Q3 2024 Public Board Meeting

Welcome everyone. Lots has happened since last time, or 3 months ago, May 15th. We've added a new product to our portfolio, which is Iffy. I'll share more about it in the future. We're also going to add another letter to the alphabet by changing the name of our company. Instead of expanding the title of every single YouTube thing, Gumroad Open Board Meeting, then it was Gumroad/Flexile Open Board Meeting, then it was Gumroad/Flexile/Helper. So instead of continuing to do that, we're going to change the name of the umbrella corp to Antiwork Inc. That is a pretty good summary of what we're trying to do with our product suites, which is to automate all of the unfun parts of work such that all the stuff that's left is actually fun, playful stuff like building features, designing features. And what's not fun is when things go bad, things break, you have to manage your cap table, dishes and laundry. Wearing clean clothes is fun. Eating is fun. How can we spend more time doing the stuff we really want to do? Some people try to motivate people by saying we're going to go to Mars. Some people need to work on that. I just want to make Earth better. Earth's pretty cool. There's water and trees here. The more time we can spend doing, hanging out with the trees and the water, the better. And the way you do that is you make robots do all the stuff we don't enjoy doing. Software is like a very early form of robots. Running Gumroad, Over 13 years requires work. Future product offerings will help eliminate a lot of the manual labor. Revenue should continue to go up and cost should stay the same or go down as a percent over time. Flexile, Helper, and Iffy should allow us to reduce our expenses as a percent of, uh, the revenues of the parent company. And if we're able to do that for Gumroad, then we should be able to sell these products to other customers. And get them to use it as well. The hope is that these products eventually supersede Gumroad in impact and scale. I had when I was 18 years old and I've learned some stuff over the last 14 years. In April 2024, we did $600K in net income, $650K in May, and then in June, only $132,000. What happened? Where's the difference? Basically, we paid taxes for the first time. So we paid about $600K in taxes to the US government. So that's a new thing for us, basically. For the last 12, 13 years, from 2011 to now, we've burned about $9 million getting to where we are today. We raised VC to do that. And the way taxes work is if you have net operating losses, you can discount them against future profits. This is why like Amazon has never paid taxes. We're no longer there. We actually have generated more profit in the last 15 or so, 12, since the pricing change, about 18 months, the first 13 years of the company. We now have to start paying taxes, which is a good thing because I like roads on top of the trees and water. A lot of people look at this. It's hard to like, I invest in Gumroad at $100 million valuation in March 2021. How do these numbers like, what is the connection? I highly recommend I send all investors our detailed P&L monthly. I highly recommend this. I can drop it into Claude or ChatGPT and say, I'm thinking about investing in this business. These are the metrics. Run a discounted cash flow analysis and tell me how much this business is worth. Or pretend you're Warren Buffett, how much would you value this? Or pretend you're Sequoia or whatever. And actually, it's a really good way to, I think, basically learn about how businesses are valued by different people. Tyler runs the company fund. He did some work for us putting together this valuation report as we're looking to potentially expand our alphabet. One way to do that in an accelerated basis would be to buy companies and products and add them to our portfolio. We should be able to run businesses better than the people previously running them. We'd pay them in cash and equity. And you know, you have to make a case for that equity being worth something. Tyler helped us with that. And he put together this nice report, which everyone can read in full. In 2021, we raised at $100 million valuation and then $125 million valuation. He considers that overvalued relative to the company's core metrics at the time. But over the next 3 years, we've made a lot of changes. Doubling our revenue, growing our net income by 800%. The business has grown into these prior valuations in a way that most private tech companies have not. Required some hard decisions, but Gumroad is much stronger than in 2021. And based on these new metrics, the $100 million valuation at which Gumroad currently issues equity to freelancers may be slightly undervalued. And compared to other public companies, of course, have liquidity associated with them. So they're always valued at the highest multiple that exists right now. Even everyone said the market crashed last week, but it went from a PE ratio of 27 to 23. It was like still very highly valued. And then anything in the $100 to $200 million range seems to be a fair risk-adjusted valuation for Gumroad equity. We're doing a buyback and we're doing this kind of auction process. So we're going to learn basically like how, what that number is for at least the pool of people already invested in Gumroad. And of course, I think $100 to $200 million is a reasonable range. We've grown the team. We added 3 people since the last board meeting. Now there are 30 of us all working hard to make that number go up because we all earn equity in the business and dividends and all of that kind of stuff. And how do we do that? We ship stuff. Let's talk about what we shipped the last 3 months and then what we want to ship next and get people's inputs on them. The core goal of Gumroad, just to remind everyone, is just help creators make more money. These are features we shipped to do that. First up, automatic abandoned cart emails. We shipped this last quarter. We allow people to spin up a workflow, say, hey, you left something in your cart. Very simple email reminder, but the problem is you have to set it up, which is annoying, right? Most creators don't want to log in and set it up. So we just set it up for everybody. So we just went through, set it up for everybody. And that led— that changed, I think before, maybe $1,000 a day or something like that. Now $12,000 a day in volume. This is $12,000 creators were not earning because someone added something to their cart and then left or entered in the wrong zip code and left. Now we send people an email reminder. And yeah, that leads to almost like a 3% lift in GMV, which is pretty crazy. We shipped gifting for memberships. Before you could only gift one-off products, like very simple because you're just absorbing the cost. But when you have a membership, you know what happens when you gift someone like 3 months, what happens at the end of 3 months? There's more complexity with that. We solve those things and you can now gift membership subscriptions to people, which is awesome. We shipped Bulk reviews page. So from your library, you can now leave reviews easily. Before you had to go into each product and leave a review. Now we have a dedicated page you can use to do that quickly. There are about 185,000 reviews on the website right now. About 1,000 reviews are added every single day. So about one every minute. So far 14 reviews have been added since this call began. You can also respond to reviews. So a lot of people are like, reviews are cool, but what happens when someone leaves a 1-star, 2-star review? You can now respond to them. They'll go to the person who left the, left the rating for you and shows up on your product page. A huge focus for us has been besides helping creators sell more to their audience, we want to drive sales. We've tried to turn Gumroad.com into a destination. It's like Main Street. We want it to be visually interesting. It's hard to build a destination site, but we can do that. We've made some good progress. We've added this new carousel to the top, which is really pretty. I really like it. We're also adding wishlists. Wishlists have done super, super well, surpassed my expectations. So we're kind of really doubled down on using wishlists as a way to drive discovery. We're adding this curated section that recommends products you may be interested in. This homepage generates about $20,000 a day in additional creator earnings. Discovery now exists in the mobile apps that we have. So both iOS and Android, you can now discover and buy, fully buy. Products using Apple Pay and Google Pay. Thousands of people open up Gumroad for iOS, Gumroad for Android. They want to listen to a podcast, read a book that they've purchased, consume a video. And all of a sudden there's a new tab, Discover, in which they can now buy stuff. It's still early days, about less than $1,000 a day are going through this, but there's a lot we can do to make this go up. For example, we're going to add search. So the core action will actually be search, search for products, filter, and buy. We think that will do pretty well. The Discover homepage, the majority of sales is still people searching for specific things or categories of things. That should help quite a bit. The other nice thing about the Discovery stuff is now that we have Discover in the mobile apps, we're more incentivized to make the apps better. The more we add to the apps, like analytics and sending emails, the more time people will spend in the apps and the more likely they are to actually click on that search button and eventually buy a product. Last quarter we shipped collaborations. Collaboration is an awesome feature of Gumroad because it's something that if you wanted to do outside of Gumroad would be incredibly difficult. We shipped this, but one issue is that we were basically taking the fees, the Gumroad fees out of the primary, the person who created the collaboration, which meant that sort of 10% went from— you sell something for $10, it's not a clean 50/50 split in terms of how much money you both make. The Gumroad fee is split equally between collaborators, which should make it a lot more popular, a lot simpler to understand. Over $1 million has been made with Flab. You can now sell things on Gumroad that are not necessarily digital products. This is a very big deal given that we started out as a place to sell digital products. So we've added 3 product types, services, commissions, calls, and coffee. Coffee launched a couple of weeks ago and calls and commissions just launched basically now live. Commissions allow you to collect a deposit upfront, 50% deposit, calls Similar to Google Meet or Calendly, you can set up your times and people can book. We want to build this into a single solution that you can use and manage everything at once. Coffee, similar to buymeacoffee.com or Ko-fi, people are probably familiar with. People want to be able to collect tips, they want to collect donations really easily. You don't need to give them a PDF. They just want to give you $10. There should be a really easy way to do that. And now there is. Now that we have these product types, we can make them more interesting and distinct. For a product like coffee, you don't need a cover, but maybe something on the side. Let's do demos. Are you ready for a demo? Connor, are you ready? You're not ready. Maybe you're ready. I'll solve. What do you call a monkey with 7 legs? 7-leg monkey. What would I call it? Okay, I'll start by saying we announced our, we launched this, which is really cool, public design system. So now we put our Gumroad design system, which I think is awesome. On Figma. If you have ideas, now that Gumroad has these 3 product types, there should be more. And maybe you want to be a designer at Gumroad, you have access to this design system, you could actually potentially build an idea for us to consider. So yeah, if you have any ideas of how to help, this design system is a great way to do that. It should also be a great tool for people who just want to learn, like, what do designers do? What does that actually look like in a sort of real context? And we'll be doing the same with all of our design systems for our products. It's great for sales and marketing because it shows everyone how awesome we are at shipping good, high-quality, pretty stuff. So proof that you can get a sense of even Flexile. Let's say you don't want to sign up for Flexile. You can look at the Flexile design system and be like, okay, cool. I can wrap my head around the values of what they're trying to accomplish. And also it's great for recruiting. A lot of designers who are interested in this kind of stuff will learn about Gumroad this way. I'm surprised more people haven't done it. If you want a job or freelancing gig. Use this design system, suggest an improvement to Gumroad. At the end of the board meeting, we'll ideate what we want to build next. Right now, if you wanted to suggest an improvement, you'd take a screenshot of Gumroad or rebuild it using Tailwind. You could actually drag and drop some stuff and end up with something pretty good quickly. Having it public would improve the quality internally. When it's internal, you get lazy and don't keep everything up to date. But if you keep getting emails like, "Hey, you messed this up. This or there's a typo here." We'll fix it up. Just like these board meetings, by the way. Public board meetings, I think, like, increases the quality of it. It's not purely a public good.

Awesome. Okay, I think people should be able to hear me now. So I'm just going to show you how to make a commissions product. You'll see the new product type down here. I'm going to make mine $10. When someone buys it, they're initially charged half as a deposit. When you upload files and complete the commission, they're charged the other half. On the content page in the product editor, we added a new element to our rich text editor. So you can put custom fields in on the content page after someone buys a product. They'd fill these out. We expect with commissions, there might be more detailed responses you want from people, and it isn't necessarily something you'd want them to fill out before they click pay. Ideally, you want people to check out and pay as soon as possible, have as little friction as possible between them adding the product to their cart and checking out. So that's what these are for. And you can use these with any products now. We designed them with commissions in mind, but you can use them for whatever products you want. I'm gonna publish this and show you. It says secure your order with 50% deposit on the checkout page. It says payment today, $5, payment after completion, $5. And when I get to the content page, you'll see that those fields are there. I can fill in my name, upload a file, and if I go to sales, you can see my name in the file. You can see the one charge for $5. And if I want to upload something, this is the commission I just completed. I upload the file and then I'll submit it, mark the commission as complete. And once that purchase goes through, there it went. You'll see there's a second charge. If I go back to the commission page, you'll see there's now a downloads thing where I can download the file that I just got. So that's how commissions works. We're excited to see what people do with it. Should I do calls too?

Yeah, calls.

I'll do calls. Calls is another new product type. You can add different call durations. So let's say you want to do a 30-minute duration. That costs $10 and maybe 60 minutes is $15. Then you can add in your hours of availability for the next week. I'm available from 9 to 5. You can put some limitations on when people are allowed to book calls. So in my case, I'll say they need to give me at least 24 hours of notice. And I don't want to do any more than 2 calls a day. And we have the standard product settings. So go ahead and save and continue, publish. And if you want to have some content to prepare people for the calls, you can do that here. You can specify different content for different call lengths or use the same content for both. I'll show you the call page. You have 2 options, 30 minutes and 60 minutes. And then you have a calendar where you can see the availability for the next 5 days. So I'll do Tuesday, 12, go ahead and add it to my cart. And if I were to check out here, it would schedule it for then. This is on production, so I'm not going to do it just because I don't want to pay myself $30, but that's pretty much all there is to see. Once you buy a call, the seller can specify Zoom or Google Meet URL, but it's just like any other Gumroad product. And I will wrap up by showing you coffee. I don't have one already. So we have a restriction. You can only have one coffee product. I don't think I do. It looks like I don't. You can specify up to 3 suggested amounts. So I'm going to say 10, 20, and 30. There you go. And then you can do a custom call to action here.

I'll say tip.

And published it. The reason you can only have one is that it uses your Gumroad domain. Mine is kunerman.gumroad.com/coffee. You can see the 3 amounts that I put here, 10, 20, 30, and then other is available. Let's say I want to be really generous and do 100. Click that, it'll take me to the checkout page. Happy to answer questions.

Are you thinking of ways for people to put the details before paying?

You can do that with normal custom fields feature we've had for a long time.

But the nice thing about having the custom fields later, for example, you want to do a purchase before, you can do that after, and then you can add a tip. Maybe talk about tipping next. We should add that in the box. For the commission product, is the advance amount percentage fixed at 50% or can it be customized?

It's going to start off as just 50%. If a lot of people want to customize it, if you have good reason to, to have it be different, then write it into us, tell us. It wouldn't be a whole lot of work for us to support different percentages, but as of now, it's just 50%.

Basically, the question that was answered is the commission 50%, that'd be customized. Not yet, but soon, probably. If people ask for it, just ask for us, ask for it, and we'll— it shouldn't be too much more work to add. And then one more question. How about tipping and buying the product together? Add a coffee and another product.

You can do that. Gumroad products you can buy together applies to those as well.

Someone said I've blocked them on X, please unblock me. Give me an email with your handle and I'll unblock you. Sorry about that.

Those are the new product types and we hope people make a lot of money with them.

Yeah, money. Just as a recap, we got these 2 product types, commission calls, coffee. That's what we got done for Gumroad. Not too bad for 3 months. Uh, and then if you're more curious about what does this mean for the business, someone asked, you only gave us 3 months of data. Well, we, there's a lot of public board meetings out there. You can check out for the nuance. This is our GMV, it's monthly GMV since the beginning. This is our revenue. The core difference was that in Feb 2023, we increased our pricing and hopefully we've justified that. Hopefully this was a good example of a bunch of features that will help creators earn more. Tipping has bumped up GMV by about 4%. That's Gumroad. Now we have 3 more products to talk about. Our second product is Flexile. Flexile is to Gumroad as AWS is to Amazon. It's a way to make our core business run more efficiently. To start with the numbers this time instead of ending with them, it's really not growing that much anymore. In July, we lost our largest customer. In July, we'll be down, sadly, because every month was up and to the right for quite a while. While we do have product market fit to some degree, if this was like its own little business, probably would not be enough to sustain development, but that's okay because we're not looking to convince anyone to give us more money. We have enough money and this is what we are choosing to invest in and we will continue to invest in it because they create enough value for us. And over time, more and more people will get that. And it's going to take many years to get these products to where they need to be, impact the world. So what did we get done? Last quarter we shipped Roles. This quarter we shipped trial periods as well. So now you can trial someone, you can give them a reduced rate during the trial. In our case, we were trialing people at, let's say, our software engineer is $150 an hour. It's really hard to generate that much value quickly. A reduced trial rate allows us to trial a lot more people, be a lot more flexible with a potentially lower bar to allow someone to trial with us instead of this, oh, $200 an hour. Also allows people to ramp up slower and get onboarded. Mostly about trialing more folks and allowing them to succeed. A good example of just solving our own problem and then hopefully solving other people's problems and generalizing to a solution. Trials are becoming more popular. Paid trials are great. We want to support that. And then we also added project-based rates. Now there's two types of contracts you can have on Flexile, hourly, and now we support project-based and full-time. If you want an annual salary, you could also do that, but it's under a rollout. If you want that feature, email support@flexile.com. We'll turn that on. For you. And we ship project-based rates. This is by Raul, who's in the chat, but he also recorded a video demo.

I'm an administrator on Plumroad, and I can go into roles now, create a new role that's project-based. Let's say I want to hire a videographer and the rate is $5,000. Now, as a job applicant, I can go to this URL. And apply. We received your application. As a company administrator, I can see I have one applicant. I can invite the contractor whenever I finish my project. I can create a new invoice which has the project-based rate automatically filled in. I can send this for approval directly to the company.

We now have project-based rates, the number one feature request. we're Anyone waiting for that, we now support it. And then the second most asked for feature was discovery. Similar to Gumroad, discovery is something everyone wants. We call our discovery feature on Flexile the talent pool. People who work for any company on Flexile will be able to basically list themselves publicly, uh, type in a little bio, how many hours they have available per week and their rate. Any admin on Flexile can see people who've listed themselves publicly. Hopefully over time Flexile becomes like Upwork for different kinds of work, mostly around software engineering, product design, customer support, things that are related to what we solve. The hardest thing about these talent marketplaces is like, how do you vet the talent? You have thousands of people, who's good? Everyone struggles with this. The secret sauce to Flexile is you'll see who they work for. So it should be possible to say Connor Kunderman. I shipped commissions for Flexong as an extra 5 hours a week. In terms of revenue, we'll take a 10% recruiting fee for the first year. And I think that will generate a lot of revenue in a very value-aligned way. You click on someone, read about them, and send them a message. Pretty simple. This, for example, is Hasib. I'm going to email him actually after this board meeting and see if he'd like to pay trial with us. Look out for that. We're moving to a world in which we're embracing Tailwind, and that will allow our designers to basically help improve our product offering a lot faster without— basically right now they're bottlenecked by engineering on certain things. We also ship company updates. If you're listening to this live, you probably got an email from me. And I want Flexile to be the number one tool, the only tool really that a CEO would need in order to manage their entire cap table. All investor relations, which includes sending them updates, board meeting videos, potentially supporting Google Meet, Google Calendar invites. I'm excited to have polls. This is a feature a few people have asked about. If you have all your cap table codified into this software, could you do a poll? Like, for example, how much should I get paid as CEO next year? And just do a poll and see what people say. And right now they would be like aesthetic polls. You're just getting feedback, but why not real ones, right? Board meeting decisions, board consents, decisions have to be made in a legally applicable way. FlexApp will be able to do more of this stuff. We have shipped expense cards. So currently you can expense via Flexile. We have a way to add an expense to an invoice. But the issue with that is let's say you book a flight, costs $1,000. You're basically like the company gets like a 0% interest loan on your $1,000, right? Because you're covering that cost for the few weeks before your invoice is due. So expense cards basically allows you to get an expense card, add it to your iPhone or Google Pay and use it for expenses in real time. They're controlled on the admin side per role. Engineers get $100 a month to spend on software. The software will manage those constraints for you. So that's Flexile. All the stuff we got done for Flexile, product 2, project-based rates, work trials, talent pool, company updates, cards for expenses. Our second product is Helper. It's a customer support tool that is like basically built in 2023, 2024, like just assumes that AI will get really good. And just like what anti-work, like what are the unfun parts of doing support? Let's say there's 5 tickets that look exactly the same. Could you get rid of those repetitive actions and focus on the fun creative stuff instead? We use Helper ourselves. We deprecated our old solution we were using, Help Scout, this week. So we really have to use it now. It's going to get really good really fast. You can start new conversations, send attachments, search, unsend emails, visit something you just sent. Assign someone over the coming quarters. It's only going to get better and better as AI is going to just, one, help us ship faster, which is nice. Actually, the unsent emails feature, the visit link, about 80 or 90 lines of code that Cursor and Claude wrote all of it using Cursor's Composer feature. Not a single line of code for that feature was written by a human. AI will basically get smarter and that will allow the models to be able to do more and more actions. Certain features that we can't really build because they assume the model It would be at a level it's currently not at yet. We'll get there in the next year or so. For example, if someone asks for a refund and that refund is asked for in the duration that matches the refund policy of the product they purchased, we could potentially issue a refund on behalf of the creator. Like that entire action could be automated. There's lots of cool stuff to save time and then offer it to our creators as well. Creators have to do a lot of customer support. If this looks interesting, go to helper.ai. Give it a shot. We're charging for it by the end of the month. It'll be $100 a month and that gets you 1,000 automated replies a month. So about 10 cents a reply. In the customer support industry, people say it's about $4 to $20 a support ticket if you calculate all in cost. In theory, there's a lot of cost savings there and it'll free up the support team to spend more time on writing help docs. No indie hacker enjoys doing a bunch of customer support, but if we can get it to a point where it's able to do a lot of it for them and they have a simple iPhone app to approve and deny stuff, it's going to be pretty cool. We started with email, built on Gmail. You connect support@gumroad.com, support@flexcel.com as a Gmail account, and then basically wraps around your Gmail inbox. But we're also going to be adding in-app chat soon. And then basically say, hey, I don't like this tipping thing. It'll say, go to checkout settings. The tipping section to turn off the toggle. And then maybe there's even a button there that's I'll do this for you. Just click here and it'll take you there. Our goal was getting response times down to 15 minutes. With in-app chat, that should go from 15 minutes to zero. Ideally, it's almost instant and should help us help you more. Our fourth product, our latest and greatest is Iffy. Iffy.com. Iffy is a content moderation API. One line of code. We want to make it super easy for any person who runs a website and accepts signups to moderate and make sure everyone is using it in the right way. Imagine you have a bar, you want to make sure people who enter the bar are not going to abuse your bar. That's what Iffy does for Gumroad and soon other products too. There's another reason it's fun to work on new projects, not just because I have anxiety, but also because it allows us to experiment with new tech stacks. For example, Iffy, it's our first basically TypeScript-only app. Gumroad's 13 years old, Flexile's 2 years old, Helper is 1 year old. All of them were built by someone familiar with Python or Rails, Ruby, and that was our approach. Iffy is our first project where we took like a frontend-first approach to it. So each product, as they are newer, right, has a better tech stack, allows us to ship a higher quality product. But also sometimes new tech is not good. It's like buggy or broken or slow. It allows us to experiment and make sure these things are legit. They actually do allow us to ship a higher quality product faster. We can apply that to Helper, Flexile, and maybe Gumroad over time. For example, we first started using Tailwind with Helper. Tailwind scares people because it looks weird. There's a lot of concerns around like design system consistency. And we've invested a lot in Flexile and especially in Gumroad. And it's like tossing it to move to Tailwind, even though we just add Tailwind to the process. It was a bit scary. Testing it out with Helper was a good way to try it out. Even after working on Helper using Tailwind, I was still a bit on edge. It's a lot of type. I don't like typing. I like talking. But as of June 20th, Claude 3.5 Sonic came out and I went from being like on the fence of like Tailwind or not Tailwind to like completely on the Tailwind side pretty soon after that. The biggest downside to Tailwind is writing Tailwind if you're not having to do that because Cursor AI is writing that stuff out for you. It's just so much better. We will be switching to Tailwind for Flexile. It probably doesn't make sense for Gumroad, but we'll see. And it should be hopefully pretty notable and obvious 3 months from now. You should be using Flexile and it should feel like it was built by designers. Right now, designers design, then engineers build. Ideally, designers can build it, engineers can QA and fix it. Uh, I would like to move to a world in which instead of designing before building, we actually build before we design. At the end of the day, we have 4 designers and 15 engineers. So if you can unblock the engineers from shipping because they're blocked by design and design's blocked by engineering, it's gonna be a pretty awesome world with a lot more stuff getting shipped. It's noticeable in our August update. We're shipping a lot more. Divi is a completely frontend stack and it's awesome. It's really nice because it allows us to get rid of the translation layer between the backend and the frontend, similar to how Tailwind gets rid of the CSS design system translation layer between Figma and GitHub, there's all this like crazy design system in the CSS. That's 7,000 lines of code. Think about all that translation layer has to be understood by both sides. It's costly. Similarly, you have Python in the backend, then you have JavaScript, TypeScript in the frontend. If you have everything in a single language, you only have to define those things once. If it works well, we're going to do the same thing with Flexile and move everything to a Node backend. If you didn't apply to Gumroad because you dislike building web apps, which is you make one small change to a CSS file and somehow a page all the way over there that you haven't thought about in 3 years breaks. Or if you're like, I really wanna use Tailwind, I'm only gonna build stuff, design and build stuff in Tailwind going forward. Now you can at Gumroad. Hopefully you'll consider applying. Normally we showcase the roadmaps we have for each product. Things are moving so fast that I think we'll be done with all of the stuff in our roadmaps by the end of the year, which is the goal, right? Imagine we normally plan out 18 months. If all this stuff is happening that allows us to ship 10 times faster, we could be done with everything by the end of the year. We're not going to be done. Like the road continues. It's just that we have to like, we're like laying the pavement as we go, which personally I think is really exciting. It basically means that the team is going to be talking to customers all the time and solving their problems in real time. And yeah, but what that means is a lot more creativity. The theme of AI and automation is the job has to become more creative. The engineer may have to say, what should I ship instead of I'll just ship what the designer built, which changes the job. But my hope is that it becomes more creative, more fun, and more stuff getting shipped hopefully makes the business stronger and means we can pay people more with AI and automation. People get paid more per minute because they can think more. Let me go through the monthly updates fast. Optimistic definition of power. Is iffy.com owned by Gumroad? Yes. This is partly why I wanted to change. A lot of people were like, hey, I invested in Gumroad. Flexile looks dope. Am I an investor in Flexile or not? And so that will help. You invested in Gumroad. Now all of a sudden you also own some Flexile shares and some Helper shares and some Iffy shares. And maybe it makes sense to spin one of those out. We'll do that. But Antiwork will be like the parent. Company. Actually, I like this because it's an A shape. It's on Figma. I'll share my screen and show you right here.

It's in Figma.

We may eventually publish our components. I don't think they'll be super useful because we don't use Tailwind and most people want drag and drop. You can see our stuff here. It's still early. We still have to add a lot to it. You can see our buttons, modals, pills. There's a cool feature called Figma Code Connect. Soon it'll allow GitHub and Figma to be almost entirely in sync. If you're an engineer familiar with type checking, You'll be able to get type checking on your Figma files. Imagine this didn't exist in Figma. You, the engineer or the designer is building it and they create the variant small in the component, in the button component, or in the pill component. And then they try to publish it. It wouldn't work. It would basically like a test failure saying, hey, you added this like small variant, but there's no small variant in Figma. And that's really cool because right now everyone deals with this issue, which is it's really nice to have a design system in Figma. It's really nice to have a design system. In code, but technically you should only have one design system. You have your Figma design system, your GitHub design system. Figma Code Connect came out in April, so it's still early, but it's gonna be pretty slick. Just to go quickly through some of the monthly updates that I send the team, you can see all the stuff that we've worked on. I wanted to show you how much more we're starting to ship in April. A few things. This is a small feature. The website builder's big. You did one transaction per cart, which is cool. In July, we started shipping a lot more, and in August, you can see a lot more stuff getting shipped. That's because I can mock the feature up really quickly instead of having to type it up in Notion, and then the designer has to look at it, and then the engineer has to look at it. I can just go in the code, use Cursor, flesh it out, and something that would have taken weeks to ship can now be shipped in like 2 days. Give people a visual, because that would be helpful. I've always wanted to do this, so we have a whiteboard. If people have ideas of what we want to build, you can ignore some random stuff. A couple things I want to talk about quickly. Jevons paradox. This is useful to know. A lot of people are scared about AI because they're like, well, if it takes my job. But basically the thing about AI is like adding more lanes to highways in LA. There's a lot of traffic. So what do you do is you add lanes so more people can drive, reduce the traffic problem. And that's true if like you just ended things right there. But what ends up happening is people learn that there's more lanes in LA. There's less traffic in LA. And so what happens? More people move to LA. Solving the problem increases the demand. This is called— so as long as market's induced demand, right? And so this is what we're trying to do with Gumroad, make it better and induce demand. The demand will be induced only if you actually lead to a better UX. So that's called Jevons paradox. People can read about that. And then this is just a chart. So there's this thing I think Coletta Perez came up with this idea of the innovation cycle. So you have the new world and the old world. Imagine this is like innovation scale. The old world would be like paper books. There's this interesting thing that happens when you go to the new world, because not everyone is there yet. In this phase, you have this ebook, a digital paper book, but it's not really embracing the new world yet. And so only once you get to basically 90 to 100% penetration, Do you end up with, for example, a website? If you go to Gumroad and you buy a product, you're basically getting access to a website with pages on it, not an ebook. An ebook is old thing that you need because people are used to this thing, but it's really a transitory form. The way I like to think about things is there's this 3-step process, which is like old, then new. To apply that to the way we ship, the old is here, right? So the old is Designer comes up with design system, then the— some engineer implements it in code, and then the designer uses design system. There's a lot of back and forth here, right? That's kind of what happens. And then there's a bit of back and forth. And then they design feature. Once people are happy with that, the engineer ships it. The happy moment is when the product gets better. That's what customers are excited about. And this is designer, engineer, and feature system, right? The system is designed, built. Then features design and build. You are designing the system, going from design to building, but in a tailwind. In Tailwind land, the designer can do it. The designer designs the feature, then the engineer builds the feature, right? There's, in this world, in Tailwind land, there's no sort of design system, or the design system at least comes after the fact. Think about how much faster this is, how long this line is. To ship anything before the product gets better. You can't ship anything before you update the design system, the CSS files. It's massively slow to ship a new feature. All of a sudden here, it's this two-step thing. There's still more work involved if you still have the design system you're maintaining. Because of Tailwind, you can ship the feature much faster. You don't have to think about the system until it's actually live and being used. So that is way more efficient and that's cool. But it's still confusing. Even I was like having a hard time wrapping my head around it. In this world, the roles are different, right? This is a completely new world. In this world, we no longer have designers, we no longer have engineers. That's gone. In this world, we have design engineers and systems engineers. In this world, you have full Tailwind and AI, right? Fully AI. This is like AI awesome. You can't even, you can't even spell Tailwind with it. That's how cool it is. And in this world, the design engineer can actually ship, right? They can ship the whole thing, just like I was able to ship that visit feature on Helper. So all of a sudden there's no back and forth. This can be shipped by the design engineer. The designer can now use natural language AI to write code instead of having to learn JavaScript. I don't know Tailwind. I've never looked at the docs. I can just say, hey, Cursor, make the margin double it. And look at the design and improve it visually, which is the goal. They don't look at the code and are like, oh, the code is so spaghetti-like. It's, ooh, the button's pretty, click. And the system can be updated. The designer should be doing all the work to design. It's actually more like design-build-design. This is more like build. And then the systems engineer can reincorporate changes into the codebase. You want two roles. The design engineer is the one who can design and build features and just ship mega fast. This is almost like a PM, right? PM who can design and code. Then you have badasses, like at our company, we have people like Maya and Chirong, especially Maya. She's like amazing at thinking about systems. She probably doesn't want to think about all these little featured little things. And so you can have people who are more focused on the systemic application. What parts can we reuse? Basically, these people may be spending more time on Lego blocky things, and these people are actually using them to build skyscrapers. But it's different. It's going to take time to adjust. This has a compound role where designers think mostly about visual and logic. That's still true, but like CSS, there's a line called separation of concerns. The button should just be a button. You shouldn't have to think about what it looks like. It's nice, actually. At one point. Old world, right? CSS is an ebook thing, right? When you're making ebooks, you want CSS, you want all the paragraphs should look like this, or all the buttons should look like this. You can still do that in a Tailwind land. It's a sort of a more of an engineering slash design problem, which is what are— you want a button, but then the button should have variants, right? And things like that. But maybe you don't want a variant. Maybe there's one button somewhere in the website that is just a crazy button. We have this crazy button at Gumroad, which has like multiple drop shadows on it. And I was even thinking it would be cool if you like hovered over it and based on where you hovered, this is all the shadow, move crazily. But you wouldn't want that really in the design system cuz it's like crazy. Maybe in one place it makes sense to do that. You build the thing and then if it ever gets made twice, then the systems person is, oh yes, we can like refactor it and put it in two places. So basically we're gonna change things up a lot. The thing that allows us to do that is admitting in every hero's journey, there's a truth you learn. The truth is that at the end of the day, there's like some tie between the aesthetics and the logic. The button component should have some level of visual stuff built into it. And then AI really supercharges this because if you violate this principle and you say the button is aware of its visual state to some degree, all of a sudden you don't have to provide all this extra context and then the AI will actually be able to do both of these If it already sees the logic in the code, if all of a sudden it can also see the visual in the code, you can have the AI become the design engineer. So basically, maybe another way of thinking about it is like all of this stuff is now here and there's these 3 new roles that now almost exist. But now you have new roles that can exist in the organization because this person is moving so fast that you need more people helping, right? You might only have 1 architect, but thousands of construction companies. So that's my rant. And then I wanted to actually ask what we should build next. What are you laughing about? This is my brain on drugs. Just kidding. What should we build? No one wants to contribute. That's fine. We'll build whatever I want. What should we build? I need help. The exciting thing about having no roadmaps is you can think bigger, right? Sometimes you're limited by what's next, but I think we can think bigger. All right. Flux. And these things can talk to each other too. I'll add some email marketing improvements. It's annoying right now. I'm doing this, by the way, with the board being stopped sometimes with my cursor course that I'm doing on Gumroad, Coding with Cursor. Oh, I want to send everyone an email, but what if I accidentally set one of these toggles? I would much rather have AI here. I can say email everyone who's bought my product in the last 30 days or 7 days, and this will turn into filters. This is going to be awesome. This is going to be way better. Email marketing is going to be core to Gumroad in the next 6 months or so, we're gonna make it better. Focus on sales. The best way to make sales is to email your audience. So we're gonna make it easier to do this. Maybe there's even automation here. We could suggest emails. We could even give you like the placeholder copy and you could edit it, things like that. It'll be pretty, pretty cool. Merchant of Record reseller, Jan 1st, 2025. Another big one. This is very exciting because it means we're gonna absorb all of the tax liability. And you're never gonna have to think about any of that stuff ever again. Upsells and the content editor, I'm excited about. So a lot of people, I think 8 out of 10 purchases on Gumroad are free products. That's great because you get their email, you build a relationship, but it's hard to do upsells. You have to email them. But what if we were showing you the cool calls feature? Imagine you sign up for a course and at the end there's a button, call me. It's like stuck, help, and all of a sudden you have a whole new way of monetization. You can focus on building your email list, building your free product offerings, and then put these call things in here, and you're going to start making money. So I'm very excited about this. Flexile. Do we plan to allow more customization so that we don't have to build a separate website? Yeah, totally. So yeah, basically he was asking about more customization. Definitely. Both the website and the emails. Emails are also like very boring right now. We should definitely have more customization. Gumroad was built in SQL land and it's just annoying because every time you want to do something, you kind of have to like add columns to a database, which is crazy if you're just like, hey, I want to customize the email template a little bit. Doesn't really need to know about MySQL. This is going to require probably doing a lot more stuff in like Mongo or something like that. Another lesson learned from 13 years. Mongo used to go down all the time, but now It's great. Mongo's pretty good. It's the most valuable tech company in New York City. Flexile. I have a controversial feature, which I'm very excited about, which is a time tracking app. The way Gumroad works currently is everyone submits an invoice at the end of the month. I think there's a bunch of changes we can make to make it better. One is people should be able to submit as many invoices as they want. If you want to get paid every day, maybe you should be able to submit an invoice every day. An annoying thing about it is it's annoying to track your time. I don't want to actually see what anyone's doing. I don't want access to your screens, but I do think AI can help with this. So you don't have to track your own time. You just download this app and all it does is basically take a screenshot every 5 seconds or 10 minutes or whatever. And then it helps you track your time and you have a button that fills out your invoice for you. This is all the admins would see. I have like none of this data. There should be a firewall here, so I don't want admins to deal with, to see any of this, nor does any worker want to see that. But I think this will help a lot. The other cool thing is everyone gets a single hourly rate. So you're a product designer at Gumroad, you get paid this per hour. You're an engineer at Gumroad, you get paid this per hour. You're a customer support person, you get paid this an hour. You know, say $100, $150, $200. It's not super efficient because what if actually, and basically these are basically based on supply demand of the market. Right? It's not actually like supply demand for what the company needs. It's like there's very few lawyers and so lawyers are really expensive, right? In theory, right? Otherwise you would have a cheaper lawyer come in and offer the same service for less and you'd use them instead. So generally supply demand. But once you work at the company, I think it's important for us to say the most valuable thing is to do design or to do support or to do engineering. And so I would love for some sort of way to say, actually, maybe there's a way to move these roles, but these are almost more like tasks that you do. In theory, let's say there's like migrating Flexile to Tailwind, super valuable, would really help us move faster. Maybe there should be a role just migrating to Tailwind and it's $300 an hour, right? We're not gonna create a job like that, but what if there was a way to create this in Flexile and then if you're using this app, it just automates this for you and it says, hey, you mostly got paid $200 an hour or $300 an hour, you spent 4 hours migrating pages to Tailwind and got paid $300 an hour for that, right? I think there's cool stuff we could potentially unlock that Flexile is uniquely positioned to because of the way that we work with the hourly contractors and things like that. This would probably be an incremental approach. I think there's something cool.

I have a lot of ideas about things that could be part of this because I feel like one big thing that we should focus on is several areas where it is difficult difficult to build new features due to decisions made 10 years ago. So yeah, the refactor, the purchases table. Yeah. Could help us too, the splitting views and collapse so much easier, those sorts of things. I think that one and also refactoring the, for email, refactoring the posts logic and data structure could like really help with that. 'Cause that can be a little bit of a nightmare to work on.

This tends to happen at companies where only one person touches. Basically what happens is there's so much complexity and then you're like, okay, Vishal is the only one who understands this, so it's on him to do everything related to emails. That's not great because basically it goes from like 15 engineers could work on it.

We tend to look at the front end and build off the existing backend because doing data migrations is so tricky. So I think like I almost wanna have a once a year, like state of the union where we're like, what needs to change in our, what are we gonna make work this year in our data architecture?

Yeah. I mean like this Mongo thing, right? When you have to do those mega SQL migrations, we made the decision to use MySQL in 2011, basically.

I think there's a big role for SQL. User preferences or something. Maybe those don't need to live. Like we do flags for those things, which are easier to work with.

We've used Bitflags as a hack so we didn't have to do migrations, but that's a hack that maybe would be better solved by a big user preference dictionary in Mongo. You don't really need to do sums. On people's preferences and analytics and all this stuff is also expensive. We spend thousands of dollars a month because we have this assumption that to run Gumroad, you need a MySQL database that is huge, hundreds of millions of rows, billions of rows. So yeah, I'm, I agree. AI makes refactoring much faster. Factoring is that it just takes a long time. And it's like very boring, menial work. Often you're rewriting a bunch of stuff in a different language or whatnot. There's a lot of creative work in starting the migration and finishing the migration. But I think of this like flying a plane. The human takes off and lands. The AI can do a lot. You've done one page, you showed me how to do it, or you migrated something. It makes refactoring faster. But then two, there's more value in refactoring as well, because AI struggles with like really old stuff. Most AI data, the business use case is going to be like, are we going to train the AI on MySQL, which not that many people still use today, or are we going to train it on like the stuff that everyone is using right now because we want customers for our AI? Doubling down what's working and more modern technology and reducing technical debt becomes more valuable than it used to be, which is exciting. It's more valuable and faster to do. AI influences that. It means working on our products gets better. One of the worst things about working generally is you have to deal with debt, right? Like when you work on your own fun personal website, you're like, I'll just use the release candidate and who cares? It's fun. So hopefully Gumroad can get there where it's almost, almost there in terms of should not feel like you're working on a 13-year-old thing. Competitors, that's the scariest thing. At least a new competitor to launch, they have no tech debt and they can just ship super fast. Innovation cycle will continue for us. It'd be cool to have an mobile app. I also think there should be like a reply function of some sort. If we don't reply, if a human doesn't reply in 24 for hours, it automatically replies with AI and says, hey, human is still looking at it, but we wanna get you a reply. There's a lot of room to make Helper better.

Very excited. Team updates.

Okay.

Put a thing in Slack, what we did and what we're doing and task.

It's like a task of a task. 100%. Yeah, exactly. I think ideally Flexile and team updates and even Helper, like, It should just autofill with your GitHub. Everything you do is somewhere, you know, it's a support ticket you answered, it's PR you opened, merged, reviewed, or Figma mock. It is useful to make sure everyone is productive. If you pay people nice hourly rates, it's useful to know what they're doing. It's useful for people to make sure that at the beginning of the week, they've really thought about what's the right way to spend their time. We can automate almost 100% of it. Maybe these things tie together where you have some form of not performance management, but some form of being able to be like, oh, we're underpaying for this role. Actually, there's not enough of this happening. Not enough support tickets are getting answered. There's a lot of cool things that even Helper, I think, could probably sink into Flexile. I'm pretty excited about this. Oh, one idea, if Helper could become our feature requests place, because we don't really necessarily need a road— like, why does the roadmap exist? Generally, it's because we got an email from someone. And they're like, hey, we want this feature. And then I'm like, awesome, we'll build it. And then 3 weeks later I reply being like, it got built. But AI could make that a lot faster. What if there's like an answer by opening GitHub PR? Our roadmap is basically informed by creators. So, you know, there should just be a thing that runs on every ticket and you can just see all the requests and vote on them. How cool would it be if you could open a GitHub PR that syncs to this and then when it's merged All of these get replied and boom, everyone gets an email saying, hey, we shipped that feature you asked about. And it's completely automated or largely automated. This is our back office. If you own a restaurant or any real business, you have a back office that does like all that paperworky stuff. And then you have your like front of house, the restaurant, like the people who take your order and tell you why your food is not there yet. Helper is like our way to just really improve our customer experience. Flexile is a way to make us more efficient. They're both this nice yin yang, which will be awesome. But yeah, team updates is going to be, it's going to be really cool. And we'll be able to potentially get rid of that aspect of Slack. And that's also just generally a nice thing is we use Slack for a lot of stuff. We use Notion for a lot of stuff. We use Figma for a lot of stuff. We use GitHub for a lot of stuff. Each tool, you join the company, you have to understand, okay, GitHub is like this and it interacts this and you have to do this ritual to ship anything. Maybe we don't even need Slack. Maybe we don't even need Notion in the future. And then boom, it's just GitHub, Figma, FlexOut, Helper. I don't know. It's probably a reach. Probably me becoming— having a manic episode, but I think there's some cool stuff we can do. Some audience— audience. Yeah, Mark. Yeah. I emailed them and said, hey, why'd you churn? And they didn't respond. But there's a lot we have to do. Makes Flexile really compelling. There's just so many people in the marketplace who offer similar payroll-y type services like Deal and Rippling. The core of what we do is this equity per hour model, like flexible equity. Right now it costs, so our lawyer, for example, most lawyers, it's $13.95. Like you see these, this is how much we pay. People, people think this is high, right? $100 an hour. $200 an hour. This is how much we pay our lawyers. $13.95 an hour. Lawyers are basically software engineers, except their software is US law. I can't do this myself, so I pay them. This is the reason equity is so uncool, because it's just expensive. I've talked to so many people who are like, I'd love to give equity, but I'm not going to pay $10,000 to give someone $5,000 in equity. Like, the math doesn't really make sense. There's all these features we have to build in order to make equity basically free. It should be almost free to set up a stock option pool. Rafael's working on the lawyers feature. We're basically building DocuSign into Flexile. We're going to build all these legal agreements into Flexile. The cap table exercising is now there. Tender offers, buybacks, we're building into Flexile. Cap table management, equity splitting, self-serve dividends. Every CEO listening to this is familiar with this, where you join the company, you get an option grant. That option grant needs board consent. Because the board has broker complications. You have to loop in lawyers. They're paying $500 an hour. All of that should be automated soon. Gifting shares is another one. I gifted people a bunch of my shares just to learn how annoying and expensive and slow it was going to be. And it was annoying and expensive and slow. We're going to build that into Flexcel. If you have some shares, you want to give them to your sister in America, for example, you don't have to pay any taxes on that. A gift under $10,000 a year is completely tax-free. So let's say I have millions of dollars in government equity. I can give $1,000 to 1,000 people and no one has to pay any taxes on them. And then they get dividends with that, which is awesome. Because you're not giving someone $1,000 in cash, you're giving them $60 a year for life, like access to the dividends. I think that's super cool, a novel idea. But gifting stock is a pain. You have to get permission from Excel, email them, all this crap. It should just be automated. It should be like, Excel is cool with me doing this. And then there's one agreement, they sign it, the software knows if we nailed that, then there's like in our internal thing, it's we want to do all of that before we really hire a sales team. I really think Figma, for example, Dylan, if you're listening, I'm going to pitch you on Flexcel. I can't imagine how many millions I spent on lawyers and ops for anything related to their cap table. None of it's going to autofill because it's some weird iframe Morgan Stanley built in the '90s. We're going to build all this and try to get Figma, Stripe, SpaceX, companies that are private but almost public, flexible equity, dividends, and liquidity. So NASDAQ for private companies. Oh, if you use these services, like you should just get invited to Flexile onboarding and then everything should be automated, right? You start your trial, you get access to the proper tools, you succeed, you get access to the rest. All that should be automated in the vision of Work, right? I just want to hire people. Hey, work. And then they can start shipping. We don't have a logo yet. So if anyone has logo ideas, suggestions, let me know. I think this is time off calendar. I'm a little worried. I like the fact that we live in a metaverse with these products. I like Helen's idea of integration with GitHub. So basically time off calendar and you just put the time off in Flexile and then there's like a sync because often you'll mention someone in GitHub and they're gone for 3 days and you don't know. It would be cool to automate. All of that. Jono says Flexcel Equity feels geared towards people who know equity. Would be great to have an equity for dummies experience with simple explanations, interactive examples, estimated returns over X years. I agree, Flexcel should have this built in. When you apply, you see your hourly rate, your equity, 0, 80%, whatever. If you pick zero, it should show you make the same amount every year. If you go up You make less the first year, but then you get dividends the second year. AI, I think, will help a lot with this too, because you can even have, like, with Josh, we're building this cool thing on the Coding with Cursor section on livestream. Basically, how familiar are you with equity? You could be like beginner or a dummy. I'm an equity securities lawyer. This could change based on your comfort. No one understands equity. I spent hundreds of hours learning to build this out. Every time, like tender offers, I have to learn about how they work. Could you do a tender offer where someone can have vested options and participate in the tender offer with their vested options? And the way that works is the options turn into shares, but for only a split second after they're sold. And share certificate avoided complications. I think Flexcell, just like Gumroad, basically helps people wrap their head around selling a PDF on the internet or making money on the internet. I think Flexcell has the opportunity to make equity accessible, or you want to be really cute, equity, equitable, which is basically equitable just means accessible, understandable. Everyone can have access to this thing. It doesn't mean equal necessarily, but equitable ownership in a thing is equity. For example, taxation without representation is not equitable, and I would argue that equity is not equitable. It's ironic, but it's not because it requires $13.95 an hour. Like, why do VC-funded startups give people equity? It's not just because VC-funded startups are better than non-VC-funded startups. It's just that the VC-funded startups raised money so they can afford to hire a lawyer. If you're a bootstrap business, will you have enough money to set up your equity? The problem is if you wait until here, the business is now worth $10 million instead of zero. So now to give someone equity, they have to have a higher exercise price and all this crap. And most of the time they're like, that's not worth it. You don't protect that by not having an option pool set up from the beginning. We should make that easy. That's a lot of my focus, making this self-serve. First I go and then scaling it to the rest. Ryan says in-app analytics for helper analytics makes sense to me. Analytics that could even sync with Flexile. You know, you should not have to say I answered 300 tickets this week. It should automatically be there. I wanna incentivize everyone to answer support tickets. Everyone should be answering support tickets, doing design, doing engineering. Soon everyone will be doing customer support and everyone will be a super person. What will the engineer do if the designer is building? I think they'll be designing. There's always engineering work to do unless you run out of ideas, then there's nothing to do. But for example, let's say we're never building a single feature ever again. There's still dozens of hours a week in just embracing new technology, like taking AI and including it into the software, there's always gonna be that 0 to 1 that engineers are gonna have to do. That's never gonna go away, right? Let's say AI does everything, every single thing AI can now do. Well, who's gonna set it up? Someone still has to connect the AI to GitHub, right? There's always gonna be engineering work. So feature goes from 0 to 1, then 1 to N. Right now, designers go from 0 to 1, then the engineers take the design system and build. Every feature is slightly different. You don't want the same layout on every page. Page doesn't make a lot of sense. There should be some components. Engineers, you probably have anxiety because you're like, oh, I do all of this stuff. And if designers are doing this, what am I going to do? And basically you get to do this stuff now. You get to do Lego block building, upgrade Python to Node, upgrade AI models, work with prompt engineering, make it better, faster, cheaper, simpler. Feature engineering will go away. The idea that you're bottlenecked by an engineer who knows some Ruby to ship it to production. Will probably go away. You can only ship features for something that has some context, some framework for how to ship. And so I think designer engineers will still be doing that. Similarly, like designers will still be working on the design system, but they may be spending less time actually doing Figma mocks of new features because they can go straight to code, or engineers will be able to just use Tailwind and ship their own ideas and get feedback from designers. Re-engineering is a new kind of thing and redesigning, right? Because the PMs of the company, like the people who are generalists, will be shipping. And then the really good designers are going to be the ones who actually redesign instead of design. Things should get shipped and then redesigned instead of designed and built and then shipped. Similarly, I think engineering should be okay with things shipping and then the engineer is responsible for refactoring and making things better and cleaning things up once you have data from the usage of the feature. Helen asks, asks, how do you think about what features should be built in Flexile versus using another service that focuses on that feature? But a million folks, like, for example, equity is pretty unique, but lots of folks do time tracking. Why not integrate with Harvest's API? Or are there things we should not build into Flexile? Uh, that's a good question. Honestly, I think that the cost is going down so much at building new features that the equation changes where there'll be more stuff we should do than less. But we should probably have integrations. If people are using Harvest, there should be an integration. Our team is so good, we can almost outship them.

Yeah, because it's something people use all the time for other things. Google Calendar, people use Google Calendar all the time.

It's also a cost thing where these services are free because Google makes so much money on other stuff. You'll never compete with them. I think we should integrate with FAANG or like large companies that have crazy distribution of free products like Google. Calendar, Google Meet. But the, I think the lack of a mega company that does time tracking goes to that induced demand thing where time tracking is still too annoying. Maybe we buy Harvest. Harvest is a phenomenal business, but I still think maybe it could be unlocked. There's like more stuff they can do. But yeah, I could see us potentially integrating with them as well. But I think that the payments thing is tricky because of the equity piece. We have to do the payment or lock. I could see an integration with Harvest or Toggle. Over some of the other ones. For example, we probably won't build the team updates thing. We'll use GitHub and Notion. Some products haven't shipped anything in a long time. I'll take some Q&A and then we'll wrap up. What's driving the COGS so high? Contractors are expensive. Just employee is just like a legal term. We still have 30 people. No full-time employees was a good blog post title. It's like saying no yogurt in the fridge. You have milk. Any plans to support payments like USDC? On base? We don't currently have plans, but it seems inevitable. We would love to do that because of some NSFW things. Stripe kicked off a bunch of people. We lost some volume that way. It would be cool to use crypto to support them, but we do have to make sure it's like we're supporting legitimate behavior. We can't just— that's always, always a risk with crypto. But I think they also— user experience credit cards is so good. Can you discuss metrics with a longer time horizon? Projections are pointless in my opinion. We just want to ship. We don't have deadlines, we just ship ASAP and we like shipping as long as we're profitable. Will you host in-person events again? Yes. We used to do these dinners. I want to do another Gumroad creator dinner in New York, so we'll do that. If there's no roadmap, what should we build for the job application? At the end of the day, people are gonna be responsible for coming up with their own ideas. Go to Nolts, go Twitter, see what people are asking for. Just doing what we do and speculating. Iffy, someone says Iffy will not be earning money. Iffy is actually making money. We made about $120 with Iffy, which is awesome. So Iffy is a real business, has made more money than Helper actually. What should we build to showcase? You can, Gumroad doesn't use Tailwind, but if you want to, you can use your own tech stack. It's more about the UX. If you're applying as a product designer, not to the technical quality, engineers can help with a lot. Trending products we should build, Igon feature. Someone wants more info on creators and discovery. I like this idea. We used to have it, so maybe we'll add it again, but some sort of look inside feature I think could be really powerful. But I think it would be really cool, Ken asks, for seeing someone's bio, video, maybe even chapter 1 of their book. For example, I'm teaching this course on Cursor with 4 sessions. Maybe the first one is available for free and then the other 3 are more expensive. Suhas says, how do you figure out which shipping is best when you get entries from multiple people. This is like what's in my brain that makes me employable. I've seen everything we've shipped since the beginning and I have a good kind of intuition around quality or like what will lead to the most money being earned by creators. But if you have same thing over here, but there's a bunch of friction here, right? And this will be slow. A designer's job is to build this, design and build that, right? And make sure that as this ball, the potential energy becomes kinetic. Let's say this person wants to make a dollar. Here, they only made 98 cents. But a really bad design, they make zero. So it's, that's how I think about it. And what do you call this? Work, right? Work is what you call friction, heat, work. But for anti-work, we're anti-that stuff. Right? We wanna have a super smooth ride on the roller coaster and maintain the potential energy as much as we possibly can. And if you do that, you have less heat, less entropy is generated. Maybe we can even stop the heat death of the universe with our B2B SaaS. Too late to explore the world, too early to explore the universe, but just in time to build B2B SaaS. I will see you in 3 months from Bali. The whole team's gonna be in Bali. November 15th, 2024. See you then.