around Google Gemini. Transcription and light editing by Anthropic's Claude, curation by Yiren Lu :-)
Highlights
Ben Tossell: And then it just grew and grew. It was a side project for 8-9 months and then raised a small round of $350K because it wasn't a VC company. It was never supposed to be that. I thought originally I wanted to build some big company and then realized pretty quickly I didn't. So I knew that for me, it's not a VC business, it was a lifestyle business. And I'm going to ignore all of the negative connotations of a lifestyle business because it was a great lifestyle and it's nice to have that kind of "I'm in charge of this thing, it just needs to grow, it doesn't need to grow at any level." And it did grow, and it was the only thing I've ever launched that actually had its own pull momentum where people were just paying. I wasn't trying to push it onto people to say "I need to get sign ups, need to get sign ups." It was just actually happening organically, which is really nice for a change. It'd been a few years where nothing was actually doing that.
Ben Tossell: And I'd like to say I have a list. It's not really a concrete like, checklist that I go through when I'm on calls with people. I'm mostly looking at developer tools because I just find the consumer AI world very difficult to judge. At the moment. It feels like there's loads of these companies that are launching and getting like 3 million ARR that are a product that is something that chat GBT already does or does in a, in a certain way and then is going to improve over time. And lots of them, I feel like I know they're the wrapper companies basically, is how they're described.
Ben Tossell: And I think, I don't know whether this is my lens of how I built company before, is that they don't need to be VC backed company and they shouldn't be, and they should just go and be rich from being a bootstrapper. Like that sounds great, you should definitely go and do that. So most of the time when I'm saying no is because I just think it's probably not a VC backable company.
Ben Tossell: So I'm really looking for if all of this AI stuff plays out and if every company in the world is going to touch AI or needs some sort of AI infrastructure, what are the picks and shovels that are necessary to build that world? And that's where I'm focusing at the moment. Not that I haven't done any consumer AI, but I'm just more comfortable with investing in people doing the building blocks.
Ben Tossell: And really I'm looking for founders that are just like outstanding and have done so much with so little time or resource because I just feel like that's so necessary in this space because everything's moving so quickly that you need people who are hacking away like tinkering on stuff, like putting stuff out there because they just want to and find it really interesting.
Ben Tossell: Matt Schumer, who is a founder I invested in, he's one of those folks on Twitter that's just like every time there's a new model, there's a new thing come out. He's one of the first people to ever put out a bunch of like hey, so I tested it on a bunch of these things I built into this system. This is what I found out. And it's just like, you can't keep up with him. So I'd rather keep up with him by writing him a check and let him go and do his thing. But they're the kind of people I'm looking for and similar sort of demi at Picolabs.
Ben Tossell: But the fact that they built that whole model with, like, four people in three months, that was on par with Runway, it's just like, it was just an insane, insane level of output for such a small team and such a small time. And I'm trying to get to the point where if I'm on a call, it's hard to say no, and I want to burst out saying yes during the call. And if I don't get that feeling, I've got, like, I take time off after it. It's probably not convincing myself to say yes then. And if I am trying to convince myself, I feel like I should say no to that.
Ben Tossell: And, like, they all need to be truly, like, billion dollar outcomes. So I think that's, that's always, like, a hard, hard requirement that needs to happen. I think, because I built make out in the way that I did, lots of the companies I see feel like you should just go and do this, but just like, raise some money from angels or raise, like, less than 500k if you need to go and build a team and that's what you're trying to do it for.
I think my tolerance for no experience or bad experience. Signing up for an agent platform is just like near zero. Either needs to work straight away or don't bother. There's one I signed up for. You connect your email, then you say, find emails in your inbox, like when they're most straightforward agent type use cases that people talk about. And then the first, like, link was a spam email. I was like, okay, I'm out of this. Like, that's. That's got to be baked into that product somewhere. Like, that can't be the user experience of someone. And I being in the space and knowing about it, like a normal person's gonna think, that's just rubbish. That's no, that's no good.
Full Transcript
Logan Kilpatrick: Welcome back to Around the Prompt. This is Logan Kilpatrick. Today we're joined by Ben Tossell, the creator of Ben's Bytes. We chat all things AI, automation, low/no code tools, investing in AI companies, and all the exciting stuff that Ben covers on a daily and weekly basis in Ben's Bytes. You're really going to enjoy this conversation, so stay tuned.
Logan Kilpatrick: Yeah, Ben, I'm actually curious, what's the profile of people who read Ben's Bytes? Does it tend to skew more like super technical people, or is it actually more like folks who are really interested and curious about AI but don't have a software engineering background, for example?
Ben Tossell: Yeah, I think the second biggest bucket is engineering, but I think it's mostly founder types or people interested in AI generally. Anytime we feature no-code type AI stuff, people seem to be really interested in that. We started off covering papers early on and then dropped those because people aren't really that technical. So if that's the bar of the really technical reader, then we don't do that because it's not our readership.
Logan Kilpatrick: I'm curious if you see back from when the low and no code space was booming, when you were doing Makerpad, when you were doing Zapier. I think there's still a lot of upside in that area. I'm curious if you're seeing any common trends between that space and what's happening with large language models and AI right now, specifically through the lens of democratizing who can use these tools. And I'm curious if there's any learnings from what you were doing with Makerpad that you think will play out in an interesting way with what's happening with LLMs and AI today.
Ben Tossell: Yeah, I mean, that was my whole spiel, I guess. That was my whole thing - I don't want to learn to code. I find it too hard. I've tried. I can't do it. What are the other options for me? I don't feel like you have to learn to code, and usually it's because you're trying to do a workflow or you're building something that isn't a really technical product.
Ben Tossell: I started building things like an Airbnb without code. I wouldn't suggest Airbnb being built on no code, but you could build that kind of thing. You can get customers, you could prove that without needing to hire developers or learning to code for nine months to then build a really crappy version of that. And then you're really emotionally invested in that thing as well.
Ben Tossell: I was all around democratization of software, really, and that came through no code tools at the time. This was 2018, early 2019. It was just that you could do more than you think without having to breach that technical barrier. And then that obviously leads to automation and lots of "if this thing happens, then this thing", using Zapier a lot for that.
Ben Tossell: Then it feels like it's almost a natural step to AI, where the code can be spit out for you. You kind of still need to understand how some of these pieces work, where you go and deploy it, and all the rest of it. And then there's agents and stuff. Naturally, people are building tools that let you build your own chatbots. Even Zapier has their own AI chatbot builder and things like that now.
Ben Tossell: So all, if you wait enough time, it'll all become accessible to everyone. I think it's just been really quick with LLMs. You've seen people like Andrew, who was a Zapier consultant, and he was that for a long time. I knew him from the no code space. More recently, he's using LLMs to generate the code, deploy it on Replit, and then he's now building more and more with code, not necessarily writing it from scratch. That's happening more and more often.
Ben Tossell: I think even with no code, people would get to a stage where they're like, okay, well, I want to build this as a real thing now. We'll take this into a coded platform. I think the best place for someone to learn that you need to do that is when you've built all the building blocks, you understand how software connects to each other. You understand that this is your front end, the database is your Airtable, in my instance. And then Zapier is like all the connective tissue, and you start understanding a bit more behind what does an API do, what's all the connective tissue, and where do things go. I think that helps you understand pieces of code. Not that I can write any now, but it's more I can understand some of what's actually happening in code.
Logan Kilpatrick: Yeah, that makes sense. I'm curious also if you have specific areas or ecosystems that you think are best suited, like if I'm a founder who has no technical experience, I should probably go and try to build something in these five different ecosystems because the other ones probably require a much deeper technical understanding. I'm just curious if there's specific industries or areas that you think are particularly well suited to use low/no code tools from the perspective of "I'm a founder trying to build something".
Ben Tossell: I think it lends itself to the indie hacker kind of tribe - people who want to build a business. It's not necessarily a VC-backable business. I think that's the perfect place for doing that. If you've boiled down a lot of businesses into their simplest parts, like an Airbnb marketplace, you can look at things like that and think, okay, I can build a marketplace or a community or all the sort of digital product information things. You can build courses or stuff like that.
Ben Tossell: Even SaaS applications, my brother builds with Bubble and has built our whole sponsorship system - taking payments, people can track things, we can look at a dashboard. All of that works with no code. So I think if you're not looking to build this huge company, which I don't think most people should, then that's where it sort of lends itself quite highly.
Ben Tossell: And even with AI specifically, like I said, you can build an AI chatbot. If you think that is your big idea, then you can build that with something like Zapier. That's what I'd encourage, because even if nothing else, it gets your first users in, gets some paying customers in, gets people to tell you how good or bad you've actually built this, and you'll understand what it actually needs to be much better. Whether you're then going to go and get someone to help you build it with code or anything else, it's more "am I building something that's worth pursuing and customers actually want and want enough that they're going to pay for?" Like, is this worth my time at all?
Ben Tossell: So yeah, I think it's the early stages, I guess it was what MVPs used to be. I think MVPs are very different now. It kind of feels like that.
Logan Kilpatrick: Anyway, just a quick follow up - I'm curious if you have seen the same thing, but I feel like I've been getting pitched on a lot of these companies where it's like "we'll automate the entire software creation process end to end and you'll just show up and be like 'I want a copy of Slack' in one line, and it'll just write all the software and create that." I'm curious if you've seen those pitches and if you have a perspective or opinion on whether or not that will actually be possible and be something that picks up steam. It seems like it makes sense, but it also seems wildly ambitious.
Ben Tossell: So this is the question - have I seen pitches that are like "will help you clone Slack with one line", or is it for software that helps you do things like that?
Logan Kilpatrick: Yeah, I think it's what's your perspective on those pitches? Like, do you think that's a viable, as someone who's done LLM stuff and done low code stuff, do you think that will actually work in the end, or it's a little too ambitious and not actually going to be practically possible?
Ben Tossell: I think people overlook why people are on these companies. I was listening to a podcast that was discussing 37signals and their product. They've got Campfire, which is a Slack alternative in its very basic form, and you pay for it. Once you've got to deal with hosting it, there's no real support, all the rest of it - all of that stuff is why you pay monthly or yearly for lots of these services, something like Slack.
Ben Tossell: I mean, I don't think I've ever needed to contact Slack support, actually, but the reason you are on Slack is probably because that's where people already are. Or if you're building a community on Slack, people love to complain about communities being built on Slack, and Slack is not friendly to those kind of companies, communities, myself included. I've definitely complained about that. And yet I'm still like, "yeah, but I'll just do the same thing, I'll figure out that around the rest of it", it's because people are just there.
Ben Tossell: Given what LLMs can do, I think there's a whole world of sort of single use software and single use tools that people will be able to spin up, deploy on their own machines and have it be really customized to what they're doing. But I think to say or look at a company and say "well, it's just a load of this kind of stuff, I can do that myself" - it's kind of the same argument that has been going around for years where technical enough people can look at Calendly and say "I can build my own version of that." And it's like, yeah, that's sort of where open source kind of Cal.com comes from. But there's still so much other stuff you got to think about when you're building things like that.
Ben Tossell: Yeah, I mean, if you're trying to clone something just for the sake of "that's got a massive market cap, they're missing out on this kind of customer, then I want to build the version for that" - then I just, I don't know if people are really missing the point so much as like, technically you could build it again, but that's not everything as part of that tool.
Speaker C: Ben, I guess diving a little bit deeper into Makerpad itself, I'm curious to hear from someone who is similar to myself as far as maybe not having the most coding experience or no code, low code. How did you actually go about creating Makerpad with that? What kind of tools and automation did you use on the backend so that individuals who don't have that coding experience can go in but still reap the same benefits as having coding experience or coding knowledge as well? Curious to kind of dive into some of that.
Ben Tossell: Yeah. So it happened through all the failures of other things I tried to launch. I was trying to launch a number of things after leaving Product Hunt, and I was building them all with no code because that's just how I could build stuff. So it would be using Carrd or Webflow and then Zapier and Google Sheets or Airtable as a backend. I've just launched thing after thing waiting for some spark of validation, I suppose the spark is anyone willing to pay me for this thing. Trying to look for a really quick validation on is this worth pursuing in any capacity? I never really got there with any of those.
Ben Tossell: Other than every time I tweet about something that I launched, people would always be asking me "how did you build this? I know you don't code, so how have you built this thing?" So it's the whole "those who can't do, teach." And I thought, okay, well, I'll just build a way to build all these ideas that I have. I'll just remember to record my screen and then I'll use that as a tutorial platform.
Ben Tossell: There was a Rails tutorial business that was doing well at the time. I was like, well, it's just that basically, for no code, there isn't anything like that. So I'll just create that. I built it with Webflow as a front end, Zapier as all the connective tissue, and then Airtable as a backend. That was basically it.
Ben Tossell: I'd then go and record videos and stuff with whatever I was using at the time, Screenflow or even Loom. I think the first few were Loom. They were really bad quality ones that didn't really have much editing and all the rest of it, but that was enough to get people interested.
Ben Tossell: Actually I launched it as just a Typeform saying "I will do this thing. These are all the things I built. I'm going to launch a tutorial platform. If you get lifetime access for $50" and had 15 people pay. So I was like, okay, I've never had that before. So that's something that's enough to go and do the next step, which is put up a website and actually put some tutorials up there. So that's what I did.
Ben Tossell: And then it just grew and grew. It was a side project for 8-9 months and then raised a small round of $350K because it wasn't a VC company. It was never supposed to be that. I thought originally I wanted to build some big company and then realized pretty quickly I didn't. So I knew that for me, it's not a VC business, it was a lifestyle business. And I'm going to ignore all of the negative connotations of a lifestyle business because it was a great lifestyle and it's nice to have that kind of "I'm in charge of this thing, it just needs to grow, it doesn't need to grow at any level." And it did grow, and it was the only thing I've ever launched that actually had its own pull momentum where people were just paying. I wasn't trying to push it onto people to say "I need to get sign ups, need to get sign ups." It was just actually happening organically, which is really nice for a change. It'd been a few years where nothing was actually doing that.
Ben Tossell: So yeah, that's how it was already built. And then it just grew from there. I hired a few people and we were off to the races.
Logan Kilpatrick: Got you. So it does sound like my follow up question was going to be more so as far as the market space when you were kind of going to market with this. So basically, there wasn't anything on the market such as Makerpad at the moment. There was some more that were a little bit more code demanding, but as far as no code or low code, Makerpad was kind of the first to market. Am I correct in saying that?
Ben Tossell: Makerpad was a buffet of all different things you could build with no code. It was kind of like, I know lots of code academy existed at the time, and that's very much learn to code. That was like a buffet of all the different pieces that you would want to learn. And you can choose what you want to do, but there are some paths within that where you're like, okay, I'm going to learn python, so I'm going to go through this system to get to proficiency and all of that.
Ben Tossell: But with no code, it felt like the bar was so low that you could spin up something in a few hours or a few days. So it didn't really matter about teaching people fundamentals beforehand and, like, having you go through some sequence, it was more just like, hey, sign up for Zapier, sign up for webflow, and sign up for airtable. And this is how we're putting all of this together right now, to build this one marketplace or this tutorial site, things like that.
Ben Tossell: So I just kind of felt like that's maybe how I like to learn, is just, if I want to see something, it's. If I look at something and it's like, that's interesting, I'll learn that thing. Or what actually happened a lot was people wouldn't necessarily follow the tutorial exactly, but it's more just showing them that you can do this kind of thing without code.
Ben Tossell: So this thing won't look exactly like with coding and learning coding, it's a right answer or wrong answer most of the time, whereas with no code, it's kind of like, well, you can build this slightly differently and have the same outcome. So I think people looked at it as an inspirational, oh, that's a really interesting way to put these tools together to build out that kind of system. And I can do that, but I've got a very different idea over here that does a very different thing. But those kind of pillars actually still make sense in my context, which I think is helpful because it's just then relatable to your situation, rather than people coming to you saying, hey, no, I want to build Airbnb for dog homes. I need to know exactly how to build that.
Ben Tossell: If you look at the Airbnb tutorial, you can do that basically, or you look at any marketplace tutorial, you can basically do the same thing. And then it's up to you if you want to try and figure out all these weird and wonderful features that's going to make that a billion dollar company. But, yeah, it was kind of show you how to do stuff, but also a bit of inspiration of, this is what's possible, by the way, so go wild.
Logan Kilpatrick: I was going to say, after the success of that platform, I guess I'm curious to hear kind of the conversation with Zapier and why Zapier kind of thought that Marketpad was a good fit to eventually acquire. What was the discussions between that and how did you get to that solution?
Ben Tossell: Yeah, as I mentioned, it wasn't a VC company. So I was getting to the point where it's either we've got to really figure out this business and grow it and do the next step. What is that next step? Not sure. Or is there a better home for Makerpad where that can still exist?
Ben Tossell: And actually, Zapier came to me because someone tweeted that airtable or Zapier should buy makerpad. So Wade sent me that tweet and said, hey, we should talk. And he'd been on the Makerpad podcast, and we obviously feature Zapier heavily on Makerpad. And, yeah, they just were interested and we just talked about no code. And when we thought the whole no code space was going and what is going to happen to everyone? Given the powers of building all of this software stuff, little did we know that sort of the LLM revolution would be coming around the corner as well.
Ben Tossell: It was more they want to acquire a community of people building with no code that are probably likely to use Zapier to connect all of the things they're using to build those things. So if you think about it like indie hackers and stripe, that as an acquisition is anyone who goes and starts a business or starts an indie business, they're going to need payments at some point. So they look at, okay, well, we'll probably use stripe and you don't need it to be really in your face of advertising stripe within that platform all the time. It's just in the back of your mind or in conversations there.
Ben Tossell: So that was similar to Makerpad and Zapier, where we never took money from Zapier to promote them in any way. It's just that's the tool that I liked using. So it was happy that. It was a happy sort of coincidence that it all worked out that way. And then, yeah, we wanted to join forces, but I think there was a big community that we had lots of people building stuff that either worked somewhere already or starting new things, and Zapier wanted to sort of continue to be on those minds of those people.
Ben Tossell: So, yeah, it was a pretty easy discussion, to be honest. It took a while, as all acquisitions do, but, yeah, it was a no brainer from our side. Yeah.
Logan Kilpatrick: Do you think there's an opportunity, Ben, for the maker pad of the AI space?
Ben Tossell: I'm glad you asked. So, yeah, basically, I've been thinking about Makerpad for AI. I even bought the domain whilst I was at Zapier, literally makerpad AI. And it was something I pitched internally when I was there. I sort of said, hey, we should do some AI tutorials. And actually we connected when I was at Zapier about doing some tutorials with Zapier and OpenAI and doing all that kind of stuff. But in that. In those conversations, I ended up leaving.
Ben Tossell: So I left after two years because I just. I just wanted to go and do my own thing again. And I had Ben's bytes up and running and that was just sort of. I just wanted to go in that direction and be my own boss again. Whilst doing that newsletter, I've just not seen anyone do this buffet style anything you're trying to get chat, Jimmy tea or Claude or any of these other models to do. How do I do this? How do I do this thing at work? How do I have this conversation with an AI model and get the thing that I'm trying to get done?
Ben Tossell: So that's what we're building, actually, and launching it shortly. End of March should be the launch of it. But, yeah, we're basically creating tutorials how you can use AI for work. So whether that's having a conversation with chat, GPT on generating blog and social ideas, or mimicking someone's writing style, all those kind of things that are little tasks, but they happen often. They happen across a broad range of different departments in companies, and people just want to know how to use these things.
Ben Tossell: I think the only things I've seen are courses about like introduction to chat, DBT code academy. Have one that did 60,000 people sign up for it or something crazy. And it's all the basic stuff. And like, there's loads of that basic stuff out there. People want to know real specific use case stuff like, I'm in marketing, what are all the marketing jobs tasks that I have to do? And then how do I just talk to an AR model about that? And I'm sure that in the future we'll start having things where there's a workflow involved and we use APIs and things like that. Yeah, that's what we're working on at the moment is a buffet style. How you can use AI for work.
Logan Kilpatrick: Do you have an yet? Is it going to be part of the Ben's bytes?
Ben Tossell: It'll just be Ben's bytes. So we're basically going to like, we've been a newsletter business since pre chat GBT, so we started before that and that gave us a nice bump. So appreciate that. And yeah, it's been a newsletter for a while and people love it. It's still growing. It's just not like, it's not the business that I love or I'm interested in the most.
Ben Tossell: So I've been thinking for a while about like, what does turning a newsletter business into a business powered by newsletter look like and what makes most sense and what are most of our readers looking at or like really interested in? And when. When it's on there, they're always clicking that thing.
Ben Tossell: So essentially it'll be on. It'll just be bensbytes.com. there's a tutorial library there. The newsletter will still carry on and it'll be, yeah, lots of tutorials. There'll be case studies. We've been doing these Ben Spice pro posts. So, like, looking at how companies are using AI internally, talking with people like stripe and other folks, like that clear bit, cb insights, how they're using AI, how they're empowering their employees to use AI.
Ben Tossell: So we're going to have case studies like that on the site, tutorials and then. Yeah, the newsletter as well. So there'll be a whole range of like, if you're trying to learn about anything in AI, hopefully it'll be like the one stop shop where you can find all that information. Then if you're one of these people who want to keep up on everything, the newsletter is there as well to email you every day about everything that's going on.
Logan Kilpatrick: This sounds so exciting. Me and Nolan have actually been talking for a long time about doing something in a very similar essence to this. I'm curious, do you have a team of folks? It sounds like there's a lot of stuff, and even just doing the newsletter every day and keeping up with all of the things that are happening in AI is hard to do. Do you have a team of folks helping you with Ben's bites today?
Ben Tossell: Yes, there's three of us. Three or four of us. Yeah, four of us at the moment. So my brother's working on building the site stuff. Then we have Keshav, who's been writing the newsletter with me since January 23. And then Shanice, who was my coo at Makerpad, she left Zapier, moved to Singapore, and then now she sort of hit me up and said, hey, what are you working on? Should we do something together? So I was like, okay, yeah, why don't you come and help with Ben's bytes?
Ben Tossell: And then the plan was never to do any educational stuff. And even after make, I was like, I'm done with education. I'm not gonna, like, go into that route again. I think it's just not seeing anyone else build it. It's like, I just know what that thing looks like. And I kind of built something similar before for a very different kind of audience, a different purpose, but it's just like, I think that. And people email all the time saying, hey, I want to learn about this thing. Do you have anywhere to go? I'm like, I don't know, YouTube maybe?
Ben Tossell: So, yeah, I think that's. We're just building that because I think the opportunity is there. I think there's a big need for it. And yes, we're excited about trying it in the AI space anyway.
Logan Kilpatrick: And I guess kind of pivoting a little bit here. Ben, putting on your investment cap, I know that you did a few investments into small, low code, no code AI startups. I'm curious from your perspective, what you look at when making these types of investments, especially with just the influx of AI companies coming to the market. I'm curious to get your thoughts, and I don't know if there's a specific list you go through that has hit all the different criteria or what that looks like for you.
Ben Tossell: Yeah, I had a fund called the Makerpad fund, which is where I invested in no code tools. There's a lot of no code tools that I invested in there, and then I now have the Bens bytes fund. So that is a fund dedicated to AI, if you couldn't guess that. And, yeah, so I've made eleven investments so far with the Benz bytes funds that started that was live from October.
Ben Tossell: And I'd like to say I have a list. It's not really a concrete like, checklist that I go through when I'm on calls with people. I'm mostly looking at developer tools because I just find the consumer AI world very difficult to judge. At the moment. It feels like there's loads of these companies that are launching and getting like 3 million ARR that are a product that is something that chat GBT already does or does in a, in a certain way and then is going to improve over time. And lots of them, I feel like I know they're the wrapper companies basically, is how they're described.
Ben Tossell: And I think, I don't know whether this is my lens of how I built company before, is that they don't need to be VC backed company and they shouldn't be, and they should just go and be rich from being a bootstrapper. Like that sounds great, you should definitely go and do that. So most of the time when I'm saying no is because I just think it's probably not a VC backable company.
Ben Tossell: So I'm really looking for if all of this AI stuff plays out and if every company in the world is going to touch AI or needs some sort of AI infrastructure, what are the picks and shovels that are necessary to build that world? And that's where I'm focusing at the moment. Not that I haven't done any consumer AI, but I'm just more comfortable with investing in people doing the building blocks.
Ben Tossell: And really I'm looking for founders that are just like outstanding and have done so much with so little time or resource because I just feel like that's so necessary in this space because everything's moving so quickly that you need people who are hacking away like tinkering on stuff, like putting stuff out there because they just want to and find it really interesting.
Ben Tossell: Matt Schumer, who is a founder I invested in, he's one of those folks on Twitter that's just like every time there's a new model, there's a new thing come out. He's one of the first people to ever put out a bunch of like hey, so I tested it on a bunch of these things I built into this system. This is what I found out. And it's just like, you can't keep up with him. So I'd rather keep up with him by writing him a check and let him go and do his thing. But they're the kind of people I'm looking for and similar sort of demi at Picolabs.
Ben Tossell: But the fact that they built that whole model with, like, four people in three months, that was on par with Runway, it's just like, it was just an insane, insane level of output for such a small team and such a small time. And I'm trying to get to the point where if I'm on a call, it's hard to say no, and I want to burst out saying yes during the call. And if I don't get that feeling, I've got, like, I take time off after it. It's probably not convincing myself to say yes then. And if I am trying to convince myself, I feel like I should say no to that.
Ben Tossell: And, like, they all need to be truly, like, billion dollar outcomes. So I think that's, that's always, like, a hard, hard requirement that needs to happen. I think, because I built make out in the way that I did, lots of the companies I see feel like you should just go and do this, but just like, raise some money from angels or raise, like, less than 500k if you need to go and build a team and that's what you're trying to do it for.
Ben Tossell: And there's just
Logan Kilpatrick: Yeah, well see, but Ben, the conversation, and you mentioned Matan. Im a huge Matan fan, not an investor and factory, but I think this idea of droids and agents and everything like that makes a ton of sense. I'm curious, actually, if, like, very tactically for you, you think there's a near future where, like, Ben's bytes essentially becomes an agent. Like, there is some agent that is just like running the newsletter, keeping up with all the things that are happening. It knows the writing style. You have, you know, hundreds of episodes or hundreds of versions of the newsletter that have come down. Like, do you think that's actually like a real possibility in the next few years for something like that to be like, completely automated away?
Ben Tossell: Yeah, I mean, I'd be surprised if we couldn't build something almost like that now, just finding the right person to do it. And I think what we, what we try and make sure we are doing is have human in the loop, I suppose, is how we'd say is that we're curating the stuff that we're seeing. So we see a lot of stuff. A lot of that. Most of it is like things that I've found and I'm saving manually through an automation into slack and stuff, but it's still like that. That kind of stuff feels difficult to then see how that would be automated because it might be through a Twitter feed that is on your for you tab, or I've just like, clicked on a couple of things, then all of a sudden I end up somewhere, like, I don't know how that sort of translates into an age, like, agent behavior, but certainly, like, writing it, putting it together could be more suited to an agent.
And then we have these summaries that we do the big news stories of the day, and we sort of use a mix of AI to condense some of the information and have Keshav sort of write some of the stuff himself. So I think we like having the human element as part of it. There's definitely probably things that we could have handed off to an agent, probably now if we looked into it a bit more I just feel like every agent experience I've had has never been great.
I think my tolerance for no experience or bad experience. Signing up for an agent platform is just like near zero. Either needs to work straight away or don't bother. There's one I signed up for. You connect your email, then you say, find emails in your inbox, like when they're most straightforward agent type use cases that people talk about. And then the first, like, link was a spam email. I was like, okay, I'm out of this. Like, that's. That's got to be baked into that product somewhere. Like, that can't be the user experience of someone. And I being in the space and knowing about it, like a normal person's gonna think, that's just rubbish. That's no, that's no good.
So, yeah, I'm sure. I'm sure we could do stuff, but then trust it until it's really robust and I've got Matt or Matan making it for me, then, yeah, we'll hold off on that. But I think they've got bigger things to do.
Logan Kilpatrick: I'm curious to get your perspective on this. This has been one of the reasons why I've always been so bullish on Zapier. And I think especially now in the last year, it feels like they have all of the building blocks to make really useful because they've kind of solved the connection problem of talk to your gmail, talk to notion, whatever it is. And like, to me, that's actually, like, the other stuff already exists. Like, the LLM layer is, like, working and successful. It's like all of the integrations that need to happen. And I'm curious if you think that's going to be like a. Yeah, I'm curious to get your take on that. But also, like, if they have, like, a outsized opportunity to be, like, one of the large players in the agent space as well.
Ben Tossell: Yeah, well, I hope so from a zapier shareholder point of view, and I agree with you, I think they've got all of the building blocks to really capitalize on this whole moment where normal people, knowledge workers who aren't technical, what is their entry into the AI agent space or automation agent AI stuff going to be like? I think Zapier is probably perfectly situated. Take advantage of that.
And I think Mike, one of the co founders, tweeted out, they've got AI bots coming soon. I don't know if it's like this week or next week. So I'm interested to see what that's going to be like. And obviously you can just assume what that might be. I actually don't know. But yeah, they've been building their stuff with interfaces. They can build your own chatbot. You can build all of these like rag applications and everything else that you see lots of people building from scratch. So it feels like a very good position for them to be in to have all that sort of ready.
Logan Kilpatrick: And then, Ben, before we get to our last question, and we'll link all this to in the podcast notes, but where can people find you? Where can people interact with you? Obviously we got Ben bytes, but where would you kind of position the people who are interested in following up after this, after hearing this pod?
Ben Tossell: Yeah, I mean, bensbytes.com, we do have the.com and yeah, on Twitter bentossel tosse double l. That's the best place to find me.
Logan Kilpatrick: And then we, we end every pod by asking our guests, you know, we're early on in 2024, but curious to hear from your perspective, one thing that you're very interested in seeing in 2024, and then one thing that you are still a little bit nervous on, that maybe you don't wish comes to fruition in 2024, but gets pushed down the road a little bit further.
Ben Tossell: GPT five, I guess, is one of the ones that I can say, looking forward to that. And then, yeah, just, I don't know. I'd like to see Google launching stuff without fumbling. Like, I think there might be some sort of big change over there and I think it should happen. And then like, people get excited about what they start doing as well. I think they've got all the ingredients to do things. Well, not currently, maybe with some of the. Just like different layers of people and process and policy, and they've just got to get all that fixed, I think, for, for an AI world. Otherwise, I think they're going to be in for a tough time. So be interesting to see it, especially from a newsletter writer point of view. It's giving us lots of stuff to talk about. So long. May it continue, I guess.
Logan Kilpatrick: Yeah, 100%. I feel like this is a great time to have a newsletter. There's so much cool stuff that's on the horizon, just even in the last week, it's been crazy to see. So, Ben, thank you for the time. This was awesome. Huge fan. Ben's bytes is incredible. If folks aren't subscribed, subscribe. It's the easiest way to keep up.
Ben Tossell: With everything that's happening in AI, guys, cheers.
Logan Kilpatrick: Yeah, thanks again for being here Ben. I appreciate it. Awesome.