Everyone on your team builds software now. Who keeps it safe?

**[00:00:00]** Armando: Hello everyone and welcome to The Web Talk Show. My name is Armando Perez-Carreno and with us today is Cory O'Daniel, CEO and co-founder of Massdriver. How are you Cory? Cory: I'm doing pretty good. How you doing? Armando: I'm doing great. Thank you for asking. I am really excited about our topic today because as vibe coding and just software development has evolved a lot over the past few months, a couple of years, it has allowed many people to be able to build things that just wasn't a possibility before. But there are things that we must consider before launching things into production. So please give our audience a little introduction about Massdriver and how you got to build it. Cory: Yeah. Yeah. So Massdriver is a what's referred to as an internal developer platform in the in the software space. The idea is that there's a vast amount of infrastructure stuff in the cloud that development teams need to be able to manage but they frequently don't have necessarily like the full breadth of expertise to do that. And the reality is like most organizations like not a single person does right. So like you look at a database and you might say oh I know what database I need but like the costs the compliance security of that there's like three or four different parts of the business that have like stakeholder share in that right like security team is like hey this secure thing compliance team is like hey here's the legal frameworks that we have to adhere to right and so it's not so simple as saying like oh I'm going to go get that database and I'm done right it's like oh I have to make sure like everything in the business is adhering to like our standards and non-negotiables so the idea with Massdriver is that it should be as simple as it sounds which is a crazy idea. I just people like get database, right? And so the idea is your operations team, security, compliance teams, they take the tools that they already have and then we have this packaging tool that you can run. It packages it all together and then puts it in Massdriver. And so from there, and this is where we've kind of gotten into the citizen developer and the business developer is from there, you just diagram. **[00:02:04]** So like the kind of stuff you do on a whiteboard or like when you're diagramming stuff like thinking through how you want like a system architecture to work, that's all you do. And so you say, "Oh, I need a database. I want to put it on this network." You drop database on canvas. You connect it to the other network and it builds what you see. And so what a lot of people see is they go, "Oh, it's AI." Like you're drawing it and AI is building the thing. It's like, "No, there's literally zero AI in the box whatsoever." Your operations team has pushed all of their tooling into Massdriver and we're just taking the diagram and backing it out to code that has been written by probably a human, but more and more likely an AI. So like the real goal is like to make it as simple as it should be, right? Like I diagram my system. I have living documentation of what I've built and the system is what I have drawn which is not usually not usually how it works. Usually you go do a bunch of and then draw it and then say I'm going to keep this drawing up to date and then not a single person does that. There's one guy he's saying I do that. Yes, I know you do that guy but the rest of them don't. Armando: Right. Cool. Cool. Well, that I mean we're going to dive a little deeper into all that because it's it's not the way like you said most people see it and I think it'll help a lot to understand where you're coming from to understand that's a many of our listeners are not technical are not from IT and not software and then they might think okay so what's why why do we need that like what could go wrong otherwise so can you tell us a little bit about your background and what gave you the push to build something like this? Cory: Yeah, so my background, so I've been in the cloud space since literally like the like the modern cloud like AWS's and whatnot in the early 2000s kind of came into existence. So my background was working in healthcare data centers and so I moved to California early 2000s like right when AWS was launching so like a minute ago, right? **[00:04:08]** And I when I moved to California, I changed career paths. I was like, I don't want to work in data centers. Like I don't want to do racking servers anymore. I want to write software. And so I got my first job writing software right as like the cloud is coming out. And so my boss walks in one day and he stops at my desk and he's like, "You used to work in a data center, right?" And I was like, "Yeah." And he's like, "Cool." And then leaves. And I was like, "Oh man, fantastic question, software engineer." fantastic conversation as well. And so about an hour and a half later, an email goes out to the entire engineering staff that we're moving to the cloud. And everybody's like, "What does that mean?" Like like it's not even like a super common term. Like the software engineers know what we're talking about, but everybody's like, "What do you mean we're moving to the cloud?" And my boss announces that I'm going to be leading this initiative. Oh wow. Because I'm the only person on the team that has software development experience and data center experience and I'm sitting here thinking this is my first software engineering job. So I wouldn't really say that I've got much of the software engineering experience. And so I got kind of like like pigeon holed into this role that inevitably would be called DevOps. Software engineers that are hearing this are like it's not a role. It wasn't a role. It was an idea. It was a culture idea. But it slowly kind of calcified into like a job. Mh. And so I got kind of pigeon holed into this role very very early. And so from company to company it was like oh this person has built platforms for the software team that like run on whatever data center cloud etc. And so like every job after that I was kind of like in this like I don't want to say no man's land but I was in this place like building dev tooling to make it easier for these developers to access cloud. And so as I kind of go through these organizations over you know 20 years or so I end up working for not for with Google. I have to use the words there very carefully. I was a consultant for Google. So the word is with and so a lot of the companies I'd worked at were like smaller companies you know 50 to 100 developers. **[00:06:14]** It's still a lot of developers but it's not like Google big right. And when I start consulting for Google, I'm going into companies where people have like 4,000 developers, like just way beyond like what I've ever worked with. And I'm going into these orgs and I'm like, these huge companies making billions of dollars a year have the exact same problem magnified that all these startups have. Like the engineers are sitting there and they're thinking, "Oh, we got to do it. I need a database for this new app that I'm working on, but I got to make the database. Got to think about disaster recovery. How do I secure it? How much should it cost? Like all this other stuff and their PM's like can you build the feature that I need? Like I need a feature and they're like well I need a database and to get a database I have to figure out all this other stuff. But when you get to big companies like that process which might be like a week of an engineer just dropping everything and like figuring that stuff out in the startup and a larger enterprises there's become tickets. You put a ticket in a system and then you're like I'm just waiting on somebody else to do it which might be today it might be 3 months whatever. And I was I was dumbfounded. I'm like, wait a second. Like I thought we all did DevOps and we all were professional about this. Now I'm getting these big organizations and it's just like, oh yeah, yeah, yeah. When we need something, we type it in Jira and somebody else does it for us. And it's like, oh, that must be extremely efficient. And so that's where the idea of Massdriver came from. I was like, okay, so this problem that I've seen for years at very small scale, like medium-sized businesses exists still in large organizations. And I was like, how do we make it how do we make it where it feels as easy as it is, right? And it's like as I'm working with these big companies, I'm like, how do you think about infrastructure? Like developer that doesn't know how to set up a database or do like how do you think about infrastructure? And the guy that I'm talking to, so I'm like workshopping Massdriver in my head, but I'm like working with people because I'm trying to build them a platform, too. I'm trying to help them. And he's like, I kind of just think about it like I think about on the whiteboard. He's like, I've got my database here. And he draws a box. And he's like, and I've got my app here. And I draw a box. And then he draws a line between him. He's like, and I just assume like he draws like all the security and cost stuff just kind of happens in here. **[00:08:20]** And I'm like, oh, I'm like this is okay. And so like that's literally where the original idea for Massdriver came from is this guy, he's like he knows his stuff. He's a senior platform person working on APIs for this big ad company and he's just like database and app box on screen. And I'm like that's really like that's the fidelity that you have into this. And so when we first started working at Massdriver I was like that's the fid fidelity I want. I want it to literally look how engineers just kind of scratch something on a whiteboard and like that should be how it works. And so we kind of worked from that backwards like okay if it was as simple as this what is the operational pain to make that sew. And so we've been just chiseling away at making the operational pain side almost painless which is pretty nice. Wow. So, yeah. So, that that was maybe like five or six six, seven years ago now where I left Google and I was like, "Okay, I'm going to go do this." And so, I went out and started talking to VCs. I got 170 VCs to tell me no. 170 170 173 I think to be specific. And then we got into Y Combinator. It was crazy. Do we have one of those? Like I feel like I tell my friends this and my co-founder this all the time. Like I am amazing at stumbling backwards into pure luck. Like that's how I got my first job in California. That's how I got my job at Google. Just stumble backwards into it, bump into somebody at a coffee shop, start them and they're like, "Hey, we're hiring this role right now for what you do." And I was like, "Oh, that's cool. Like network is your net worth. Give me the job." And so why combinator? We apply 5 days after the deadline. So like it's already done. We're like after the deadline. After the deadline. We're like 2 weeks from the next batch starting. I've not sold a single thing. And then in between like so we've applied late. In between then and like the start date, somebody emails us back from Y Combinator. **[00:10:20]** And just so happens to be that morning I got my first sale which was like it's like decent five figure sale for like a startup product and I got my first angel check for like like I think like 100k. So all of a sudden we go from a startup that like been around for 4 months like bunch of VCs told us no. We literally have like zero momentum. Applied for YC literally because somebody this is during COVID I'd never considered YC. Like there's zero chance I moved from Southern California to freaking San Francisco. Like sorry guys, love it up there for the weekend, but like SoCal is where it's at. I'm staying here. But they still had remote YC because of COVID. So I'm just like whatever, we'll apply for it. We're never going to get in. We have zero traction. Bing bang. One day we go from like no money in the bank to like 150K. YC responds the same day. I tell them about, hey, we just got our first angel check and we just closed our first sale this morning. So then we get our interview the next day and I'm like, "This is insane." Like they're they're talking to us. They interview us and they're like, "We usually let people know same day." I'm like, "Great." Like we're totally getting into this thing now. We don't hear from them. And I'm like, "That's very weird. I assume they would have at least told us, beat it, like scram." and then so we get an email from the next day and they're like, "This is very awkward. Can we do your interview again?" And I'm like, like, "Well, that can't be bad." Armando: No. Cory: Yeah. I it's inconvenient and extremely weird and I'm like okay well yeah let's do it again. So we get on and it's completely different people and it's like the same questions and I'm like this is so weird like it just feels like Bizarro World. So I answer the questions and then like an hour later they're like you're in and I'm like it's like 3 days after YC started so we're like starting late and it turns out there was one spot left. Somebody had backed out. That's why we got in after after the deadline. We somebody backed out or was removed for whatever reasons. There was one slot left and they do these like groups or batches, I can't remember what they're called. Batches and groups. **[00:12:20]** The groups are the smaller things in YC. The group that we were interviewing for first, there was a direct competitor in that group. And so the group was like, "These guys are great." They're like, "Proble is there's a competitor in the same group and they don't do that." So they had to like shuffle another person out of a group and they're like, "Let's take somebody from you, put it over here, and then you have to rein with that group." And so it's just like just the amount of like weird dominoes that fell to like make it so. And then it was like once we were in YC it was like like everything just kind of started taking off. We started getting traction. VC started coming to us and I was just like you told me no four months ago. I'm going to tell you no now fool. Armando: That is incredible. I like those stories because a lot of times people especially entrepreneurs can get into a rut when like nothing's happening or like it feels slow or whatever the thing might be that day they might feel like I don't know how I'm going to but you never know. It could be that the next day everything just magically aligns. H Cory: For sure. Have you seen It's so funny you say this. Have you seen that meme? It's like it's like a two guys underground and they're digging and there's like a big wall of diamonds and like one guy is like one step away and gives up and the other guy is just like way at the beginning and he's like ah and it's like ah never give up. It's so funny. It's like I look at that meme all the time and I just think never give up. Yeah, I get that. Like that guy was so close. But there's also like the sunk cost fallacy of that meme like those diamonds could not be there and that is every single day as a startup. If you're sitting there and you're thinking, "Okay, I just got kicked in the balls 75 times in a row. What's the chance that it's going to happen a 76th time?" And then it doesn't. And you're like, "That's great. What a relief. I didn't I didn't get swatted again." but it is it's just like it's just it's sometimes it's like you just got to keep trying and trying and banging your head against the wall. And it's hard to figure out like am I banging my head against the wall and like the wall's going to eventually fall down that I'm going to make it through it or is the wall going to split my head? Like I'm not sure which way this is going to go. But such a part of life. **[00:14:24]** Armando: It's it is and it's a it's a gamble to a point. You have to say do do I keep doing it? But you have to look at outside factors as well. Like sometimes it's just pure persistence and you know in your soul that you're going to pull through and there's no other evidence that you're going to pull through but you still do it and you do it and that's amazing. But and then for some people it's like no no it's just not going to work. There's some things where you have to step back and look at the evidence, but I mean some of the greatest inventions out there are just pure persistence to a point that you had to make it possible. And I mean we were flying rockets to space, right? Like that's still mindblowing to me that someone decided, oh, we can probably get out of the atmosphere and they achieved it and where it was impossible before the first person that does it then everyone else like, oh, it's possible and but through two years later they can do it. So, it's like this weird unlock when somebody else has pushed through. Cory: Yeah. And for anybody who's a space or physics nerds, I'll just point out really quick. While my company's name is Massdriver, we do not actually put things through atmosphere. So, it's very funny analogy you had there. I'm like, that because that's actually what the company is named after is Massdriver, like launching payloads into space, launching them into the cloud. I was originally a physics major, so that's where the name comes from. Me, my homage to wanting to do something more interesting than software development. Armando: Ah, wow. But whatever. This is very interesting. Cory: Yeah. Our Massdriver is a lot less interesting than things going into space, but I still think it's fun. Armando: Yeah. Yeah. No, it's a cool name. I like I like the name. It's very curious that what we expect sometimes is completely different than than what things are because like you said, some people are like, "Ooh, database. **[00:16:28]** Yes, I'll take that one." And the amount of complexity behind that decision is something that that most people can't realize because they've never interacted with these multi-systems where it's not just about can it store data. It is can it do it fast? What's the latency? What's the security look like? There's all these little things like you mentioned at the beginning that if you are not an expert at that and you just want the feature, it's you're like why why isn't this moving so fast? But at the same time, when you are the person who's tasked to do that, it can take so much time from you. You just want to make the feature that they ask you for, but then you're trying to figure out everything else might just put a wrench in it for so long that then the feature becomes unnecessary or obsolete at some point. Like there's there's so much going on. How did you figure out a way to make it visual, right? But then actually have it work behind the scenes. I mean the concept sounds amazing but what actually if you can take us through like what happens after you do that diagram and tell it go like what is happening behind the scenes for this to magically go into existence. Cory: Yes. This is I guess hopefully I don't get too far in the weeds for anybody here, but like the way the way that Massdriver works and what's interesting about it is like again there's no like there's literally zero AI to our product. But and this is not in our marketing materials, but this is like our internal design spec. And I hope nobody takes this the wrong way except for like the one person that will and I'm talking about you buddy. We designed Massdriver. **[00:18:28]** I said earlier like we wanted to make it like as simple as it is to talk about it. My app uses a database. Like that's the extent of like how you'd say oh I need I need my SQL. I need I need right like so casual the way you say it but like the amount of effort behind that is intense. And so when we designed Massdriver we're like we have to design this for the absolute dumbest software engineer on your team. And like that was that was our guiding light for like four years is like this has to be so simple that literally the dumbest person on your team can do it. And the reason why we did this is like we're staunch believers in the idea of something called self-service and infrastructure. It's like the ability for an engineer to get what they need and move on, right? Like that's what creates good products, fast products, good MVPs, good time to market, right? Anytime you artificially stall that drive, whether it's in the PM, the developer, the designer, like you're you're negatively impacting the business, right? Like whether it's from a revenue point of view or an experimentation point of view, right? So we got to get these people through quick. And so what happens is if again like the dumbest person like sorry to say the dumbest person it's just like the easiest way to think about it. If the dumbest person on your team can't act with autonomy, right? You're going to create classes of people in your organization that can act with autonomy and can't act with autonomy, right? And so then what you end up with in many of these organizations is you get teams where work is routed to because you know these people can get things done quick, right? And then you start to see teams where it's like, oh, we have the senior people and we have the junior people, right? The senior people can change infrastructure, the junior people can't. And that sounds perfectly reasonable to almost everybody on the planet, but immediately what happens is kind of like a fresh juice layers will separate, right? And so what happens is the seniors that are allowed to make changes all of a sudden are doing all of those changes and now the juniors are writing software, right? And so now the seniors become gatekeepers, right? And so like this thing that seems reasonable just creates another layer. So, okay. Well, now what we used to do is we'd open up a ticket and now we don't open up a ticket. **[00:20:30]** I just ask my senior guy on my team to do it and he goes, "I'm busy as slack me." Okay, they'll slap you or I'll open a ticket for you. Right? We're right back to where we started. Right? If the dumbest person on your team can't act with autonomy safely within company's non-negotiables, then you're never going to truly have self-service and you're always going to slowly backslide to where we were, right? And so when we're designing Massdriver, really what we try to do is have as little opinion in the box as possible. So it's funny like we'll do demos with people and we do pretty amazing demos. And most of our pilots, we build that, we build the pilot software for people based off of one conversation with them. And when we hand it to people, they're like, "Whoa, like how does this do this?" And it's like good oldfashioned software engineer, just a little bit of software engineering. And so at the heart of Massdriver, we have this thing called it's called a type system. And this essentially for people that aren't software engineers, it's essentially it's a concrete or like a almost like a contract for how you're going to talk to a computer. Except for in Massdriver, each of those things that you're talking to is the box on the screen. So your organization says this is what database looks like to us. It's not just you saying, "Oh, I need a database." We're saying, "Hey, you need a database and there's some very simple controls that we're going to expose to you about that." Right? And so instead of getting into the computer science side, I want to like bewilder you with buzzwords for a second just to like hit you with how hard this can be. And it's funny because like the way that this pans out in organizations, people do this. They just do it in weird spots. So let's say you're a developer and you said, "Hey Cory, I need a database." Okay. How much should it cost you a month? You don't know. You're the developer. You don't give a That's the CFO's responsibility, right? I'm like, I don't really know. Like it's an important feature. So whatever important features cost us a month and then I as the operations person say I guess I'll guess I'll guess for you and that'll be somebody else's problem in like two or three months when the bill's high. **[00:22:38]** That's fine. How often do you need this database up? Like how how much can it fail? Oh well I mean we work on a global web service. It needs to be up 24/7. Okay. If it went down what would happen? Like if it's if it's gone for like five minutes, it's like, "Oh, the world would end." Why why would the world end? Well, because if it's down, I'm going to get paged and woken up in the middle of the night, which would ruin my evening. Wouldn't really ruin the end users. End users aren't going to notice the thing's down at 5:00 a.m. Or, you know, for 2 minutes or whatever. But as software developers, we go, "Oh, I don't want it to be down because if it goes down, then I look bad, right?" So you're like, should be up all the time. Well, that those two questions I just asked you are at opposing ends of the spectrum. I can keep this database up forever no matter what. It's just going to be very very expensive. Right? So, if you say, "Oh, it should be cheap, but it needs to be available." There's a there's a median that we have to come into agreement with you. And we haven't gotten into security, compliance, ownership, or about eight other questions that kind of jerk this config this way or that way. So now the problem is when I ask you those questions as a operations person, you're asking me, right? Let's say you open that ticket. I'm literally responding with those questions in the ticket and then you're answering them or you're slacking me. We have this conversation and what's funny is John comes along. John goes, "Hey Cory, I need a database." And guess what I do? I ask him the same freaking questions and he gives me the same vague answers, right? It's groundhog day over and over again. So the first thing to think about like when thinking from like the Massdriver's worldview from the operator's point of view is what are the most common questions I ask this developer to kind of work through what they actually need from this database. Let's make that a form. So now when you drag that database onto the screen, the first thing that happens is this form opens up and it asks you those questions and you would say, "Hey, wait a second. Am I filling out a ticket?" No, no, you're not filling out a ticket. Normally those questions would be in a ticket but as you answer these questions what actually happens is the infrastructure team like the software like the software development team that manages cloud they register things that can act on these questions. **[00:24:48]** So now like oh these are my four questions that I ask every developer and when the answers come through it executes this layer of logic that makes that so. And so now as a developer I come back along and I say oh you know what like I said that has to be available 24/7. I didn't really answer anything about cost because like costs and availability are kind of tied one to one. So we just take the cost question out. You tell me how available this thing needs to be. We'll back out the cost for it like what the cost limitations, budgets are, etc. Right? So I come along as a developer, I'm like that database is really expensive. My boss is mad. Like I said it needs to be available 100% of the time. I'm going to lower that to 99% of the time. You're like oh that doesn't reduce cost by 1%. That reduces cost by like 25%. Right? Going from 95% available to 100% available. That is an near exponential curve of costs, right? It's very easy to get something 95. Very hard to get close to 100. Almost impossible to get close to literally 100, right? And so now what happens as a developer, I can come in, I drag box. Google box, great. Here's four questions, five questions, couple of questions that I definitely know the answer to and they're phrased for me. How what is the uptime of your service? Now that's a different question than how often you how available do you want this thing or how many regions or zones or where in the world do you want it? I say how what is the guarantee you giving to the business that your feature will be available and you go 95%. And I go great put that in the 95% box. I'll figure out the cost and availability for you. But that was a question you know the answer to right and so now you're coming in you're following a standardized path. You draw box drop answer for questions database exists. Now I connect the lines right and so kind of going back to that contract that contract exists all throughout Massdriver how I interact with one of these boxes and as you draw the boxes and connect the lines how the lines talk to each other too. So now database connects to app database understands its app understands its database right computer level not like you know personally reading words and so now they agree on let's set up some firewall rules let's set up some permissions to access this thing and so now like all my access my compliance my security is handled by that same type system right and so now the person using the system it does feel like a whiteboard I go and I draw the box on the screen I draw the line to my app the guy in there in the room with me. **[00:27:06]** We're walking through stuff. He's like, "How available does that need to be on?" I'm like 99% of the time, right? Like just cut answer those kind of questions. It's like as simple as having that conversation comes into existence. Armando: Wow. Cory: So hopefully I didn't lose anybody. I tried to minimize the tech talk and buzzwords there while also putting enough in there to make it annoying because it should feel annoying. Armando: Yeah. Well, because it's a lot of work. I it makes perfect sense because it's like people don't realize this but a lot of it before AI like stuff happens with software right you don't need AI for for a lot of things it's like when you go on an e-commerce site and you see a product what can you do you can view the product you can add quantity you can add to cart you can check out you can pay magically the product appears at your house right but what really happened is the software has specific speific things like you said like a contract that the everything behind it can understand you do this action this happens then it goes to the other software for distribution logistics and eventually gets to your house right now this is sort of very meta because you are making a tool to diagram things that then get built or executed which is great but the logic stands right it's a it's a discrete operation. Cory: Yeah. Yeah. And what's what's amazing about this, like what lucked out for us was again like our north star was we're building this for the dumbest software engineer on your team. And again, people are like, "Man, I know who you're talking about on my team, but I wouldn't call him that." But now everybody has the dumbest software engineer on their team, and that's Claude, right? That guy is he's smart, but he's a dummy at the same time. And what's interesting that's happened is the you know most of the users of mass it's funny it's like our users and our customers are two different people. Our customers are the people that tend to own cloud compliance security operations engineers etc. **[00:29:08]** The users of Massdriver are their customers which are their software engineers right and so what's happened over like the past two years with like the advent LLMs and code generation is when you look at organizations now they used to have like sorry I'm starting to laugh because it's just it's this whole thing happening was very funny to me watching it happen because again like my background is IT there's this concept of these people that have existed in IT that create something called shadow IT people have heard this term before and the term to refer to these people that that create shadow IT as citizen developers. And so you can rewind time like back to the 80s and these people still existed, right? So like let's go not 80s. I don't know if anything existed in there. Let's go back to the 90s, right? So let's go back to the 90s. Surgery. But so if I go back to the 90s, I worked at a hospital or hospital chain in Florida called Florida Medical Clinic and had a whole bunch of offices across the state, couple data centers, and all of a sudden one day I get a call at our help desk and it was like, "Hey, the MRI server is down." And I'm like, "Wow, that's a that's big news because I didn't know we had an MRI server. What are you talking about? We have an MRI. MRIs need a server." I'm like I'm blown I'm mind blown by this is somebody calling from like the MRI center like MRI servers down and I'm like where is this thing and they're like oh it's under my desk and I'm like it's under you have a server under your desk with like patient information on it. They're like yeah I'm like how'd this server get here? And they're like oh we just we upgraded the MRI machine to the new thing and they sent they just sent it to me Doug whoever working in the office. They just sent it to him. He just hooked it up under his desk and I'm just like like how did you like network and everything? Oh, we just found like a network cable and plugged it in. So, it worked great. Right. It was the '9s. And like that's how most was in the 90s. Like like Betty and accounts payable just bought some software. They just mailed her a server with a thing on it, right? Or a disc and they bought a $400 Dell and stuck disc in, right? That's citizen development. And like for the longest time, we kind of like like calmed the chaos of citizen development just because like the access was hard. **[00:31:15]** It's hard for me, a person not in technology to introduce technology to the business. This changed in the 20s or early 2000s. You started getting like Office online, right? Google Docs. Oh, okay. I'm starting to put all this. I don't need an app now. I'm just going to use this spreadsheet as app. Okay. Well, that's still citizen development. You're introducing technology. There's compliance, risk, cost factors in there. And then there was this brief moment right before the LLMs where there was the no code movement. You draw you drew everything. Everybody thought math was like a no code tool at the time, right? And so then you started getting this mass proliferation of like marketing pages, dashboards, etc. Where like the business was creating stuff now. And now with LLMs, what's happening is there's people that were doing no code and introducing all this like shadow it for years. It's like, wait a second, the means to creating software is now accessible to all. Which means I don't have to go into a spreadsheet and like fiddle with something. I can just say I wish this existed and the computer will do it which is amazing. It's super cool. I genuinely like I might sound sarcastic right now. That is awesome. Like as a person who's worked trying to make it easier for developers to do something like my goal has always been like the business should run efficiently. I want the market to tell us whether to slow down or go fast not like gum in the works, right? And now the idea that somebody in business can sit back and go, "Hey, there's this thing that I've wanted forever and it keeps dying on the backlog like developers never get around to it. It's not a priority for the business, but it sure as hell would be a priority for my team. It'd make us more efficient, might make the company more money. I want that person to have that. I just don't want it to be my problem tomorrow when they do, right? I don't want to be, oh Todd made this database app and he made a mobile app for SDRs on the team and like Cory, their database is down. That's your problem now. I'm like, I don't want that to be my problem. I want you to have that autonomy, just not for to create problems for me, right? And so what's wild about this is it's very easy, and this is where like most of Massdriver's business is coming from today. **[00:33:17]** These people sit down and they'll sit down with Claude and they'll like, I want this app to exist. And then it does, right? I was talking to a friend the other day. He's not a software developer and he's like, "Yo, I've got like an app. I got like 20 users. Like people are starting to use the thing." I was like, "Cool. Like, where are you where are you hosting that?" And he's like, "I don't know." And I'm like, "Holy I'm" I'm like, "Holy what do you mean by you don't know?" I'm like, "What? Like, who have you given your credit card to?" And it's like, "Chat GPT?" And I'm like, "Well, chat GPT is not running your app. I know that for a fact." And it turns out like like get I get into it. I'm like looking at his code and I'm like, "Oh, it's running on Supabase and it was just like Claude or ChatGPT or Codex, whatever it was, signed up for Supabase for him, set up his app to deploy there and they have a free tier, right?" So he never even thought about what cloud he was on. Like to him, he was just like, "Chat GPT, right?" And so like that's Armando: Make no mistakes. Yeah. Make no make business, no mistakes. Which make me a billionaire. That's awesome as a solopreneur, you know, or somebody who just has an idea and wants to get it up. Cory: But when you're an enterprise company and you've got 4,000 software developers, 3,000 databases, like a thousand networks, and then all of a sudden Todd rolls up and he's like, "Yo, I got to make a sales tracking database. It's going to store a ton of PII in it." Claude, let it rip. And Claude's like, "I'm going to put this on a free Supabase." that does not work in brownfield enterprise organizations, right? And so, right, and so what's wild with Massdriver is those same simple questions. Oh, hey Todd, you need database, right? So, they're still sitting at Claude or Codex. They just happen to have like Massdriver tied into it. And now these desires of apps that they have all get created with all the compliance, security, cost, this entire framework around how to get things into the business and move them along. Right? So now I say, okay, I want this app. And Claude goes, "You need a database for that." And instead of it just going, "Oh, I'm gonna go put on Supabase." It goes, "I can see like Massdriver has this tool that you can link into it." And Claude's like, "Oh, we actually only have two options for a database. **[00:35:26]** We can put it on Postgres, which is a good transactional database. If you're doing some e-commerce stuff, it's going to feel great. Or we can put it on Dynamo, which is a NoSQL store, real fast, like yada yada yada, right?" It'll tell the person, "These are the two options that you have." And then Claude will actually say, "Based on what we're talking about, this is the right choice for you." And the person goes, "I don't care. Thumbs up. Pick the right one for me." And then Claude goes, "Again, those questions are registered." So now Claude goes, "Oh, I can't make it until I have an idea of like how reliable should this app be? Like how often do you need this app up? Like percent of the time?" And the person's like, "This is an internal tool. Eight hours a day." Like it doesn't really matter, right? And so now like the only options are 90%, 95%, 99.9 and 99.999. The person goes, I'll just take 90%. I don't care. Like it this internal tool doesn't really matter, right? And so now I'm answering really simple questions that even a business person can answer. Hey, like what's the budget of this thing? Like if the price of this database went over an amount, like when would you want to know? And I'm like, oh dude, if it cost more than 15 bucks, my head's on the line. Like $15, dude. Right? And so now now business people are like answering the same questions that a developer is and they're thinking that they're thinking okay what do I need right and the questions are being presented from the point of view of like all of the stakeholders involved from the back end. So now I can let you make as many databases as I want. There's no risk to me. I know that it's consistent. It's going to be the exact same as the last one. It's going to have these couple of parameters changed based off of your input. I can see every database that was created by everybody now. And if you're like, "Hey, my thing's broken." And I can I can I can literally hop in there and look at it and go, "Oh, you know, your app's not connected to the database, dude. You just got to draw that line right there. Claude should have done it for you." I don't know what he was thinking, but right. And so like it's funny like this is the world we've ended up in. And it's funny, we meet so many companies where like this is already happening. Like we had a c we had a customer, they're like, "We have 400 developers." And then I talked to them a couple weeks later and they're like, " we need to like get 4,000 seats." And I'm like, "Whoa." Oh, I'm like, "Do you guys buy another company?" He's like, "No, the CEO just said everyone's a software developer now and like we have to figure out how to deal with that." And I'm like, "That's awesome. **[00:37:38]** Like, you have the tool to do it." Would have been very not awesome if you were like doing this on your own. But like that should not be a concern of you. Like there's going to be more throughput. There's going to be more cloud cost obviously because more people are building things. But like hopefully what you find is you're getting revenue or you're generating some sort of value from their efforts. And if not, you can see what you need to shut down later. Yeah. Armando: Crazy fun. Yes. Well, and also it's super funny because what you're talking about, people do not have an idea of all the little like we're using the database example, which is good. Let's say you're in Azure and you have all these possibility of containers, right? Or different sizes of things and you're like, "Yeah, yeah, I just need it to work this way or that way." But each one just to keep it running will will need so much more other stuff for it to run or might need to be shared. Like maybe you need an an whatever naming it be like E1 or E2 or E3 or whatever, but you might learn that someone on your team is already using one and these types of tools might tell them, hey, you don't have to spin up a full one. We already have one that's using like 1% of its usage, right? Cory: Yeah. Armando: But it's a it's super awesome. I think like even even when I'm when I'm teaching people to use like claude for themselves and smaller smaller companies, right? They're it's very nice when you can have these sort of rules that say this is the infrastructure I like and these are the tools that I like and this is the right way to do it and make sure you always have RLS and like a bunch of little things that will make them build better apps more securely, right? But at the scale that you're talking about like that, you can't have that in like claw MDs. It has to be something that hooks into the process and actually makes sure that things are compliant and using the resources properly. **[00:39:48]** Cory: Yeah. And it's funny like and that's exactly what our we call it the context engine and Massdriver that the thing you essentially plug into cloud. So claude, you know, like the way that most of this works is like when you're when you're working with cloud directly, it's generating code, right? And when you're using it with Massdriver, instead it's interacting with essentially our API like our type system, right? And so it's still generating your code for your app, but when it goes to do stuff in the cloud, instead of generating code to manage the cloud, it's saying, "Oh, I don't have to generate code." Which is great. Actually saves you a ton of tokens, right? Like it's funny. Like you look at this is going to be a really weird one but I put together this workshop recently where there was a workshop for like a large HR company. The entire workshop actually might not even be able to see if I can find this really quick. It's really Do we do video on the podcast or Armando: Yes, Cory: It is. Okay. So I'll Yeah, there we go. This looks like crap. My handwriting is crap. But where's my camera? This one right here. Right there. So, there's some words on there. I not I drew this on the airplane. I was on the way to the workshop. I was on airplane. I drew that picture. I uploaded the claude in the Massdriver context engine and it created the entire workshop while I was on the plane. Now, what's crazy about this is all of those resources behind the scenes were already in Massdriver. So, Claude only had to kind of orchestrate the architecture. And so what does this mean? If I did that workshop without claude and without Massdriver, it was easily man it would probably have been taken me two weeks to put that workshop together. With Massdriver, all this stuff is reusable. So like every time I've made a database for a workshop, I just use the same database component. Like oh same database. Like I might change the questions on the outside to kind of like gear it towards one audience or another, but the core component of it's the same. And so that picture I took generated the whole thing took like an hour because there was no code to generate. **[00:41:52]** I just had to literally piece stuff together, right? Which is crazy. So this is a workshop like for like standing up clusters and databases all this crazy for a big healthcare company and it's like the entire thing was done in less than an hour. The wilder thing costwise is I ran that same experiment. The token count was absolutely insane. Generating it from nothing, but generating it using the context engine where I can see what exists and what doesn't. It was less tokens than it takes to process the Cheesecake Factory menu. It was like 9,000 tokens for to build the entire thing. And so, which again, like in nerd speak, it's like that's almost nothing. It's negligible. And it's just like like like you think about this every single day. You're talking to this thing like, "Hey, help me with this. Help me with this. Help me with this." It's like if you have a bunch of pre-made decisions that are now deterministic, right? A, it takes any risk of claude out of the equation, but now it's like it's just executing stuff super fast instead of grinding through code and then saying, is this code right? And then you going, I don't know, try it. Right. Instead of going, oh, this code's right. I know that because the ops team already certified it. Mark's ready for production. I'm going to execute that code now. Right. Yes. Yeah. Yeah. One of the things that really stands out like when we're in workshops with executives, it's like yeah, yeah, the AI stuff and some of them are like glazing over, but once you mention the like deterministic tools, like yeah, but look, you can do this. It will process the data. No AI. Let me turn off my Wi-Fi. And they see it like we can do that. Armando: It's magic. And now like bringing it down like you're saying that that what was actually approved by the security team and compliance and everything. Now you know that anything that your LLM assistant will do is going to go through that and not hallucinate or make up like weird numbers or things or like spread your PII like it. This is great. I hadn't heard of Massdriver to be honest and I think the way you explained it makes a lot of sense. **[00:44:00]** I'm a software engineer so I at least for software engineers clear but I'm sure other people are going to get it because because of the examples. But it's brilliant because it's a it's a very good way to save tokens of course, but also to make sure that what you're building can be supported by the team that needs to support it because when it comes goes down eventually, someone has to be able to fix it and it's not going to be the person that built the little tool. Cory: Yeah. And like avoiding redecisions like red decisions are the worst. Like going back to that that guy, the guy that you're asking for the database and him asking the same four questions a h 100 times to everybody. It's like he asks those four questions that's time. Somebody sitting there responding going back and forth. But then there's like this like jelly electricity running in my head where I'm backing that out to like okay well what does this developer need? It's like I shouldn't have to do that more than one time. I thought through that and I maybe even made notes. I'm like hey when somebody says they need like 99% that means three zones and if they say 95 we'll aim for two zones. It's like, okay, you thought through that once, calcify it in code. Like, don't think don't think through that. You made the decision and now it's auditable and executable. Like, nobody wants to go and be like, hey, Todd, why did you pick why did you pick two zones? That's another thing that's crazy. So, now you can look at the config in the cloud and be like, it's on two zones. Why is it two zones? It's like, oh, well, our algorithm says if it's 95% or lower, it's two zones. It's like, well, it should be three zones technically for 95 or whatever, right? Like, it's just like you can now understand how you got that. It's like, "Oh, I understand how we got to two zones." It's not like, because otherwise it's random. It's like, "Hey, developer, how many zones do you want this thing in?" And they're like, "I don't want to wake up five a million worldwide." You have a toggle that shows whether or not I get woken up or not because I'm going to select the don't wake me up toggle, which is going to be very expensive for everybody involved, right? Like, Armando: Yes. Yes. This is amazing. Well, I know you have to go, but this has all been very enlightening and I really thank you for coming on to the show. **[00:46:03]** One quick question before we go. If someone wants to learn more about this, where do they go and is this solely for like enterprise use or can a small business or small organization take advantage of something like this? Cory: Yeah, so you can learn more at massdriver.cloud. You hit /ai if you want to see some of our AI specific stuff. Again, like there's no AI features in Massdriver. We're just designing the product to work well with whoever's introducing the change, whether it's software, operations, AI or CI like should be the same path to product path to production. Our typical customers are larger enterprises and we and we actually price on the site for that specifically. So if you go there today and you're a small team, you're like, "Oh, have to have like I think we I think at the minimum we require 15 users to use the product." If you're looking at it and you're thinking like, "Hey, I really want to give this thing a swirl." Like ping me on LinkedIn or hop on our Slack and ping me. I can I can set you up with like a, you know, a single user, three user account for free, so you can play around with it. If you're a smaller team, you're like, I care about this stuff. Like happy. The reason we do it at 15 is like we I a little behind the curtain on like startups like we spent a lot of time when we were in YC originally selling to startups but turns out like small teams are just they're moving so fast that they don't they don't have the same cares right and they're like we don't care about cost $7 million from our VC we don't really care about security or compliance you might be thinking wait what they say that like yeah they absolutely say that dude none of them give a about security compliance whatsoever they're just trying to get money they're just trying move quickly, right? And so we specifically moved away from startups because it was like we were spending a lot of time talking to them and they're like, "Oh, we don't care about any of that stuff that you just said." So if you are a startup or a small business and you're like, "Wait, I know we're going to be successful. I'm planning a little bit ahead. I want to make sure we're not caught with the shoe on the wrong foot or whatever and we want to get ahead of compliance security." Ping me. I'll just hook you up, let you play with the account for free. Let it run wild. It's pretty easy to get on. We are also doing We just did a workshop. **[00:48:06]** It's recorded. There's two versions of it. The second one's better, but my Wi-Fi kept cutting out, so sorry. But the workshop is end to end with Massdriver. All the code's open sourced as well. It's building a software factory, if people have heard of those terms, a building a software factory on Massdriver. So you can literally go from no code whatsoever to a functioning software factory running in your cloud account with your own private hosted models, your own private hosted sandbox and then just straight vibe coding from there. So you can once you're done you can just go to GitHub, write a ticket and computer starts building it and it's all built on top of Massdriver and then we've open sourced those bundles and those those questions. So it picks all that up. So you actually start with a very good it's all built on AWS's free tier for the most part. So like almost everything that comes out of the software factory is aiming to put it on like the lowest cheapest compute possible just so teams can get in and play with it. But if you wanted to get in and start playing with like software factories like privately hosting models that's a really great way to start again like hop on our community Slack. Go to massdriver.cloud. There's a community button at the top. Feel free to DM me. I'm online all the time. I need to touch grass but in the meantime ping me on Slack. Armando: Amazing. All right, so I'll put all that in the show notes. Thanks again for coming on the show, Cory. It has been awesome. Cory: Yeah, super fun, man. I appreciate it. Armando: Yeah, it's my pleasure. And everyone that's listening, if you haven't liked and subscribed to the show, please do so. That way, you'll be the first to know when we have an awesome episode with amazing people like Cory here. Thanks everyone, and we'll see you in the next one.