[00:00:28] Jay: Hi, everyone, welcome to the First Customer Podcast. My name's Jay Aigner. Today I'm lucky enough to be joined by Ryan Rusnak. He is the CTO of Airspace, a buddy of mine from 7CTOs. Ryan, the first time we met, you had frosted tips, so explain yourself. how we're, that's how we're gonna start this, just so we set the tone right now

[00:00:45] Ryan: I did. Me and my buddies are in a cover band, and it was his 40th birthday party, and we did a '90s theme. And he goes, "Do you remember what you looked like in the '90s?" I said, "I did." And he said, "You gotta do it." So I was literally on a conference call with-- And my wife dyed my hair while I was on the call.

And it, it was a great concert, and it was so fun

[00:01:03] Jay: I mean, it worked out. But like as I said, by the time I saw you, you were the '90s. Like, that was like the... You had the ti- just the tips left. You looked like, you know, what's his name from NSYNC. you were, it was a great way to meet you for the first time.

All right. where did you grow up, and did that have any impact on you being an entrepreneur and business owner later in life?

[00:01:20] Ryan: I don't know about like the location, but I grew up in Northern Virginia and, I don't think that made me necessarily more entrepreneurial. But, yeah, I'm now in Southern California, and I just, I always wanted to make the world a better place somehow, you know? Turns out the way you can have the most effect is doing something with a small group of people that are equally as motivated as you.

[00:01:40] Jay: that's pretty accurate. so you have a BS in psychology and then you have a master's in computer, human - computer interaction. I saw those and I thought, "What the hell do those look like today versus, like, when you got them?" Right? Like, I mean, I mean, talk about, like, what that degree could mean now with AI and, like, human-computer interaction, compute...

Like, what is it? what was it when you went to school, and does that, do those same kind of, concepts apply now with all the stuff we've seen with AI?

[00:02:07] Ryan: Absolutely. I'll give you the story. I think it applies-- I think HCI applies more now than ever, but I'll give you the kind of story. I wanted to be a doctor, so I majored in biology. No, and my pa- I wanted to major in biology. My parents said, "No, you can't do that. You have to major in business." I went to my first business class, and it was all about how to do, how to use Excel, and I was like, "This is really dumb.

I know how to use Excel. I'm not gonna sit here for 45 minutes a week and get a test on how to use Excel." Switched back to bio, realized I couldn't do any of the dissection stuff, and then I was like, "Well, and I wanna, you know, I wanna help people." So, and I guess I had a bad career advisor because, they were like, "Why not psychology?"

I loved the classes. So I really did enjoy the classes so much. Then you graduate and you realize no one hires a psych person. However, when I was in middle school, I was making like little computer games on TI-82s because I was bored, 'cause I wanted to learn how to do that.

[00:03:01] Jay: Drug Wars? Did you have Drug Wars is all I need to know. Did you have Dru- Okay, and then the Snake and all this stuff. Yeah, all the... Yeah. All right. Sorry. Didn't mean to interrupt. I had to know, 'cause Drug Wars was just the best on the TI-83, or TI-82 Plus

[00:03:12] Ryan: Right. So yeah, I was writing those things and then, you know, in, high school someone needs a website and you realize, "Okay, well, I can write these other things. Why not do that?" So, you know, I was programming and then college I worked at the New Media Center, and I was making websites for professors and little web apps.

You learn more about, "Oh, this one needs a database 'cause it's gotta do some other stuff." And then I graduated and the only thing people wanted to hire me for was doing that stuff. So then I figured, "Okay, I need to get good at this." And I was at Deloitte and I was effectively coding spec. So they had this giant spec for an application, and I built the exact application, and the user was like, "Well, that's not really what I wanted."

And I was like, "I literally built it exactly the way you had written it down." And, they go, "Yeah, well, I don't know. I don't know what to tell you." So I thought the hard part of software turns out is not writing code, it is actually building what the user wants. Turns out there's a whole field of study around this called human-computer interaction.

So I looked up the best human-computer interaction program in the world, and it was Carnegie Mellon. Miraculously, they accepted me, and it was a awesome time and it changed the way I think about software. And now with AI, I think it is even more important that you can close that feedback loop with the user, and you can make things that they actually need very, very quickly.

[00:04:28] Jay: Yep. I love that very much. so tell me about Airspace. Where'd the idea come from? tell me a little bit about your co-founder. and, you know, just what did the kind of initial, you know, phases look like of like, "Hey, we're gonna spin this thing up and we have this problem that we're gonna solve"?

[00:04:43] Ryan: Airspace is a funny story. I was at a point in my life where I was spending most of my time doing government contracting, and I would build things that I was really motivated to build, and then they would just get buried. There'd be administration change of priorities, and it would just all go to nothing.

And that was a pretty depressing way to spend your-- my time on this planet. so I thought I was pretty good at this, and I was looking for something that was really, really hard, like the most challenging thing, the best, you know, way to measure my skills, so to speak. But I also wanted it to matter to the world.

I was sick of making these projects that just got buried by the administration. So, I was kinda looking around and I met Nick, and he told me about time-critical shipping. And I go: "What is time-critical shipping?" And he goes: "Anything in the world that has to travel faster than the integrators, faster than FedEx, so where minutes matter."

I go: "Okay, like what travels on that type of transportation?" And he goes: "These are organs for transplant. When you're stuck on the runway, a lot of times you're waiting for a part for the aircraft that's been time-critical, tissue for research, biopsies. you know, if it's one biopsy, the, if they lose it or it's ruined, you might never know if you have cancer.

It's real high-stakes shipments." And I go: "That sounds like a pretty important industry. The tech must be really good." He goes: "No, it's not." And then we looked into it, and about two-thirds of the time it made it. So about 33% of the time, there was some sort of problem in the supply chain and it didn't make it on time.

So these people, this literally happens where they will be knocked out to get an organ, the surgeon will try to time it perfectly, that driver gets stuck in traffic, no one knows because there's no tech, and that person gets woken up not having received a life-saving organ. So that's the problem that we're trying to solve.

That's the problem I think we've done a very good job solving, and it feels really good. It's a great way to spend time on this planet.

[00:06:36] Jay: Well, that's it. Thanks everyone. We've heard everything. That's all we need to hear. you're saving lives. You're doing it all. no, you're... that I, we talked about, I love, the mission. like anytime, like you said, you have a small group of people, but it's the mission, right? It's like you guys have the thing.

You're like, "This is something we can have an impact on," and you like kinda just did it, and like now you have this thing that you have statistics that you're like actually saving lives and stuff, and it's an interesting like side of the tech space and world because not every company thinks like that, and I'm gonna say most of them don't.

So like, it kinda reminds me of,Martin, Nexleaf Analytics that I met also through 7CTOs. Like they do, you know, vaccine refrigeration temperature technology, and it's like who would do that other than people that are trying to like improve humanity? And that's what you guys are doing as well, and I think it's a really cool thing to have like behind your company to drive you at all times.

So I love that, and it's probably, I would have to imagine, impacts the culture, right? Like you guys kind of do, like you guys bo- has everybody bought in on this mission on the team?

[00:07:38] Ryan: Oh, yeah. I almost feel silly because I thought I was, like, just this crazy person that, like, really cared about, like, how I spend my time and that it matters to the world. And you go to a tech conference and you don't feel that crazy having that thought 'cause a lot of people are talking about Snapchat and are motivated by that, or Twitter, these apps that actually make the world worse, you know?

And, I undervalued how much a mission that matters helps a startup. So we have been able to hire people that I have no business hiring. Certainly in the beginning, like, we couldn't pay people what they were worth, and they still joined because they cared and they wanted to spend their time this way.

so it's been a true blessing having a mission that matters.

[00:08:22] Jay: I love that. who was your first customer?

[00:08:26] Ryan: Our first customer was Panalpina, a big freight forwarder. They're up in Torrance, so it was like a, you know, an hour drive to get there. And, yeah, they were the first one. The story about us closing them was ridiculous.

[00:08:37] Jay: Well, let's hear it. We're here. Let's go

[00:08:39] Ryan: So, one of the first things we did is we kind of made the assumption...

Not made the assumption. We looked at all the problems. So Nick actually was in time-critical shipping, and he had this big spreadsheet of all the things that go wrong in the supply chain, and they were all coded. So we literally, like, sorted the sheet by count and said, "This is our roadmap. Let's make all these problems go away with tech.

Let's wrap these problems with tech so that they can't go wrong." And one of those is that the shipment was misrouted. And when we looked into routing these shipments, it was kind of like the travel agent problem, right? You've got, an organ somewhere in a hospital. It's gotta go to another organ somewhere in another hospital, and there's, like, literally millions of different ways to get there, and humans are using heuristics in their head, and they, you know, can't follow the driver in real time.

And because of those things, they're saying, "Well, you know, the driver usually takes maybe two hours to get to the hospital. it's San Diego, so United's, like, pretty good there, so let's use that," when really, like, it might be a Delta flight in 45 minutes where if the driver catches all the green lights, you make it, you save somebody's life.

[00:09:43] Jay: Mm-hmm.

[00:09:44] Ryan: And that's kinda how it works. So we wrote a router that figured out this problem, the most optimal way to ship these packages. And we wrote the first version, and this was well before we, we baked all the AI into it and had some smart optimizations. So it took about 90 seconds to run. And I don't know when the last time you waited 90 seconds for online, but, I had this big hand-wavy speech about what it was doing.

You know, it's checking all the traffic. It's looking at all the flight combinations. And I made that speech take about 90 seconds. So I go to do the first demo with Panalpina, and, we're in this giant conference room. There's only one guy. There's probably 30 seats, and it's me, Nick, and this guy. And I press the button.

I do my big 90-minute hand-wavy speech, and he goes, "It's done?" I go, "Yeah." He goes, "That's the answer?" I go, "Yeah." And he goes, "If that's right, it's gonna change the world." And I go, "Okay, yeah, it is right." And he goes, "All right, we'll find out." He opens the door and, pretty much goes to his team. He goes, "Hey, everybody get in here.

This nerd thinks his computer is smarter than you." And brings in a 20-person logistics team, and it is their job to do time-critical routing. And they start barking addresses at me, and I'm typing them in as fast as possible. And, you know, I do my hand-wavy speech, and this person meanwhile is doing it, on their own.

As soon I'm talking about how I came up with it, this person goes, "That's wrong. I came up with a different flight." And I go, "What flight was it?" And he shows me, and I could see in the logs that it was in our candidate set and that because of traffic, they would've missed the cutoff time. So I said that, and everybody goes, "Oh."

[00:11:24] Jay: It's like a high school lunchroom. I don't know. You turned it into,quite a sales, I'm guessing you landed that

[00:11:29] Ryan: We landed the deal and, yeah, it was crazy. I'll never forget that meeting for the rest of my life

[00:11:34] Jay: So we were talking about a little bit beforehand. I mean, I, you know, we could bore everyone to death and talk about like AI in a bunch of weird abstract ways. But I mean, you guys have been doing it for a long time, you said 2017. I guess from somebody who's actually been in the space using it for practical real world stuff for that long, where are we at today compared to, you know, when you guys were using it back then, right?

Like, what were you using it for? I mean, I know, I kind of get an idea like back of what, where we were back then, but it certainly wasn't where we are or where we're headed. So like, how are you guys using it to- you know, today versus when you started using it before even, you know, people really were messing around with it much in 2017?

[00:12:14] Ryan: Yeah. It kind of makes me laugh at like when ChatGPT came out, everyone was like, "Oh, that's when AI became a thing." Like that's when AI was invented. And we've been using AI for like a decade before that, and, we were using neural networks. So I think neural networks are so, so powerful for businesses.

And the reason we used it, it's funny, like now everyone's saying, "What could I use AI for? What problem could I use AI for?" And that's so silly 'cause it's backwards. So previously that same router that I was talking about, what was going on was it got very good at finding the absolute fastest flight period.

That might be a flight into Baltimore and then like a hotshot drive down into Virginia. And that's fine if it's gonna save someone's life, but if it's for like a surgery the next day where this is like a cryogenically frozen vein, they want the flight into Dulles. They just, you know, don't try to save them 45 minutes and cost them an extra $1,000.

They want the one that's a little bit slower. Well, that's pretty tough, right? Now it's not an optimization problem. Now it's a learning problem.

[00:13:15] Jay: Yeah. Yep

[00:13:16] Ryan: So we had these different configurations where there's like the absolute fastest one and then there's like the economical one, something in the middle. And what we needed to do was rewrite the entire router because we were kind of building these config files for like what kind of routing the customer wanted.

We didn't really let it learn. So one of our employees was like, "I can do this. Give me like three weeks to rewrite this with neural networks." And what it did is every time the router guessed wrong, we would save that data and then we would, the, a human would reroute and that would signal about what the appropriate route was for that customer.

So we more or less built a neural net around that concept. what route is least likely to be changed by operations? And it just went through the roof in terms of accuracy, where the number of routes that the customer's confirming ops wasn't changing just went, you know, to high 90s, to the point where we thought it was actually like a bug.

We thought using routes that were too slow. We roll, we rolled it back. Ops immediately started having changing more flights, and then we put it right back up in production. So, we've got this great chart where, you can see a dip and then we, you know, lean back on the automation and it stays in the high 90s now.

[00:14:28] Jay: So what, what are you using it for today? What are you using AI for today? Are you guys, are all your, are you know, full Claude code? You guys are, you know, doing code gen stuff and fully agentic? Like, where are you at on the curve of like AI engineering today in your team?

[00:14:43] Ryan: So we use it, it's very, very embedded in our company, because we've been using things like neural nets. You know, we use neural nets to classify drivers, figure out what the best driver is. and now that, Claude Code is out, pretty much when Claude 4.6... yeah, when Claude 4.6 came out and Codex 5.3 came out, we said this is what we need to be using.

This is how we need to be coding. It is so good and, it just writes code faster than anybody else that is very, very, very good. So we're not at full dark factory yet, where the AI is-- We're writing a spec and the AI is, coding and then an AI does the poor review, then it deploys it. But I think we're at a very healthy state,

[00:15:22] Jay: Mm-hmm.

[00:15:23] Ryan: right now, which is everyone is encouraged to use Claude to, get all their stuff done.

So what's really nice is we use MCPs to pull from all the info capture. So all the meetings are recorded, and then can MCP into Notion or Google Drives. PMs can then take those to then create specs in Jira, parse them in epics in Jira, all using Claude. And we're-- So we're building all the epics with Claude, and they're excellent.

Like requirement docs have never been this good. And then the engineers can use MCPs, just reach into Atlassian, to Jira, pull that down, and then start cranking out tickets. They're using Git worktrees to work on multiple things at once. so we are shipping a ton of code to the point where we kind of need to figure out the PR problem because the people that are really relying on AI are producing an egregious number of PRs, and we're not ready to say that's an AI problem yet.

For

That's a great point. That's a-- Yeah

[00:16:22] Jay: I am curious, I mean, obviously I'm a QA guy, so I do have to ask, like, what are you guys thinking for that, right? 'Cause I'm seeing this problem starting to emerge that's like AI-accelerated dev team, shitload of awesome stuff coming down the pipe that like's 10 times what we were seeing before.

QA or whoever has like a company Claude account, and they're like, "Now go test all that." And they're like, "What? what do you mean? We have 10 times the work now?" Like, so yeah, obviously there's levels of maturity with that, but where are you guys on then? How are you planning on like continuously validating this like flood of awesome stuff that we couldn't get done before and deploy it in a way into production that we're confident that you're, you know, that you're gonna be delivering quality code and quality features to your clients?

How are you guys gonna deal with that going forward?

[00:17:11] Ryan: So we're in an environment where if we ship a bug, like it's kind of life or death,

[00:17:14] Jay: Yeah. Right, right

[00:17:16] Ryan: pretty high pressure environment. So because of that, we really want the engineers to own what they ship. So we make sure... Like at some point we actually stopped feature development and just did an engineering supercharge effort, and we made everybody work on automation.

And everyone was so stoked to work on that so we could have confidence in every deploy. And now tests run, tests pass, the automation deploys it right out to prod. We try to put most things behind flippers, so we have a little bit of control over that. engineers QA their own code in multiple environments.

but for the most part, it's all through automation. We really, really lean into automation to make sure that we can just deploy, deploy, deploy with confidence.

[00:17:55] Jay: Is there any part of that, like where there's... I fucking hate the term human in the loop. It's like the dumbest. we gotta come up with something else. This is so stupid. It just means human beings. But like, is there any, like, it does it worry you guys with that level of automate? How do you get that confident?

Like, and is there a point where you're like, "We need somebody to at least open it up on their phone to make sure that like the guy, the delivery guy who's sitting there, you know, it actually opens on his phone?" Like, is that part of this like kind of evolution of your engineering team?

'Cause I'm... So many people are at different parts, I'm just curious like where you guys are on that spectrum.

[00:18:29] Ryan: Yeah. There's two things that we're pretty obsessed about, and the first one is DORA. So how quickly can you fix something that breaks? So we deploy over a dozen times every single day.

[00:18:41] Jay: Mm-hmm.

[00:18:41] Ryan: These deploys are really, really small so that if something does break, it's a quick rollback. Even when we deploy on mobile, we actually prep a rollback version so we can roll that back quickly.

yeah, so we're really, really serious about DORA, and in order to be serious about that, you have to have really high test coverage. And the second thing is we're, we really care about the users. We want them to love our software. So there are some things that are really hard to QA with automation, like maps.

How's the map? How's the map feel? And we just have people, you know, test that out. We'll run simulations. So we'd run these whole warehouse simulations where we send people to go out to get food and get coffee, and in the real life, that person might be carrying a hip replacement or an artificial knee.

But instead they they come back with pizza and coffee. We watch on the map. Everyone's got their roles. So we do a combination of automation and role-playing, to

[00:19:33] Jay: I told you before, that one is my favorite. Like, every time I've seen that in the real world, it's almost like just the fact that everybody has to be involved, like, the test is more valid. It's like everybody kind of is like, "All right, we're doing this thing to like, like mock the field. Like, let's go."

And everybody kind of gets it. And it's a- I love that, "Hey, we're gonna, you know, let's beat it up, while we're walking around and driving around." Like, that's the only way that, like, I would be confident at the end of the day, like when I hand this over to somebody. even if I've run a million automated tests, like you said, the feel and just the user experience, I'm curious when we get there from an AI perspective that it gives us like something human.

Can it... Is it gonna be able to go like, "Is this human likable, yes or no?" Like, are we gonna get there? I don't know. What do you think?

[00:20:19] Ryan: Yeah, I mean, so I think the important thing is the context and the intention, right? So a lot of developers, I feel like developers that are the kind I'm about to describe, won't do very well in the world of AI. And that is when they're testing it, it's I click this button and it rendered this page. You know, I did this thing and it, the expected thing happened.

And when you're try... When you have an intention, right, which is I am trying to get a hip, a knee to someone who needs it, you know, or I have all these drivers that are busy, I need to get another thing on their route. How do I do it? You're looking at the application totally different because now you're trying to achieve a goal, right?

And I think when you can be that close to the user, that's when the application... There might not be any bugs in the application, but the application might not help the user. And that's just as bad. So lots of times when you have developers that can be that close to the user and in their head, you end up with products that people love

[00:21:20] Jay: Love that, dude. Love it very much. how do you... do you take any stock in personal brand? Are you out there trying to build anything up for yourself? Like, are you Blinked? Are you a social media guy? Like, do you see yourself as having any sort of, I don't know, specific kind of thing that you champion or you like talking about?

Like, are you out there in the tech space as a, as the guy or are you just, you're just, you know, crushing the shit out of things in the office and just, like, doing your thing? What is your style?

[00:21:46] Ryan: I mean, the, probably the worst, my worst quality is that, like, I'm allergic to LinkedIn. It is this self-aggrandizing, I don't know, cesspool. I can't stand it.

[00:21:58] Jay: There's a picture of a blue elephant that I'll show, that I'll send to you. That's,

[00:22:02] Ryan: yeah, I have seen that

[00:22:03] Jay: that's very accurate for LinkedIn, I think

[00:22:05] Ryan: Yeah, and maybe it's 'cause I came out of the government space where the people that got ahead were the people that knew how to work the system, and I hate-- I always hated that. So now you see, I mean, companies will let people go and then you'll see someone, you know, say, "Oh, I got promoted into a new opportunity." And it's like, man, we know what happened.

[00:22:25] Jay: Yeah, we saw it. We saw it live. Okay.

[00:22:28] Ryan: Yeah, the, there's the, the public vers- private persona is, is like a marvel of mankind, you know?

[00:22:33] Jay: Do you, I mean, do you try to keep your private stuff private intentionally? Like, are you one of those guys, or are you just kind of whatever's out there is out there and, you know? I guess some people are very, you know, intentional about their public image when they are a co-founder of a business.

They're like, "I gotta look a certain... I gotta say certain stuff 'cause we do this." Like, are, you're out there all day, every day talking about, you know, time critical logistics stuff, like, that's all you talk about? Or are you just kind of Ryan out there doing your thing?

[00:22:59] Ryan: Yeah, I don't know. I guess I, I don't, I should think about it more. I'm sure if I did, I would probably die with more dollars in my pocket. But like in the moment, it doesn't feel like it's worth spending time on. So like I'm the guy at the concert that is not holding their phone up because they're happy to hear the music, you know?

[00:23:14] Jay: My man. I love it. Love it very much. Well, you're a breath of fresh air. like a real human being, just in a world of AI and nerds and tech and just... It's great. Like, I enjoyed your energy when we met. I enjoyed it today. I think you just... I love the way that you view life and the way that you view tech.

And I have one last question. This is not about business. this is just about Ryan being Ryan. If you could do anything on earth and you knew you wouldn't fail, what would it be?

[00:23:42] Ryan: Oh my gosh. I'd probably stop climate change.

[00:23:46] Jay: There

[00:23:46] Ryan: I,

[00:23:46] Jay: Most of the time there's a pause, and there's no, there was no pause. I love that. Most of the time there's like a, "Oh, what do I do?" You were like, "That's it. Let's do it."

[00:23:53] Ryan: I think it's because I know it's gonna be a problem for my kids, and I don't feel like there's anything I can do about it. And right now, because they're three and seven, anytime they have a problem, for the most part, I feel like I can make it better.

So I feel helpless

[00:24:07] Jay: Oh,

[00:24:08] Ryan: in make- in, in protecting their future.

[00:24:10] Jay: Oh man, that hurt, dude. That, that like stuck. I hear you. Okay. I mean, yeah, I... Well, look, in this scenario, you can solve it. You solved it. You couldn't fail, so you solved climate change.

[00:24:22] Ryan: Oh, that feels good.

[00:24:24] Jay: Feels good. I'm glad I could give that to you today. Ryan, if people liked what they heard today, they wanna reach out, they wanna talk to you, like, how do they reach you directly, if they wanna reach out to you?

[00:24:32] Ryan: I mean, people are-- I mean, I would say hit

[00:24:34] Jay: Not LinkedIn. Not LinkedIn I

[00:24:36] Ryan: never, I ever open the application. airspace.com. You can always email me, ryan@airspace.com. I'm always happy to hear from people

[00:24:44] Jay: Beautiful. Well, Ryan, you were awesome, dude. Thank you for being on. we're gonna follow along with the journey and, you know, wish you the best of luck in this insane world and, you know, we'll catch up with you soon. Thank you, buddy. See ya

[00:24:55] Ryan: Thanks, man. Good luck to you as well. Thanks for having me on here

[00:24:57] Jay: See ya 

​