How to Empower Creators and Innovators with AI-Driven Tools | Joey DeBruin | Glasp Talk #38
This is the thirty-eighth session of Glasp Talk!
Glasp Talk delves deep into intimate interviews with luminaries from various fields, unraveling their genuine emotions, experiences, and the stories behind them.
Today's guest is Joey DeBruin, a passionate entrepreneur, product builder, and writer dedicated to empowering creators and innovators to make the world a better place. Joey is the Co-founder and CEO of Robo, a revolutionary company streamlining the product-building journey for founders by eliminating traditional barriers with AI. Prior to Robo, Joey co-founded Backdrop, a large-scale program that supported over 10,000 builders in launching over 1,000 projects, which was later acquired by Seed Club. Joey's journey began with a background in neuroscience and has spanned roles at companies like ResearchGate, where he led product innovations and fostered growth in the scientific community.
In this interview, Joey shares his journey from neuroscience to entrepreneurship, his insights into building scalable tools with AI, and his vision for empowering non-technical founders. He dives into the challenges and lessons learned from his past ventures, his approach to product development and innovation, and the philosophy behind Robo’s transformative approach. Joey also offers valuable advice for aspiring founders and product managers, emphasizing the importance of clarity, experimentation, and a relentless drive to create impactful solutions.
Read the summary
Glasp
Transcripts
Glasp: Hi everyone, welcome back to another episode of Glasp Talk. Today we are very excited to have Joey DeBruin with us. So Joey is a passionate entrepreneur, product builder, and also writer focused on empowering creators and innovators to make the world better. He is the co-founder and CEO of Robo, a company revolutionizing how founders build their initial products by eliminating traditional barriers like, and company. Previously Joy co-founded Backdrop Build, a large-scale program supporting over 10,000 builders in launching over a thousand projects, which was later acquired by Seed Club. With a background in neuroscience, Joy carries funds, and leadership goals in companies like ResearchGate, venturing, and History. So today we will dive into Joy's journey, his vision for building innovative tools, and his insights into fostering growth and creativity in tech and beyond. Thank you for joining us today, Joy.
Joey: I'm excited to be here, thanks for having me.
Glasp: Thank you. So, first of all, you have a neuroscience background, and your first career was as a researcher at UCSF, right? But now you are building Robo. It's like an AI kind of company, and I wonder what inspired you to start Robo?
Joey: The journey from science to Robo is a long one, so I can cover some of the key moments. I think what inspired me to join tech and leave science in the first place was I've just always been interested in the unknown, helping figure things out, bringing new knowledge, information, and life inspiration. That's why I was originally attracted to neuroscience specifically, outside of all the sciences, because, especially when I started studying neuroscience in college in 2009, there's so much unknown. There are just huge topics that were like, “Well, you know, people dream and we don't know why.” It's just these kind of fundamental questions that are just fascinating. The reason I left science was because the work of being a scientist is very well known. So the steps that you need to take and the progression of your career and the path are kind of defined. So you know that's what attracted me to working in technology, where the problems that you are trying to solve are unknown, and how you conduct yourself as a person and build your career, it's also kind of unknown. So it is just the sort of blue ocean of work. That's what inspired me to start working in startups a decade ago, and really that's why we built Robo as well. So now what we see is the technology that we have, especially with AI, to build things incredibly quickly, incredibly cost-efficiently, which means that we should be able to just connect more and more people with an idea to building that thing. That's kind of my goal: always just to help more people dive into the blue ocean and into this crazy, stressful, chaotic world of building stuff that I just think is such an amazing place to be.
Glasp: I see. And you target non-tech founders at Robo, I guess?
Joey: That is one of the big groups. So, it's interesting. Non-technical founders are a lot of who we work with because they can't build their product for themselves, either because the tools that are available to them, no-code tools, just aren't good enough, or they need to build something that's a bit more custom. So being able to work with a Robo agency — so at Robo we work with Robo agencies, right? A Robo agency is a team of people that's using AI to build products way, way faster, way, way cheaper. Non-technical founders have always outsourced to freelancers and dev shops and agencies, and I think we're able to provide a much, much better product for them. But it's interesting because there, it's actually not the only kind of people that we work with. We actually work with some fairly technical teams that even have existing companies, and they just want to build a new product very quickly without having to hire an engineer. So there are a lot of reasons why you actually don't want to write the code yourself for a product. One of those reasons is that you can't because you're not a technical founder, but there's actually a bunch of other reasons as well, and we are a good fit for a number of those cases.
Glasp: I see. I saw your tweet actually, and then your customer said, “Oh I got, like, a 50K for two months, three months development,” but with Robo, you can quickly build it in one to two weeks with 5K. That's amazing, but at the same time I wonder, nowadays we have Deit, and also Cursor, and a VZ, from Replit. I'm curious how do you differentiate from those developer tools that have AI features in them?
Joey: So the way that we see the landscape right now is that there are a lot of — if you're an engineer, there are some amazing tools for you already. So the ones that you mentioned, Cursor, Vzer, Rollet, and all these, this is just making your life so much easier. But if you are not a developer, those tools might even market themselves to you, but the reality is that if you just try to — if you've never built something if you've never built a product before, and you just get thrown into one of these products, like you're, it's tough. A lot of decisions that you make, or things that you need to understand in building software that for someone who's never done it before, are just hard. We think that there are tools that are already being designed exactly for non-technical people, but those are not as powerful, and I think that's the way that the landscape is going to be, we think, for years. So if engineers are going to be able to — I mean, everyone's going to be able to build more. So no-code tools are going to get better, tools for engineers are going to get better, like everyone's going to be able to build more. But it's more so thinking about the market. So if you have $5,000 to spend to get a product built, that's a lot better than you could build on yourself, wouldn't you do it? I think that market is going to be very durable, and even actually going to grow in our opinion, even as the tools that you can use on your own get better.
Glasp: I see. Yeah, it's exciting. But I'm curious how it works in your case. Does the customer need to prepare, like, a PRD, or document what they need, and then you build it and ship it to them? Or do you provide — maybe not coding, right? You don't provide an IDE, but I'm curious how, let's say, if you want to work with you, how it works.
Joey: So we — the way it works is that let's say you come to us, you have an idea. You say, “Hey, I want to build a new startup. I have an idea.” We work with you to define the scope of that product. We have some tools that we've built, we use AI to do some of that, but there's also humans involved. We get on a call with you and talk about it. So some parts of that are just, you know, difficult, or not important to automate. The end of that process is that we have a defined scope for what you want to build, that we estimate. So we say, “Hey, we think this is going to cost $5,000, take us two weeks.” You might say, “Oh my gosh, amazing, I thought this was going to be way more expensive, way longer.” Then, if it looks good, we actually will deliver it for you, so build it. So you're — the clients that we work with are not — it's not like we're giving them tools for them to build themselves. Like we are acting as an agency, so we are delivering the product for you that you want. We'll check in every week, and we establish channels of communication so that we can get feedback. But it's not as if we are asking clients to do the work themselves. So the way that we can deliver it so much cheaper and faster really is just the process that we use on our side. So we are using a lot of different tools and building these workflows to deliver products faster, but that's not something that we expose to the client directly.
Glasp: I see. So I checked your website, but it's a simple page at this moment. There's no demos or videos. So in the future, you will prepare some demonstration so that people can understand it?
Joey: Right, yeah, we will. I mean, I think the best demos for us are always going to be just showcasing the products that we built. “Here's what it cost to build this thing.” Because our goal isn't to build, you know, one of these AI demos that blows you away about how the product works, right? Because our goal isn't to get people to use our product. We use our product, so we don't need to demo it. The only thing that you need to care about, as someone who uses — who works with us as Robo, is that we can show you a product that we built, and you be like, “Wow, this thing looks amazing. It works super well, it's pretty complicated. That's the startup that I want to build, is just like that.” And we can tell you that, “Hey, we built this in a week.” That's the thing, that's the demo. So over the next weeks, we'll start releasing more of that kind of demos, I think, but more like case studies.
Glasp: Cool. And I love you guys to do dogfooding, right? It's great for dogfooding. You're building something that you can use, and that's the best use case. At the same time, I'm curious, now you are working with some customers maybe, but what is the biggest challenge at this moment for you? I think before, pre-launch. I don't know if you guys are in stealth mode or not, but I'm curious what's next for you and what's the challenge you think.
Joey: It's like somewhere in that stealth, you know. It's stealth light, I would say. We're public, we're announced, we're out there, we're working with clients, but we're still iterating and changing a lot. The good news for us is that we've — we've done this several times. I think we've launched several products at this point, and we've gone through the full cycle of building and selling a company or a product. So we're very comfortable in this kind of messy beginning phase. But the hardest thing for us right now is when we say “building a product with AI,” what we don't mean is just using Claude or ChatGPT or — I mean every engineer is using AI, right? Or maybe some aren't, but those are stuck in the Stone Age. So everyone — I mean everyone on earth is probably, at least a high percentage of them, using ChatGPT to write something, right? So that's not what we're talking about. What we're talking about is really imagining the whole workflow, end to end. So imagine that you have to build 100 products in the next month. What would you do? What would you need to do to be able to do that? What resources would you need to build for yourself? What workflows would you need to automate? It's thinking about building blocks. So what we are trying to do is think about, how we construct the Legos so that any product that comes to us is something that we can quickly put together with the Legos that we have. It's not like — you know, there's a concept in a product, like software development, like boilerplates, right? Which is — it's almost like you get 80% of the way there and then you just do the last 20%. So it's kind of like that, but the thing is that in the age of AI, these boilerplates can be a lot more modular. So it's more like — it's more like Lego pieces than it is like a skeleton, I guess. So the hardest thing for us right now is choosing the projects to work with. So we try to be very selective about who we work with because not everything can be kind of Robofied yet. At some point, it will be, but for the time being, we have to be picky about the kind of products that we take on, both because we want to — we want to draw a hard line. Our line is we only take on a project where we think we can deliver it 10 times faster and 10 times cheaper than an average agency. And if we don't think that's the case, then we won't work with it. And so we have to be picky on that front. We also have to be picky about finding projects where we can continue to expand our understanding of how to do this better, how to — so it's not like we just want to do the same project over and over again. It's like we want to kind of find interesting newness to the projects that allow us to continue to expand our sort of infrastructure underneath. So that is the hard part right now. It's — we've been fortunate that we do have a good amount of people kind of reaching out, wanting to work with us. So we are able to find clients, but finding the right clients is a much harder question.
Glasp: I see, yeah, that makes sense. But have you figured out which space, product, or types work best for your use case? Let's say, is it a financing product, or is it more like — I don't know — a tech product, or B2B kind of, more like consumer software is better? Have you —
Joey: So it's — what we're finding, and again we're still early on, but it's — it doesn't work as simply as you might like, in that sense. It's not like it's just, “Oh, this industry works and this industry doesn't work.” It's more like there is a set of characteristics that might make something easy to do with AI right now versus harder to do. So as an example, some kinds of products, let's say an internal tool. Let's say that you're a big — I don't know — you're a big sort of sales-oriented company, and you have a thousand sales reps. Each one of those sales reps has a manager, like a sales manager. So you may have a hundred sales managers, and maybe all the sales managers are manually training all the sales reps in how to do something. I'm just giving you — this is a theoretical example. But, and so maybe you want to build a product that can help the managers of the sales reps train their team. But in order, if you were to hire an agency to do that, the first thing that you would want the agency to do is to go and talk to all the managers of the sales reps and understand, “What are you doing right now? What makes it hard?” Talk to the salespeople themselves, and say, “How are you getting this information right now?” So you need to do that product discovery, right? That's the hard part. And then you, when you roll out that product across your huge organization, there's a lot of stakeholder management, there's a lot of other things. So the complexity of that product isn't the software, it's the kind of organizational stuff around it. So you can't peel those things apart. You can't build that quickly with AI. So that's a project that's harder for us because it's harder to scope upfront. Anything we work with, we want to be able to scope it upfront, because that's the process, right? We scope it, and then we ship it fast with AI. So if we can't scope it because it is going to change every day as you talk to more and more people, then it's better to work differently. It's better to work in this kind of iterative, retainer kind of style of contract where you're just embedding yourself in that organization. And so that's not how we work. That's one thing. Another thing might be something that's very — I guess a good mental model is that AI is trained on all the existing code and products that are out there, right? So the kinds of stuff that's easiest to build with AI are where there is kind of, like, state-of-the-art, there's, or there's existing concepts. So if you are building something that is bespoke and has not been built before, it's going to be a lot harder to build with AI than something that has more examples. So a marketplace product is probably easier to build than something that's like a custom graphic design that you're going to have to build from scratch, or something with a lot of compliance. Like if you're building a healthcare app, maybe some of those healthcare apps are easy to build, but HIPAA compliance is not something you can automate, right? So it's a lot of decisions like that — and for us, there's even, on top of that, it might be that, “Hey, we don't have a lot of experience building,” or “Like, most of the products that come to us are going to be web apps. We don't want to spend the time to build the — build the Lego blocks for a desktop app,” you know? So it's a lot of decisions like that, more so than it is, “Oh, we can do fintech and we are — we can't do health — health tech, or our tech.”
Glasp: I see. Yeah, yeah, that makes sense. So you are working on Backdrop Build, right? It's, like, a program supporting, and helping builders build projects, and which is acquired by Seed Club. So I was curious, why did you start that project? And also what was the M&A process look like?
Joey: So it was, I think, relative to other sort of M&A processes, it was great. I mean, one of the things that made it great was that we're just really close with a Seed Club team, and we have been — I mean, they're sort of their investors of ours in some ways, or the Seed Club Ventures, they're the kind of venture arm. They're one of our investors. They've been customers of ours. They've helped us through every iteration. So we are just — we know that team — there's a lot of trust between us and them. And we've always had very complimentary products. So they run a more traditional accelerator where they select several amazing teams give them funding and help them go to market. I think they're just incredibly good at building distribution channels and helping people get attention and those kinds of things. We've always been good at building platforms and software and solving problems at scale because of it. So the reason that we decided to sell Build was that it was still growing and doing well, but we were looking at what it would take to make it 100 times bigger. So when we sold Build, we were running a program every single month. It was mostly automated in terms of the infrastructure and how it actually runs, with thousands of developers joining every single month. So the question is, like, okay, how do you — and making good money, but like, how do you make 100 times more money? How do you make it 100 times bigger? How do you support 100 times more developers? And when we looked at the options, like one of the options that we looked at was, like, it would be amazing to have a traditional fund or accelerator attached to this thing, because we see all of these great projects, and some of the best ones it would be amazing to take equity in those and give them more funding, right? But so then the question is like, well should we go out and start that accelerator? Should we go out and raise the funds ourselves? It just felt better to partner with someone that has already done that, that it's like really has experience doing that. So it was just one of those things where we just didn't feel like we were the best team to take it to the next stage, and we also had a very close relationship with Seed Club, and they were totally the best team to take it to the next stage. So it was just very natural.
Glasp: I see. Sorry, this is kind of a dumb question maybe, but if you can go back to the days of Backdrop Build, how would you do it differently? Do you have some learnings or some ideas to do differently? Or would you do the same thing?
Joey: Interesting. There are some things that I can maybe — I can start with what I would do the same. So I think the things that we got right were that we were very principled from the beginning on what we would do or what we would focus on and not focus on. So to be more specific, a lot of people, when they join any sort of accelerator-type program, will naturally want to network and meet other people and have that kind of experience. But doing that — so we were — one of the principles that we had from the beginning is that we wanted to make a program that was much, much bigger than sort of any existing programs that are out there. So we always, going into it, were aiming at having thousands of people in every batch. Networking is really hard to get right, especially if you're a community builder, like facilitating the right connections that are relevant, and you want to find people where they're excited to meet each other. So you don't want these one-way relationships where the popular people are more established, more successful, or just getting tons of requests from people that kind of could use their help. So it's hard to get that kind of mechanic right. It's easier if you have a small group of 10 people where you hand-select everyone and it's curated. It happens naturally. But we were very principled that this is a problem that's going to — we know it's going to be a problem for us. We know people are going to want it. We know that we're not going to be able to deliver it well within the model that we have. So we were very principal about, even the platform we built, there was no ability — there's — you could post, you could share updates, you couldn't comment, right? And people are always like, “Oh, how can I comment on this thing?” And for us, it was always like, “Well, the moment that we enable that kind of peer-to-peer commenting,” even though we — we love that like we love that energy, we want people to connect, but we start creating this expectation that this is a community where everyone should be networking. We wanted to help people spend more time focused on building and getting the support and getting support from our sponsors, which is really where a lot of the magic happened. So I think that's the — and it was — it's hard to do that. Like when you have people that you care about that are in your community that are asking for something, it's hard to say no, I think unless you have strong principles about why you're doing what you're doing. So I think we got that right. I think the things that we probably — it's like, in hindsight, you could always have gone faster through some of the learnings. I think we — I say that now, but we tried a bunch of stuff that didn't work. We, frankly, overbuilt in the beginning, I think like everyone probably. And we were doing a bunch of things that we had — like all these events in the first few programs, we were hosting events every week. We were, essentially, like panels. We would have to go and find the guests for these panels. So it's kind of like giving people advice on how to build. Those are valuable, but there's so much of that content out there in the world. So to provide something unique is just super hard, tons of work. So we ended up kind of removing a lot of that stuff and the program was just as good. So, going back, I think we probably could have just stuck to the simple mechanic that we knew was working and focused more on just growing the value of that by adding more people rather than building out the program to be more and more and more and adding more layers of complexity on it because just — it just slowed us down. It probably — we wasted a couple of months at least doing that.
Glasp: I see. This is kind of a random question, but I think you have a research background, right? So do you have any — do you think you have any advantage in having a researcher background and working in a startup? Also, conversely, is there any difficult point from the transition from researcher to startup? Was it a difficult point to adapt to a startup?
Joey: Totally, yeah. I think there's — it's a double-edged, two-sided coin. I think the thing that is a two-sided coin is being a researcher. I think you have a strong truth-seeking mechanic. So the whole — I think what people are attracted to science is because you have this — science is like the mechanism to establish truth, right? It's like the only way that we've found, as humans, to establish what is true. It's experimentation, all these kinds of things. I think a lot of scientists are attracted to product building and startups because it's almost like — it's like finding truth via a different kind of experimentation. You find what people want by iterating on different products until you find something they want, and then you've kind of established that knowledge that, “Oh, people want this kind of a thing.” So I think that's the part of it that translates well. And especially the reason that we have always been attracted to frontier tech. So we started in crypto, now kind of really building in AI because the need to find truth is just so acute there. It's like nobody knows what's happening, nobody knows the way that the world is going to unfold. So having that just curiosity to try to peel back all of the layers and try to find what is true is, I think, advantageous. But the reason it can also be a disadvantage is because, sometimes, you just like, especially at early-stage startup work, you can over-logic this whole thing so much. We've fallen into this trap all the time, where it's like you need to be able to see the map in this clear way and have tight logic. You read all — you can read a lot of strategy articles, and then you spend all your time trying to know the universe. And it's just like, you need to move fast and make a bunch of decisions and trust your intuition and trust your experience and follow your customer and all those kinds of things that sometimes, I think, founders that are a little bit less scientific have an advantage there because they don't spend so much time worrying about what's right or wrong, they just do it. So I think it is a —
Glasp: I see. Yeah, makes sense. So sometimes founders act as a visionary and therefore go with a gut feeling, but sometimes it doesn't work and they spend so much time. But after researching, and researching technologists, you chose a marketing manager as your first position. How did you learn marketing, and what was the transition?
Joey: I think it was pitched to me — my first job basically, the way it happened was I was working at Johns Hopkins in a neuroscience lab, like 2013. I had a friend who I went to college with, I played tennis with him. He started a company right out of school, and it was an ad tech company, basically like an ad blocker that allows you to donate, or basically like a new — a new tab extension. So every time you open a new tab on your browser, there is a nice little page there that has some interesting widgets and clocks and whatever, you can customize it however you want, but there's like one small ad at the bottom. You generate revenue from that ad, and then, you got to decide what charity the ad revenue would go to. So it's kind of like opt-in advertising for charity. That was the concept. So people don't like all these ads that are out there, but if you get to participate in the revenue and help it go to something that you care about, then you might be more likely to engage or see. So interesting — it's a company that's still around. But so I was talking to him on the phone, and he was like, “You know, how's it going? How's life as a neuroscientist?” I'd had some kind of not-so-great experiences with just bureaucracy and the slow pace of being a researcher, and I was kind of whining about that. I was like, “Well, science is cool, but being a scientist maybe is not as cutting, or not as fast-paced, as I would like.” And he was — he was like, “Well, you should come work for me. You know, there's this role” — he called it “growth,” I think. Yeah, maybe my LinkedIn or something, my title ended up being marketing manager, but he originally pitched it to me as, “You could be an internal scientist.” He was like, “Hey, there's this role, there's this thing called tech, there's this role called growth. The way growth works is that you just get to run experiments. But nobody's going to — there's no bureaucracy, it's startups, you can do whatever you want. You can run 10 experiments a month.” And I'm coming from research, where running two experiments a year is a lot. So I was — it was a great pitch for me. I was like, “Wow, no bureaucracy, run — I know how to run experiments, I'm a scientist, that seems cool, and I can run them fast and nobody's going to be micromanaging me. Sounds great.” So that's when I moved out to San Francisco and started working in tech and figuring out what growth meant. I kind of wound my way from there to the product.
Glasp: I see. Then after that company, I think you became co-host at Reforge, in 2017 for the growth series. I was curious, how did you meet him? Did he invite you? They have great content, and they host a great series. I'm just curious how it happened.
Joey: So I went through — Reforge is an amazing community. I mean, it's — at this point, still around and still getting better and better. I think they just have a lot of the best, most insightful content out there on product and growth. I went through the program myself kind of when I was — so when I first started in growth, I was like, “I have no — I mean, I'm a scientist. I don't have any idea how this works.” So I went through the program myself, and it was amazing. It was super valuable for me at the time. So I just stayed — that's how I met Brian. And I just kept — I stayed involved, and then they kind of brought on some alumni to help host the future programs and mentor people, and I was able to do that, which was awesome. Yeah, I still kind of stay connected to that. I wrote a bunch of articles for them. People are often like — they read one of those articles and reach out, and I end up reconnecting or helping advise companies that way.
Glasp: Do you still keep in touch with them, like Brian and other folks?
Joey: I do, yeah.
Glasp: Cool, nice. By the way, you — I saw you went to On Deck Fellowship, and I went to On Deck, too. I saw you — I saw that in your Twitter profile. I see mutual friends, interesting. I was also kind of a researcher before — not “before,” but I was researching — I was a chemist before, in the Master's program. So I published a research paper before, then I used ResearchGate to track my research papers and how many citations I got, and how many views I got. I liked it. It's a very cool company. I was lucky to work there. I'm curious about the community aspect of ResearchGate. How — what — as you were the director of product management, and also head of product management, how — what did you work on? What experiments worked well for the community and product? Could you share some insights and stories behind ResearchGate?
Joey: It's interesting. For one thing, a lot of people, I think rightly, probably call me a community product builder. But it's funny, because I never really consider myself that, I think because maybe I have some concept of what building a community is that doesn't feel like it is exactly what I've done. But maybe the better way to frame that is that I think I've learned that a lot of companies, especially tech companies with very strong communities, the community is actually just built around a product that is just delivering a ton of utility to people. So you don't work on the community at all. The community is like an emerging property from the quality of the product that you deliver to people. So the community happens very organically and doesn't take a lot of work. What takes a lot of work is building the product. At ResearchGate, there are some — the product is huge, it does a lot of stuff. I think for people who aren't familiar, ResearchGate is a kind of LinkedIn for academia. So 60 or 70% of every scientist in the world has a profile in ResearchGate and is active there. So it's a huge platform within science. Not that science is so massive, but that's like 10 million scientists, something like that. And the way that it is built is actually — it's a very complex data product under the hood because the unsolved problem before ResearchGate was that you have all these publications, right? So if I publish — if you publish something on chemistry, it's like a PDF. What ResearchGate did was, very early on, before there was a lot of — like how you can use ChatGPT to do PDF extraction, right? Actually, that was an unsolved problem years ago. So what ResearchGate started originally working on was, “Can you extract information from these PDFs,” such as, “Who are the authors? What institutions are they from? What are all the citations in the publication? Which other —” and then you have to do a lot of machine learning basically to know that. So let's say that one of the authors is Steve Brown. I don't know, there are probably dozens of Steve Browns out there that are publishing. So you got to know, “Okay, which Steve Brown is this?” Like, so you do a lot of machine learning basically to try to create what we call this “professional research graph.” So that's this graph of information that represents research from the point of view of the researcher so that we can say, “Okay, you, Joey, have 10 publications and we know all of them. We know all the publications that are citing you,” and all that kind of stuff. So a lot of what we worked on there is just extending the professional research graph because that is really the core asset that drives that whole flywheel. So some of the stuff that we worked on was, “Can we use machine learning to extract method information from a paper?” So we're trying to understand, “What is the protocol that you used to run this experiment?” And can we extract that from millions of publications so that we can create a page that would be, “Hey, if you're looking to use this certain CRISPR method, here are the publications that you might need to use?” And the more that you build out that research graph, the more products that you're able to create. So for example, during COVID, we were able to — we built a whole product around helping people navigate this flood of pre-prints that were being created. Because when COVID happened, everyone was publishing, but nobody's publishing in journals, because nobody had time to go through the journal review process. There's just a huge amount of information. So we built a whole product to help
people understand, “What's being published by whom?” Trying to make sense of that noise. But all of it's built on top of the same kind of underlying graph of information. So I think people misunderstand that company and how technical of a product it is. It really is like a lot of data, machine learning under the hood, and then the social network is kind of just the little bit that sits on top.
Glasp: Interesting. And also, how did you guys try to differentiate from Google Scholar? Because in Google Scholar, people can see their research activities, right? They don't have the social aspect. I don't know if Google was doing machine learning stuff, but is that the same thing you cared about?
Joey: Google Scholar is a great product, and I think it has become more and more popular over time. The thing that Google Scholar always had is just search, right? Because they're connected to Google, so obviously they're going to be amazing at search. Search is the most common way that people find research publications. So even scientists, they're mostly just searching on Google Scholar. That's kind of what drives the Google Scholar flywheel, is that people go to Google Scholar and they search, and that's what makes that product work. They have a lot of the same citations and extraction and things like that, and ResearchGate was actually earlier than Google Scholar. Google Scholar came along after. But I think what ResearchGate always — it just — I mean the honest answer is it was just never a core focus for Google. So they never — there's a lot of stuff that you want as a researcher: the ability to showcase your work and connect with other people and look for jobs at different universities and all the things that get stacked on top that Google Scholar never did. In addition to that, one of the things that — I mean when I left in 2021, a lot of the recent work was on this kind of publisher relations, because for a long time — this is more backstory — ResearchGate had a very adversarial relationship to publishers in some ways because publishers were upset that this big platform was being created that was now driving a lot of the traffic, because what publishers always have is distribution, right? Like that's their whole business model. So these — in the same way that a record label might have some beef with Spotify, the publishers had some beef with ResearchGate. But over the — like before I left, a lot of — I think the tide kind of changed and publishers finally were just realizing that this is the way the future is going to work and so we might as well get on board. So a lot of it was kind of building these pipelines of content between the publishers and the platform and doing reporting for them and helping them kind of run their businesses. So I think all of that is where ResearchGate spent a lot of time in the last few years.
Glasp: But did it take time? Because publishers are sometimes so mad, “Don't use our content.” How did you guys get along?
Joey: I mean I think it took a lot of time to understand each other. It’s frenemies, right? There’s — you can find all this online, but there’s lawsuits. That was always a thing, but at the same time, a lot of, especially the forward-looking publishers, just knew that you can't really fight. There are these bigger shifts that are happening not just in research, but across every industry, right? So consumers want aggregated information. People want to use Spotify. There's nothing you can do — there might be a different product that comes along, but it's going to look a lot like Spotify, right? Like that's the consumer experience that people want, and so you can't stop that. So I think the question is just how do you get on board as a publisher? I think — you know, it's a hard time to be a publisher, not just in science, but in general, right? So even newspapers are — all that — it’s tough. It’s a tough — it’s tough for a lot of — times — that industry is not doing super amazing. But I think the best publishers are the ones that have kind of tried to work with the tech companies rather than be adversarial to them. So it was always like that. It was lawsuits on one side and partnerships on the other, and just trying to bring people into the future.
Glasp: Thank you. And also, you mentioned — we also want to know more about your writer side. You have a newsletter called Flying Penguins. I love the name, by the way. I think you've been writing for over four years because the first newsletter I saw was in 2020. I'm curious why did you start writing the Flying Penguin, and how's it going?
Joey: I’ll say, I’ve been writing before 2020. I was writing for Reforge. I used to — before I started on a newsletter I was writing a lot of guest posts for different places. I think those were — ended up — I got my job at ResearchGate because someone — my boss there — read an article that I wrote. I got all these things that happened to me because I was writing. So I was very much bought into the idea that putting your ideas on the internet is a good way to have cool things happen to you. That's why, part of the reason I started a newsletter, is just because I very much believe in creating some serendipity for yourself, and I think that signaling is just the best way that I know of to do that. But it was also a good excuse — I've always been really interested in how to create flywheels for my learning. One of the things that, at the time, when we started that newsletter — I started with Rafa, who became my co-founder years later. So that was, in some of the early projects that we did together. We started this newsletter. It was originally called The Product Kitchen, very different from Flying Penguins. It was originally called The Product Kitchen. What we did was that we would review products and talk about — do a teardown for those products and what — what's good about them, like really, more design-focused. That was something that we were doing inside of the companies that we were working for, but within your company, you can do a full product teardown maybe every month because you're going at the pace of your development cycle. We just wanted a way to create faster feedback loops for ourselves so that we could review more products and learn and get better at our jobs. Maybe we could also find some interesting people who geek out on the same stuff as us in the process. So that was why we started it, and then it evolved, as all good projects do. I took it over and have been writing it on my own as — and kind of branded it to Flying Penguins, I think, in 2022, just because — it's a funny concept that felt like it fit a lot of things that I care about.
Glasp: Why the name Flying Penguin?
Joey: So there's this famous essay that this guy, Ronald Coase, wrote, I think — I can't even remember — early 1900s or something like that, called "The Nature of the Firm." It's like, if you go to business school they all make you read this article. It just talks about why companies exist, why firms are created, and when it's better to create a company versus just having something on the free market, right? This is like a classic article. Then this guy, Yochai Benkler, I think probably in the '90s or something like that, I can't remember the exact — maybe the late '80s — wrote this post called "Coase's Penguin." It was talking about how he was — specifically talking about Linux because Linux's mascot is a penguin. So Linux is like this open-source operating framework that's still now the most popular operating framework for a lot of really critical infrastructure. He was talking about how there are certain cases where, on the internet, you have a different kind of collaboration that can out-compete a company at providing the same good. At the time, in 2022, I was interested in how digital networks could help build new stuff. We ended up starting a company around that. So — I guess the thing that I liked about it is that the penguin is this kind of waddling, slow bird. Basically, when Yochai Benkler wrote that article in 1990, he was saying, “Linux exists, but these organizations are slow. They're kind of like — these online communities are very sclerotic,” or “They’re disorganized, chaotic.” So I wanted to talk about how I think, in the future, you have these flying penguins that are — they're also these digital organizations, but they're going to look a lot more sophisticated and be able to fly. So, broadly speaking, not that it matters, but that's what I care about. I just care about how the internet is going to help us collectively build cooler stuff. So I felt like it was a good title for that.
Glasp: I love that. I think we will put the link to your first newsletter. You mentioned that it's about the economist. Thanks, by the way. I think you are reading a lot of books, also newsletters, and you're learning a lot about products. After that, where do you keep ideas learnings, and knowledge? Do you use any second-brain tools or note-taking apps?
Joey: I'm very promiscuous when it comes to my stack. I haven't found anything that I'm super religious about. I use Obsidian as a note-taking tool and I love it. I've — one of my good friends, Sari, has a company called Sublime, that's kind of a way to save things that you find on the internet. So I love that and use it. But I will say that I'm guilty — I have a very — I never took notes going through university. I never took it — I just sat there in class and listened. So I'm one of those weird people who consumes information best by just keeping it in my head. That has its problems. There are things that I forget, but in general, my real note-taking — a lot of my organization is just whatever I can keep in my head. If it's good enough, then it stays in there, and if not, it falls out.
Glasp: But you have a pretty good memory.
Joey: Good enough, I guess, to get me through the day.
Glasp: Wow, that's amazing. For writing, nowadays people use AI tools. Do you use any AI tools for your writing or brainstorming?
Joey: I do. I think, like a lot of writers, I find the tools best for creativity. I often work by creating these agents with certain personalities, and then I'll go to that agent. So maybe one of the agents is like a creative, brainstorming kind of agent, and I just build these in Claude. I'll go to that agent and then say, “I have this idea for a kind of article, and these are the kind of points I want to touch on, and this is the audience I think is interested.” Then, I work with that agent until I have an idea, and then I'll go out and write it. I don't use AI to write most of the content. But then I'll feed it back into — like I have an editing agent who tells me which parts are crap, or what's confusing. So I do that. So I — at this point I don't actually — I used to work with editors and collaborators more, and I think AI is now good enough for me to be working mostly on my own, which is great. But I still do a lot of the writing myself.
Glasp: I see. Thanks. Since time is running up, you shared many pieces of advice and lessons with us, our audience, but do you have any — because our audience is aspiring founders and product managers, do you have any advice for them?
Joey: Of course we should use Robo. My advice to people is often just that — I think people overestimate how hard it is to build stuff and build your idea, bring your idea to life. I think what's hard is figuring out exactly what it is that you want to build. So a lot of people, they have — you go to a dinner party with them, and they're like, “Oh, I have this idea for this company. I've had it for six months,” or “Years I've been wanting to build this thing, and I just don't have the resources to do it.” If you ask that person, a lot of times, “How would you do it? What would be the initial version? Literally, how would it work? How would you sell it to people?” they don't have an answer. It's not their fault. It's just that you don't need to create that answer until you are in the process of doing it. So I think there is just so much value to trying to get insanely specific about your ideas. So take all of your ideas one step further. This happens to me all the time. It's like I have an idea for something, and then I go and — I just force myself to spell it out, “How would it work?” And then I'm like, “Oh, this idea is crap.” So I think there are a lot of people that have strong ideas that would maybe realize that they don't need to spend all their time wishing they had that thing anymore because it's not the right thing. Or maybe they do it and they realize, “Oh, actually, this is easy, this is something I can build tomorrow.” So I think people are a lot closer to bringing new stuff to life than they imagine, and the answer is just, to force yourself to go one step further.
Glasp: I see. Thank you. It’s great advice. This is the last question. Since Glasp is a platform where people can share what they're learning as a digital legacy, we want to ask you this question: what legacy or impact do you want to leave behind for future generations?
Joey: Obvious question, given everything that I've worked on. I just want to — when I look back on my life, to just have helped more people do stuff that they feel like they're able to invest their full selves into. That's kind of a broad answer, and so it doesn't need to be startups, it doesn't need to be products. I just think that more people should pursue things that they feel consume them. I want to make it easier for people to do that. So that's what I hope to spend the rest of my life and career doing, and that's what I want to be my legacy.
Glasp: Beautiful. I see — I see it from Robo, what you're doing, helping people realize what they need and what they want to do. It's amazing. Thank you for joining today. We learned a lot from you today.
Joey: Thank you for having me.
