Making AI an Opportunity With Joshua Drake  🔗 **Source:** [Making AI an Opportunity With Joshua Drake](https://share.snipd.com/episode/be6f11e6-5a62-4598-a656-43fed8d9a881) ## Highlights - **Why He Refused Venture Capital** - Joshua Drake kept Command Prompt independent for 30 years and refused VC, PE, and debt to protect team welfare. - He founded the company in 1999 after seeing founders sacrifice values and mistreat teams under investor pressure. Transcript: Joshua Drake Sure. Command Prompt's actually the last independent standing, or I shouldn't say the last, I should say the oldest independent standing Postgres company. We've been doing it since 97. We have no venture capital, no private equity. And we're dedicated to Postgres and specifically open source around Postgres. Unlike my great friends at Procona, where they're more, you know, abstracted where they do Postgres and MariaDB and all these other things. We're more boutique. Emily O'Meara Was just going to ask you about Percona. So why, why stay independent? Why not take venture funding? Joshua Drake Ooh, that's actually a great question. So I came up as an adult, a baby adult, so my twenties, uh, in the nineties. Uh, I was at the forefront of internet services providers. I actually launched, I don't know, three or four private internet service providers. And what would happen is I would launch one and then someone would come along and be like, oh, you know how to do this. I'll give you a job and a little bit more money. And I did that a few times in a row. And I actually saw what venture capital and private equity can do to people. And it really, I mean, not all people, obviously there are exceptions, but when you take somebody else's money for the purpose of building your equity, you are sacrificing your ideals. Right? You no longer get to say this is wrong. No, you have to say, I do not agree, but we will move forward anyway. So after getting fired from my last job in 99, command prompt was part-time for the first couple of years. In 99, I launched Command Prompt and I decided that there was a few basic rules. We would never do debt. That doesn't mean we don't have a credit line. I mean, I'm always going to make sure people get paid, but we've never had to use it. We don't do venture capital or private equity. And the owners always get paid first. And that's not to say that our team is not valuable. It's not that. It's from personal intimate experience. I worked for a company who was struggling because they were trying to get product growth off the ground. And the founders decided to not take payroll in order to make payroll. And I watched very reasonable, good human beings become terrible people because they were under so much pressure and stress and they weren't able to rechannel that into proactivity Or positivity within the team. And then all of a sudden, you know, you had things like, I didn't take a paycheck. Why can't you stay an extra half an hour? And I was never going to be in that situation. ([Time 0:00:54](https://share.snipd.com/snip/b636b9da-2387-49a3-a9a4-6b4a31ce2d78)) - **Prioritize Stable Payroll Over Investor Equity Gambles** - Pay owners first and avoid abusive payroll setups that force founders or staff into survival behavior. - Joshua enforces no debt, no VC, and guarantees owners get paid to prevent team mistreatment. Transcript: Joshua Drake We would never do debt. That doesn't mean we don't have a credit line. I mean, I'm always going to make sure people get paid, but we've never had to use it. We don't do venture capital or private equity. And the owners always get paid first. And that's not to say that our team is not valuable. It's not that. It's from personal intimate experience. I worked for a company who was struggling because they were trying to get product growth off the ground. And the founders decided to not take payroll in order to make payroll. And I watched very reasonable, good human beings become terrible people because they were under so much pressure and stress and they weren't able to rechannel that into proactivity Or positivity within the team. And then all of a sudden, you know, you had things like, I didn't take a paycheck. Why can't you stay an extra half an hour? And I was never going to be in that situation. ([Time 0:02:51](https://share.snipd.com/snip/80f8e6fb-bc8b-41f3-ba62-bd47ee7f5da9)) - **Founders Who Eat Beans Lose Control** - Accepting investment that requires founders to live on 'beans and rice' signals misaligned incentives and often produces bitterness. - Joshua says founders who accept those terms almost always walk away bitter in his experience. Transcript: Joshua Drake If you're going to take investment and the individual or firm that is looking to invest in you says you're going to eat beans and rice because you've got the equity. Okay. What they're saying is we don't want you to be successful. We want us to be successful. Okay. Any per, if you are so desperate that you are willing to be in a position where you are the lowest paid team member to make your idea come to fruition, you will be abused. Period. I have never seen, I am 53 years old, in my entire professional life, every single founder that has been willing to grunt labor for the chance walked away with nothing but bitterness. I'm sure there's exceptions, but that is what I've seen as a whole. And don't get me wrong. I'm talking specifically about taking investment. I'm not saying you bootstrapped it yourself. So you're eating ramen. Okay. That's a totally different discussion. We're talking, you said, you know, I'm going to take in $15 million. Here's my MVP. They get 60%. I get to keep 20 and they're keeping the other 20 for potential future investment or employees or stock options or whatever. And they're like, you're going to make beans and rice money. ([Time 0:06:40](https://share.snipd.com/snip/96e1977c-bede-45f1-b00d-c6799cbef287)) - **From Alarm To Acceptance About AI** - Joshua’s stance on AI evolved from alarm to pragmatic acceptance: initial IP theft concerns gave way to seeing AI as inevitable and transformative. - He expects AI to expose long-hidden bugs and raise development velocity and quality. Transcript: Joshua Drake That's a great question. Well, you know, I think it's actually changing the good. Initially, it was for the bad. Initially, it was one. We got a lot of reactionary opinions because it's, quote unquote, not human. Okay. Even though it actually is in a way which we could get into if you want. And then the second round was you got a lot of bad actors. I mean, you've seen, I can't remember the repository, but there's a Rust repository that's starting to show quite a bit of malware because of AI. And a lot of people had to shut down bug bounties because of just ridiculous pull requests generated from AI, things like that. I initially, and I was, I mean, if you look at my posts from a year ago forward, it's very much been an evolutionary attitude change toward AI. Mainly because it's a criminal enterprise. I mean, it's organized crime. These companies got together and said, we're rich and we're powerful. And we've taken investment from every major other company in the world that matters. So we're going to ignore all the laws and steal all the intellectual property. That is literally what they did. There is no arguing that I've got a class action suit that I was a part of from Anthropic. They got a settlement because of it. Okay. So that was where I started. But as I moved forward and I started realizing where we're at, we're at a critical juncture point where you have no choice, but to accept it, this is coming period. And not even coming, it's here. And i have a description of why i say that um and when you even have old old school linux the kernel linux kernel post chris ql which i thought would have rather closed the project than accept Ai um you have these companies now wanting or actually making use of it, I think it's going to basically now do one thing. One's going to expose everything that was broken, which we've already seen that. I mean, with the whole fable mythos story, they found a 27-year bug in OpenBSD of all things, which is largely considered the most secure open source operating system on the planet. But the flip side of that is that we're going to see two things. One, a much higher velocity of development and a higher quality of development because now we have tools that a peer review is now a peer review of work that was done in the background Over four hours that you can summarize, put your eyes on, run a unit test against, and clear. Versus download the patch or download the tree, apply the patch or recompile the tree, test, look for things like buffer overflows. All these things that frankly have been cryptic, but mundane. And they're no longer cryptic. They're all highly visible. So if open source embraces it correctly, and there's some things that need to happen. ([Time 0:09:01](https://share.snipd.com/snip/cb6c6245-ea36-4533-ac4b-5cdc59df6eca)) - **Require Legal Accountability For AI Contributions** - Require durable logs, notary of actions, and contributor agreements when accepting AI-generated patches. - Joshua recommends projects tie to nonprofits or legal frameworks so patches come with contractual responsibility. Transcript: Joshua Drake Durable log and notary of action must happen. Right. We have to, you know, we're going to start seeing more and more open source projects, I think, go nonprofit or associate with a nonprofit so that they have legal standing so that you can Sign contracts. So that if JD comes to you and says, I have this Claude patch, we'll say, the project will say, okay, but here's our contributor agreement. You are responsible for that patch. ([Time 0:12:27](https://share.snipd.com/snip/4c342caa-acd2-448c-aa18-73773590d9ac)) - **Open Source Is A License Not An Identity** - Open source is primarily a license, not an identity; conflating identity with a technology leads to threat-based stress. - Joshua urges technologists to identify with people and outcomes instead of a specific project or license. Transcript: Joshua Drake Well, and I'm pretty opinionated on this because I used to be a director for software in the public interest. I'm a founder of United States PostgreSQL, things like that. One of the mistakes that people make is they assign their identity to open source. Open source is nothing but a license for software. Now, that doesn't mean someone didn't create an open source community, but that's not open source. Open source and or free software is only about the rights that you obtain or declare by the use of open source. Full stop. Okay. So if you identify I'm an open source guy, you're going to have problem with this versus I'm a technologist who happens to use open source. Now, I understand when you start getting into like the Free Software Foundation and RMS and even, you know, say someone like ESR, who's not really involved anymore. You know, there's some ideology there, especially with, you know, GPL. And that's okay. But again, the GPL was created literally so you had software rights. That's no different than what's happening today. The velocity is just different, right? Now, if you're talking about licensing, I think the jury is still out on that. We don't have any precedence for, I rewrote Python in Rust, and therefore I do not have to adhere to the Rust license. That precedence does not yet exist, and we're probably a few years out from it due to lawsuits. There has been examples of that recently. But I mean, even if we go back to, you know, the Google Dart Java API type deal, right? Did they violate Java's license because they re-implemented the API? I mean, even that's not a precedence for a complete rewrite in a different language. So the solution to that, and this goes back, I mean, the solution outside of lawsuits to that, and this goes back to what I was talking about with associating with a legal framework, is Things like patents. Patents and trademarks are how you're going to end up protecting open source of significance. There are things that even Linux can do, even though there is question about the validity of the Linux mark, legally speaking. Trademark, you can say, look, you can say you're a clone, but you cannot say you're X, right? ([Time 0:15:00](https://share.snipd.com/snip/92a466c8-0092-444c-8d09-d808ba627ef0)) - **Tech Identity Causes Threat-Based Stress** - Identifying with technology triggers threat-based stress that harms health and collaboration. - Joshua links threat stress to shortened telomeres and recommends identifying with community roles like 'people advocate' instead. Transcript: Joshua Drake It's why the conferences that run Postgres Conference, their motto is people Postgres data. Okay. And the reason is, is that if you're identifying with the technology, you're ultimately, and we've seen this decade over decade over decade, you know, VM versus Emacs, Linux versus BSD, Microsoft against the world. If you're identifying as part of a technology, you are constantly, I'm going to get a little weird on you here, you are constantly going to be a part of what's called threat-based stress. And threat-based stress always comes about from an attack on identity. You can see it in the social norms that we have today. We don't need to go into that, but you can see that, right? And as a side note, threat-based stress will actively shorten your telomeres, which are directly related to your lifespan as well as your disease profile. So if you are identifying with things that put you under threat-based stress all the time, and yes, this is actual science-based. Go ahead and look it up. It was research done by a Nobel prize winner called the telomere effect. You were actually shortening your lifespan. So basically what I'm saying is, dude, let it go. Identify with the community, identify with the people, be like, I am a developer advocate. I'm a community advocate. I'm a people advocate for people that use Postgres. But Postgres itself is irrelevant there. ([Time 0:19:02](https://share.snipd.com/snip/516e5807-57ed-4fbd-b577-ed4939024c8b)) - **AI Turned Weeklong Ops Into A Three Hour Task** - Joshua used Claude to build a complex interfacing tool and summarize 20 years of Redmine notes in three hours. - The AI-created deliverable replaced what would have been a week of systems admin work. Transcript: Joshua Drake Further than our purpose allows for about 20 years now because we made people tons of money. It's not because we're smart. It's because we made people tons of money. They don't care how they make that money. If they can make money from people that aren't as smart as we are, they're going to do that. Right? So what's going to happen is technology is going to go back to where it wasn't unique and i'll give you an example today before this podcast i using claude i was able to uh create an entire Interface between claude and using AG Ledger as my durable log and node reaction. I was able to clean up a whole bunch of Incas stuff that I didn't even know I had on my machine. I was able to connect to Redmine and update an entire database of over 20 years of Redmine and summarize it into an executive report that I could review without taking days and days and Days and days to write it and research. I did all that in three hours. The Claude A.G. Ledger Incas work alone would have taken a solid systems administrator, of which I am, a week. I did it in three hours. ([Time 0:28:39](https://share.snipd.com/snip/4f7098f7-24f9-46cc-bb75-cc287b4008b8)) - **Use AI For Scaffolding But Keep Senior Review** - AI currently scaffolds MVPs very well but still needs senior review for UX and security. - Joshua recommends using AI for documentation and scaffolding, then have experts finalize and harden the deliverable. Transcript: Joshua Drake And this is how I actually got into AI. I am off grid. And I mean, when I'm off grid, I don't have the nearest power from any kind of utility is over a quarter mile away. I am literally off grid. Everything runs on solar. And sure, my solar equipment all had, you know, it's got its little Bluetooth dashboard and things like that that I can use on my phone. But I wanted a centralized portal. So I said, I'm not going to take the time to learn all these libraries and build it all out. So I said, create me a solar dashboard that works with Victron equipment and JBD batteries. And it did take me about 12 hours. Now I could do it now in probably an hour and a half, right? Because I was learning how the prompts work. I was learning its attitude, its behaviors, its personality. And yes, it has all of that. But I now have a solar dashboard that I can connect to via a web browser. It's also got a text-based user interface written in Rich. And I can see what all of my critical systems are doing at any given point. Those are features, right? So will we see a situation, let's say this year, where all of a sudden you hear 100% written by AI and it's not something that's complete crap? No, you are going to need, the features will be MVP level. They may work, but you're gonna have to have a UX person to move this pixel this far over, fix the mobile layout. You're going to have to have a senior developer review the code base, preferably using something like plan mode and cloud. ([Time 0:35:07](https://share.snipd.com/snip/86240bba-1235-4351-bf4e-8d8a15dd1cd5)) - **AI Will Shrink Junior Roles And Rework Teams** - AI will reduce demand for many junior roles, project managers, and marketing jobs, while shifting teams to AI-supervised staff. - Joshua predicts small-business software will rely on a few developers plus low-cost AI subscriptions. Transcript: Joshua Drake Different industries are going to be affected differently. I think the biggest industries that are going to be affected are legal and health. Our industry is a little bit insulated because you still want a human to verify that the config file that you just wrote, when you restart it, you didn't just open up the bank to people In some bad actor country, right? There still has to be a throat to choke. So I think that we're a little insulated as a whole. I do think it's going to deeply affect the number of junior roles that are new to hire. I think it's going to deeply affect things like project managers, product managers. Instead of having five product managers, you're going to have one that reviews what AI says, and you're going to spit polish it and send it off. It's going to decimate marketing. Because marketing has not changed in 150 years. The mediums have, but marketing itself, the theory of marketing has not changed. It's going to decimate it because it knows all of it now. So you can go in and say, build me a marketing plan based on the cited research from, and it just will. But to directly answer the question, no, it's not there yet, at least at a user interface perspective. But it's close. And any small, where I think the impact is going to be the biggest, let's say you have a company of 25 people and you have two developers. One of them is infrastructure. One of them is a client facing front end. And you pay a bunch of people in Ukraine or Poland to help you out. You don't need that anymore. You need two $200 a month cloud subscriptions, and those developers are going to punch out everything that you need. It's not going to work on the higher enterprise scale, but higher enterprise isn't where all the money is. ([Time 0:37:10](https://share.snipd.com/snip/6b20c688-92e5-404a-8374-d724f177ca12)) - **Make AI An Opportunity Not An Excuse** - Reframe AI as opportunity and invest in people to stay relevant. - Joshua warns those who cannot reframe will be laid off while others use AI to gain advantage. Transcript: Joshua Drake You've got two choices. You can think of AI as the end or AI as the beginning. It's up to you to make it an opportunity. There are opportunities for every one of us in tech to make AI something that is beneficial to us. There's also opportunities to make it not beneficial to us. Those are the people, the people that cannot reframe, cannot consider the opportunity, cannot take advantage of the opportunity. Those are the ones that are going to be laid off. ([Time 0:39:37](https://share.snipd.com/snip/6eb8de38-530c-4d5f-b2ff-f4bf1e66a165))