Why Mobile Releases Still Feel Like Chaos, with Runway CEO Gabriel Savit

[00:00:00]

Matt Klein: All right. Welcome to Beyond the Noise Signals, Stories, and Spicy takes the show where we dig into the stories of the people shaping the future of app-based computing with a special focus on mobile. I'm your host, Matt Klein, co-founder and CTO of bitdrift, as well as the founder of Envoy. Each episode we'll talk with engineers, founders, and technical leaders who've transformed the way their companies build and understand what's happening inside their systems.

We'll dig into the challenges, the breakthroughs, the lessons learned, and wrap it all up with their hottest takes. Let's dive in.

Today, I'm thrilled to have Gabriel Savitt, who's a former iOS engineer and now is co-founder and CEO at Runway, a mobile release management platform that helps teams automate, coordinate, and monitor their app releases with extensive experience shipping apps at scale,

Gabriel [00:01:00] developed a love for designing and building awesome mobile experiences and a dislike for the friction and overhead required to actually get app updates out the door. Outside of Mobile, Gabriel is known to travel to countries people haven't heard of, and he spends as much time as possible underwater.

He holds three passports and a very expired pilot's license. Thank you Gabriel, for joining us. That's an awesome intro.

Gabriel Savit: Thanks so much for having me. Um, good to be here.

Matt Klein: You know, it's, it's funny, obviously we're here for a technical talk, but, I am also a scuba diver, although it is long expired. Um, so I, I, I used to do it a lot when I was a younger person, but haven't done it in a while.

But it's good to hear that you saw it out there.

Gabriel Savit: Yeah, I, I had a long break too. I, I started as a kid and, and didn't do it for a while and my partner actually got me back into it and super avid now, so it's never too late. You should get back out there.

Matt Klein: Where do you go now? I mean, I'm sure people are here listening for technical talks, but we might just, you know, talk about scuba diving the whole time.

Gabriel Savit: Sure. I'm, I'm up for that. Um, so, so we actually live [00:02:00] in, um, a small town in Baja, California, in Mexico.

Matt Klein: Oh, fantastic. Okay.

Gabriel Savit: Right. Right by here. There's a, a fairly well-known, uh, marine park that is just one of the best places we've ever dived. It's part of the reason we decided to, to move here and live here.

Um, so spend most of our, our sort of time diving there.

Matt Klein: Fantastic. Um, I promise, this is my last question about scuba diving, but one thing that I'm curious about is, um, at least my experience was also having started when I was a lot younger, is that, there's been a lot of coral reef die off, right?

Gabriel Savit: Mm-hmm.

Matt Klein: Like, and I'm, I'm just curious, um, do you think that, the scuba diving has gotten measurably worse in the time that you've been diving? Or, or I don't... Just tell us briefly about

that.

Gabriel Savit: Yeah, no, for sure. I mean, unfortunately I think it's pretty, pretty widespread. There are areas that are like doing better than others in that regard, but, um...

the great barrier reef. When I was a kid, I grew up in Australia and, uh, dove up there and this was, you know, [00:03:00] already things were deteriorating. But a frame of reference then versus now. Yeah, it's changed dramatically. So yes, this, this stuff that's happening as well.

Matt Klein: Alright,

maybe we'll have to do another whole episode.

Just on scuba diving, but let's, let's, let's get back into the tech talk. So yeah, so you have a career that's really entirely on mobile. Would love to start by just learning about how you, how you got into mobile. I mean, what, what excited you about it? And, and, and tell us how you ultimately end up starting a company around it.

Gabriel Savit: Yeah. Pretty random and kind of meandering path. Um, but let's see, I guess starting at university, I was always interested in sort of engineering and, and technical stuff and, and building things and, and that kind of thing. Um, going all the way back, I actually sort of worked on - or, worked on... did some robotics, Lego robotics, I think I what it's called first or something that league, where you program little robots with this sort of visual language.

Was doing that sort of at an earlier age, but when it came to sort of university and actually sort of higher education sort of... [00:04:00] picking a path, as it were, didn't know what I wanted to do. As you mentioned briefly, like I love, planes and flying pilot's license, so, engineering plus aerospace,

I - I thought I'd sort of try that as a track. As part of that, of course part of that major, you do some compsci, but pretty, pretty basic, pretty limited stuff. So had some exposure there, but obviously wasn't sort of anything like... let alone mobile apps, like any kind of like production, software engineering, let's say.

So really started, when I took time off of university to co-found, my first startup, which was a sort of hyper-local marketplace thing in New York. Took, a bit of time off school to build that with my brother and some, some friends. We weren't technical in the sense that we could build a product, right?

So we either needed to outsource that stuff or figure it out ourselves.

Matt Klein: Sorry, when you say like, hyper-local

marketplace, could you say more? What, what was that about?

Gabriel Savit: Yeah.

This was, this was kind of in the early days, maybe I'm dating myself here, but early days of Groupon. And so the, the kind of, crowdsourcing demand, model, we wanted to kind of flip it and actually start on the demand [00:05:00] side.

So Groupon was trying to like, sort of engender

Matt Klein: Yeah.

Gabriel Savit: Demand by putting out these crazy offers and kind of get foot traffic into businesses that way. We wanted to start on the demand side, so like... In someone's neighborhood, they could hop on this app and sort of... put in a request basically like we're a party of four, we wanna sort of go out for a meal.

What have you got? And so businesses could respond in real time, almost like in a, a chat like way or with automated rules that would sort of take in all that info. Some cool stuff like, you know, you could tie into the weather with APIs pulling the weather. So if you're an ice cream shop, you might sort of increase incentives on a cold day when there's less, you know, less.

Matt Klein: Yeah. Makes sense

Gabriel Savit: folks in

Matt Klein: makes sense.

Gabriel Savit: Your business. So yeah, it was that sort of like two-sided marketplace, but starting on a demand side. Um, and, and tried to scale that out. It... didn't go anywhere ultimately, but it was the, the reason why I sort of first got into to mobile development. I worked on the iOS side.

My brother did Android, and that sort of divide has, has stuck actually to this day. Really picked it up on the fly. And there was a whole lot of learning to - to do, of course. We had some [00:06:00] contractors helping, but ultimately, I think a lot of folks know how this works. It's, it's tough, especially with a startup to like get stuff done... right,

and quickly. Um, so kind of took over that code base and just learned by getting in there.

Matt Klein: Yeah.

Gabriel Savit: And uh, playing around.

Matt Klein: Were

you, were you entirely self-funded? I mean like how did you, how did you do it?

Gabriel Savit: That was like friends and family...

Matt Klein: yeah.

Gabriel Savit: Sort of small, small, scrappy deal. You know, we were kids, it's a very different, different journey than, than now with Runway, of course.

Um, but very, very scrappy, had a lot of fun. And, and yeah,

Matt Klein: I also, I also started a company when I was around 24 or 25, and it was a total failure. Um, but it, it is interesting because I feel like you learn actually a lot from that experience. And I, I guess my first question for you would be what, you know, what were your big learnings from that time that I'm sure you take forward to Runway now?

I mean, I'm, I'm sure there's stuff that you learned then that you said, ah, I'm not gonna do that again, or I will do that again.

Gabriel Savit: For sure. Yeah. I mean, first of all, like tying back to the, the previous point, like [00:07:00] concretely, like that's what sort of taught me, or, or, you know, brought me in the direction of software engineering as a sort of production practice.

And mobile of course. So like concrete learnings more on the like startup business side. Um, a a ton of things, you know, uh, in my role at Runway now, I am rarely touching code. I try to, but I sort of sneak it in. But doing a lot more on the business side. Everything I'm doing now and everything sort of, I started...

knowing on the business side of sales side, I definitely learned aspects way back then.

Matt Klein: Yep. Sure.

Gabriel Savit: I didn't know it at the time. It was a very different kind of sale. We were going door to door to businesses in New York, which definitely, definitely throws you into the deep end and, and you learn some stuff there.

But a lot of the discipline, a lot of the sort of approach and, and ways of working with folks and understanding, um, how... you bring any kind of market, excuse me, product to market. A ton of learnings there. Um, a lot of learnings around people and building teams. I think very, I

Matt Klein: was gonna, yeah...

Gabriel Savit: very small team at the time.

Matt Klein: ...say that I, I, I, I feel like [00:08:00] my biggest learning from that time was... Your, your co-founders matter. Your, your people matter. It all, it all matters a lot.

Gabriel Savit: Yeah.

Matt Klein: Sorry. Keep, keep going. Yeah.

Gabriel Savit: Yeah, no, a - a huge amount and, and that learning, not even from a, coming from a place of like, things went horribly wrong with co-founders or the team, but you start understanding kind of like how you can really set up for success with the right people and, and the, the kind of effect that has both good and bad.

So, yeah, just, just a ton coming out of that. But again, you know, we were, we were kids at the time, this is sort of

Matt Klein: Yeah, for sure.

Gabriel Savit: Middle of college. So a lot of just learning in general... and super fun of course, but eventually wound that down, went back, finished my degree... still aerospace. And then sort of the next, formative step, I guess in the journey was

working on my senior thesis project with a group, we decided to build a sort of small scale air traffic control system with an unmanned plane, a drone, and back then

Matt Klein: Oh wow.

Gabriel Savit: Drones weren't even like a

Matt Klein: Yeah.

Gabriel Savit: A thing in the way they are [00:09:00] now. So it was really just, you know, one of these hobbyist, like RC planes.

And we fitted it with a sort of onboard flight system. So rather than you control it from the ground, gave it a sort of central nervous system and then tried to mimic the interactions that normal planes have with air traffic control. So you could sort of give it directions for places to go in the sky and, and, you know, altitude and, and, and stuff like that.

And it would kind of figure, figure out how to get there versus direct control.

Matt Klein: And, and, and, sorry, just so I understand, because it's interesting, you would program it before it took off or there was like a, there was like a remote component to it where you could give it like an air traffic controller would, you would give it instructions while it was flying around

Gabriel Savit: Exactly.

The latter. So yeah.

Matt Klein: Okay.

Gabriel Savit: It was still a remote component, but it wasn't direct control. So you would feed it instructions in sort of a, a standardized way.

Matt Klein: Right.

Gabriel Savit: Um, it would interpret it and it would, you know, take the plane where it needed to go. Um. This isn't like groundbreaking stuff, but, but at the time we thought it was cool and there were some interesting aspects that kind of [00:10:00] our advisor, uh, encouraged us to, to explore.

But concretely we used a, an iPhone as the flight computer. So that's what we kind of just chucked on the plane. That was the central nervous system, and...

Matt Klein: how did you,

Gabriel Savit: course someone had to, yeah.

Matt Klein: I mean I, I'm, I'm just really interested now, how did you hook the iPhone up to the aircraft controls?

Gabriel Savit: Yeah, there was a, I mean, I barely remember the details here, but there was like a raspberry pie type thing that plugged into the iPhone.

There was a whole lot of like,

Matt Klein: yeah,

right, right. Okay. Yeah.

Gabriel Savit: Rigging that needed to happen there to get all the, the pins and the ports. Right. But, um, but yeah, it was basically talking to a, I guess were they called like Arduino or something?

Matt Klein: Arduino,

Gabriel Savit: yeah. It's like pretty similar, yeah.

Something sitting in the building,

Matt Klein: but like you had the phone talk to it over like a USB port or something, and then that would give it the instructions and then it would do all of the flight control programming and stuff like that.

Gabriel Savit: Exactly. Yeah.

Matt Klein: That's fun.

Gabriel Savit: Yeah.

Matt Klein: All right.

Gabriel Savit: So, so yeah, it was, uh, it was a cool, cool con contraption. It admittedly didn't fly so well. Um, but, uh, but [00:11:00] we did get a, a paper or two published, um, about it, which is kind of crazy. And, that was my like, yeah, next formative step in terms of mobile development. So I was the one that was building out the, the app that was running on board to power the flight control system.

Matt Klein: Yep.

Gabriel Savit: Um. Super hacky and, and looking back, it was really funny. I, I, you know, had very little idea of what I was doing. I think I had it running on a... just like piggybacked off some sort of boilerplate code from, um, I forget what it's called, some like CD framework, like gaming stuff that would just like kind of run it and it was kind of doing this stuff as part of that loop.

So super, super hacky, but that was the next, you know, mobile related project. At the time I was actually still working on a side project app with my brother called Tab. It still exists. It is our internal Guinea pig. Uh, at Runway. We use it to, to test a lot of stuff in, in production or pseudo production...

Matt Klein: it's actually funny that, that, that you say that because I, I feel like at bitdrift. Our main problem is that we don't have an app, so we can't, we can't really dog [00:12:00] food the system, you know, and it's something that we've talked about a lot, like I realized that we're skipping ahead to, to the companies.

Um, but we have talked about doing an app. Just so that we have something that we can fully test the system on. Um, but it's obviously very nice that you already had an app.

Gabriel Savit: If you guys wanna borrow Tab. It is, you know peak guinea pig at this point so you can get in there we'll...

Matt Klein: we should, we should, we should talk.

Gabriel Savit: ...give you some access.

Matt Klein: Yeah, we should talk offline actually. That, that would be fun actually.

Gabriel Savit: For sure. Yeah. Yeah.

Matt Klein: All right. Anyway, so, so, sorry. So you made that app, and that app does what, exactly?

Gabriel Savit: Yeah. This was sort of in that weird transition period where we're figuring out like the, the previous company, are we ending it?

What's happening? Uh,

Matt Klein: yeah,

Gabriel Savit: we were sort of like... it was losing steam. And, uh, I don't wanna go too deep down, down the story- storytelling here, but, we were in New York at the time working the company and, and you know, a lot of our friends were in the city. This was over a summer period, uh, doing internships and stuff.

So a lot of group dinners, a lot of like, you know, going out and, uh... struggling [00:13:00] to. Split the bill at the end of those group dinners.

Matt Klein: Oh yes. Okay.

Right, right, right.

Gabriel Savit: So very much, especially in that demographic, we, we found later this is a more global thing.

Matt Klein: Absolutely.

Gabriel Savit: It's very cultural too, it depends. But in that demographic, you know, poor college students, um, you wanna split the bill fairly.

Some folks drink, some folks don't. Some folks order a ton of food, some don't.

Matt Klein: Yeah.

Gabriel Savit: Um, so we sort of started towing around this idea of like OCR reading a receipt

Matt Klein: Yes.

Gabriel Savit: Itemize it.

Matt Klein: Right.

Gabriel Savit: Let folks sort of assign items and, and do a, a fairer bill splitting.

Matt Klein: Right.

Gabriel Savit: And so yeah, in this transition period, stuff was winding down

at the previous company we spent like a day just like hacking on this and it, it kind of worked pretty well. And, um, so we decided to launch it and it actually took off like pretty organically. We still have... even though it's super neglected, still have a, a pretty healthy user base on it. Folks kind of across the US a little bit international, um, using it.

And so, so yeah, it's basically like spill- uh, bill splitting on, on steroids. There's a lot you can do.

Matt Klein: Okay.

Gabriel Savit: Sort of multiplayer and

Matt Klein: awesome

Gabriel Savit: OCR based stuff.

Matt Klein: And, and, and you said that it's [00:14:00] neglected. Do, do you still work on it at all? I mean, or is it It is just kind of done. Yeah.

Gabriel Savit: Like zero. Yeah. It's is really maintenance mode.

And again, Guinea pig for, for Runway. It's, it's a shame. I'd love if someone kind of took it over 'cause, 'cause some folks really love it. Um, but...

Matt Klein: right. But, but like when you test Runway and do releases, you just like make a comment change and then like deploy it. Like how does that work?

Gabriel Savit: Yeah,

yeah. No, exactly that.

It's very, very small, very small changes. Um. Which is, uh, well this is getting like further down the line, something that some teams do actually too. Like the, the whole world of like ASO, right? And app store optimization. Like, depending on your team and what your typical diff looks like and whatever, there's some teams that will ship like zero change versions just to get new versions out.

Matt Klein: Interesting, okay.

Gabriel Savit: Which is a bit of a tangent, but that's like a thing, but... oh yeah, nothing is happening in a Tab codebase. There's a lot of read me changes, for sure.

Yeah.

Matt Klein: Okay. All right. So you did Tab and then take us forward further.

Gabriel Savit: Yeah. So Tab kinda happening on the side. [00:15:00] Um, senior thesis project, the UAV, iPhone flight system.

At that point, I guess I had like, you know, some baseline of experience with, with mobile and specifically iOS dev. And so really just, just happened into a full-time role after, after school. You know, I liked it. I, I realized I probably didn't wanna work full-time in aerospace engineering, it's a very different kind of engineering.

The, the work environment, the kind of work you do is so, so different. I had a friend who worked at... who worked at Boeing. He was telling me about what he was doing. He graduated maybe a year before. She was working on a tiny, tiny piece of the flap track fairing, it's called for the 787, which is like a little piece of the wing in one spot.

Just like a joint.

Matt Klein: Yeah.

Gabriel Savit: And he was spending all his time just like on some, some design stuff for that piece. And so you really lose the... the sense of the bigger picture, the sense of like iterating and actually sort of like shipping stuff, which is very, for me, very gratifying and part of the reason why...

Matt Klein: yeah. Well,

Gabriel Savit: the mobile stuff was very rewarding.

Matt Klein: Well, I was gonna say that. And if you mess up when you're working on an airplane, the [00:16:00] implications are, are a lot larger typically than the kind of stuff that you and I work on.

Gabriel Savit: Absolutely. Yeah. Getting into the sort of quality and and monitoring side of things.

Yeah, for sure. Yeah.

Matt Klein: Yeah.

Gabriel Savit: Different stakes. Um, really, yeah, just like slow product cycles. It's a very different kind of work.

Matt Klein: Sure.

Gabriel Savit: So kind of realized early, didn't wanna work in that necessarily. Um, liked the mobile stuff. And, another friend who graduated again before me having, you know, I took time off... had started at Rent the Runway, which is a company maybe a lot of folks have heard of.

Matt Klein: Yeah, yeah.

Gabriel Savit: I think he was their first, one of their first, mobile hires. So they were just, I think it wasn't launched yet. They- he was just spinning up their iOS app, which obviously became, or still is, I guess a huge part of their business. So he reached out. Let me know they were hiring to, to grow that team.

And I thought it sounded cool. Just interviewed and... and started there. So that was the first, I mean my first, you know, job- proper job and, and first foray into, you know, real production, iOS engineering. And I quickly realized how little I knew. 'cause like when you come in, sort of like hacking on stuff and, and picking up as you [00:17:00] go.

Matt Klein: Yeah, absolutely.

Gabriel Savit: And missing,

Matt Klein: yeah,

Gabriel Savit: missing a lot of context and understanding. So learned a ton. The whole team was learning a ton. So it was actually an interesting environment, right? It's not like coming into a, I don't know, like a Google or some FANG job. Um, there was a bit of push and pull, right?

'cause the team was young, the product was young. I was also learning. So it was a very, it was a fun environment. We had a lot of... freedom and agency to do a lot, especially on the product side. And this is what really drew me there as well. And this is part of the sort of story leading into Runway.

Very much like a, a product engineer, I wasn't super excited about like tooling, build systems, lower level stuff. Like I really just liked sort of building, building UIs, building product and, and sort of experiences and, and getting them shipped and seeing how users sort of respond and, and iterating from there.

So it was a great role for, for that. At the same time as the team was growing, quite rapidly at the time... the, the needs. On the platform side were increasing and the needs around like tooling and process.

Matt Klein: Mm-hmm.

Gabriel Savit: And CI [00:18:00] of course started out without even having CI, which nowadays seems crazy. That was all creeping in and, and that's where I think sort of like...

My, my transition or my, my broadening of like understanding I guess, of this whole process and like how, how mobile works really end to end was growing. As well as my understanding of like the, the frustration and the fact that I actually didn't like a lot of that stuff.

Matt Klein: Yeah, I, you know, I've, I've talked about this with a couple of different guests and, and one thing that I think is interesting in this space, which is probably directly related to your company, so it's a good segue, is I feel like Apple and Google, you know...

they do an amazing job of getting the most apps as possible into the app store. Right? I mean, it's like that's I to, to me that's what they want. I mean, that, that, that part makes sense, but when it comes to, you know, what growing teams need, right?

Gabriel Savit: Mm-hmm.

Matt Klein: In terms of whether it be build tooling or IDE support or release management or observability or whatever, [00:19:00] it's, it's pretty lacking, honestly.

Gabriel Savit: Yeah.

Matt Klein: And, and. You know, prior to telling us about Runway, I, I'd love to understand from you, why do you, why do you think that Apple and Google don't help more with some of these things?

Gabriel Savit: It's a great question. I think it's one that we ask ourselves a lot. It's a little bit of, of... well, there's, there's a lot of facets.

I think, and this has changed over time. I don't know. The, one of the things that comes to mind is like, you look at Apple nowadays and, and their business is in, I mean, it, it's always kind of been, hardware's a big piece of it, right? Like that's a big focus. Also media, like, like they're a, they're a media company now.

Like there's so much else going on. This feels like something that's just so far down the pecking order... things work okay. And this gets to a broader point that we talk about a lot.

Matt Klein: Sure.

Gabriel Savit: Just like the status quo of people accept it. That's how it is.

Matt Klein: Yes.

Gabriel Savit: Feels like a bit of that they have so much else going on.

This is, this is just not [00:20:00] high on the priority lists. Um, the same on the Google side. Like it's... Apple's come a long way APIs wise, like their support for sort of like tooling around App Store Connect, is in a much better place. So while that's not Apple themselves investing too much there, like at least they're kind of unlocking that, that opportunity to make people's lives easier when it comes to that stuff.

Matt Klein: Yeah, sure.

Gabriel Savit: Google is, is lagging even further there. Like the APIs are just a miss. They're not sort of shipping new stuff there. Google, you could say the same. They have a whole lot else going on.

Matt Klein: Of course.

Gabriel Savit: I think this is just down, down the pecking order, like things work and they at their level probably are not seeing a huge, you know, downside to not focusing more on...

Matt Klein: yeah.

It, it's, it, it's mostly, I just find it interesting I guess from a business perspective that... pretty much every larger app, you know, has to solve a lot of fundamental problems, right?

Gabriel Savit: Mm-hmm.

Matt Klein: Like whether it be, again, around CI, dev tooling, IDE like all the things that we talked about. Um, and it just seems like a lot of these things... get solved [00:21:00] over and over and over again.

It also provides a lot of opportunity for our types of companies to come in and do interesting things, so that's fine. Um, I just think it's interesting, right? That, that there are these holes in the ecosystem that basically have to be filled either by companies or by large companies with large apps.

Gabriel Savit: Yep.

Yeah, no, for sure. I think the other thing I, not to like preempt this part of the conversation too early, but of course like just AI and the... the confusion. That's like coming with all this froth now. I think it's like even more than ever, like taking people's attention elsewhere.

Matt Klein: Yeah

Gabriel Savit: this stuff is not gonna go away.

It's obviously changing.

Matt Klein: No, makes sense.

Gabriel Savit: Changing...

Matt Klein: Yeah.

Gabriel Savit: But it's not going away and yet there's just so much distraction.

Matt Klein: Yeah, we can, we can definitely talk about AI in a bit. Um, but I, I do wanna get into how you founded Runway. So could, could you tell us just a little bit about what, what led you to that?

What were the challenges that you were seeing? Yeah, that would be fantastic

Gabriel Savit: For sure. Yeah. It took a while, so, [00:22:00] yeah, I, I worked at Rent Runway for quite a while, through a ton of growth, through a ton of sort of changes for the team. Like I said, maturing process, maturing tooling... and started noticing a pattern, I won't even say, I started noticing it actually took a, a step back, with some of my colleagues there to, to kind of have it click and realize that this was a thing.

But... in hindsight, looking back, there was so much thrash around process and tooling and how the team coordinated around releases. Um, we were wanting to ship product frequently. We had a pretty good release cadence. There was this rotating cast of, I feel like, you know, EMs or, or maybe a, a PM would come in and sort of realize, oh, like this release process is amiss.

There's a lot of noise. We're having kind of a lot of issues and, and constant hot fixes. Like, let's put some guardrails in, let's do better. And they'd... just come up with a new checklist. And so we had these long runbooks like release runbooks, right?

Matt Klein: Yeah.

Gabriel Savit: New checklist. Okay, great. You know, static checklist folks aren't, aren't following [00:23:00] it.

There's no audit trail. You don't know what's actually happened when, so you don't know that things are actually being executed in the kind of nice, perfectly designed... system. Um, a ton of just like back and forth on Slack, trying to get people on the same page, especially at a product forward company.

There's a lot of stuff launching and features and, just so much back and forth and coordination and, and realizing over time how much bandwidth this was taking. Especially when we had a release rotation, which is a common model. So folks will take turns running point on a release. That became this thing where I and many of our other engineers would sort of dread your week or two when you were running a release

'cause it was just mayhem. You couldn't get product work done. It was just not fun. And this started adding up and realizing like, why, why is it, why does it have to be like this? Why do we keep just like refreshing a checklist and nothing really changes? There has to be... you know, a better way to, to do this, to put together all these pieces.

Matt Klein: Yeah.

Gabriel Savit: And run a, a saner process here. But again, like I said, when you're in it, you don't have that [00:24:00] distance. We weren't sort of taking a step back and realizing like, there's an opportunity here. It's just like a... this is the status quo. It definitely sucks. We're realizing this. You can see it because these people come in and, and for a moment decide, oh, let's try something else...

it doesn't go anywhere. But then the stars sort of aligned... COVID happened, Rent the Runway, had some layoffs, some furloughs. We all, in different ways had time to sort of take a break. Take a step back, and this was one of admittedly a few ideas that we were tossing around sort of a core group of the, the mobile team at the time.

We were very close friends. We knew we had this idea, we wanted to start something together. We didn't know what. And we... as things were sort of on, on pause and quieting down during, during COVID, we started meeting more and more regularly to discuss just like ideas, like stuff that might be interesting to work on...

Matt Klein: meeting

Gabriel Savit: and decide...

Matt Klein: in, in person or on Zoom?

Gabriel Savit: Zoom, yeah.

Matt Klein: Okay.

Gabriel Savit: So yeah, we were, we were all spread out, uh, at the time, so not in person. We, we started a doc and we had a bunch of ideas in there, and very early on, one of the other co-founders [00:25:00] put down this idea of a release management platform and... It's interesting. We, we still have this doc and we go back sometimes and look at it.

We also sort of have this as part of a presentation. We give new hires to sort of talk a bit about the origin story and just like see so clearly that we all gravitated to that idea. There's a lot of comments there. We were talking about it a lot, right? And so, very quickly realized with that distance and taking a step back like this is, this is a problem area that we sort of experienced that all the teams we've been a part of...

Matt Klein: yeah,

Gabriel Savit: different shapes and sizes, of course. But there was, there was an opportunity here, like why, why deal with the status quo? We, that was the other sort of key theme is like at all these teams, there was some understanding, some recognition in the backs of people's heads at least that like... this was a really sticky part of their overall like development process.

Um, but this inertia or this, this idea that just like this is the way that you ship mobile.

Matt Klein: Yeah.

Gabriel Savit: Um, it's hard to deal with it. So we really try to engage with that and, and understand like is there a... is there an [00:26:00] opportunity here? Can it be sort of actually made better in a meaningful way? We had a lot of conversations with friends and connections at other companies just to validate as well our, our experiences at that point.

And, for the most part had our, our ideas sort of confirmed. Definitely a range of sort of responses there, but for the most part, teams of different shapes and sizes, they sort of rated their release process as quite inefficient and full of noise and, and that wasn't great.

Matt Klein: Yeah.

Gabriel Savit: And that's how the, how the thing worked.

Um, and so it all kind of, kind of went from there.

Matt Klein: Yeah. Yeah. I mean, I, I guess I'd love to dig in more, just because I think the people that are, you know, that are out there will be interested is, I mean, can you take us through a little bit, like what, what is the typical release process, right? It's like... what are, what are the things?

I mean, because I'm sure the product, I mean, because you've been doing this for a while, so I'm sure the product has evolved and you started with an initial set of features. So just like I, I'd love to dig in a bit more on like what were the first pain points that you were trying to [00:27:00] solve and then I guess like how has the product evolved over time?

And, you know, I guess, you know, these are many questions in one, I apologize. But just kind of looking at... what are the biggest problems, you know, that you think that developers are facing in this space right now?

Gabriel Savit: Yeah. Yeah. There's a lot to the question.

Matt Klein: Yeah. Yeah. Go ahead. Yep.

Gabriel Savit: We saw, we saw kind of the problem space as quite broad from the beginning.

And, and this actually made it hard in the early days of kind of building the, the product, certainly and even sort of building the company. Some, some parts of our initial kind of like wire frames and mockups look very similar to how they are today. We had this idea of a, a really comprehensive end-to-end release cycle.

A lot of people think of releases and release automation, and their minds immediately go to CI, just like the build process. But releases start well before that. There's a lot of git wrangling with like branch cuts, cherry picks if fixes come in. Just like getting on the same page about what is going to be shipping the sort of readiness or preparedness of feature work both across [00:28:00] your, your code, but also project management

'cause PMs are looking over there. There was always disagreements there. So starting well before the build process to sort of like kick off, feature readiness, sort of like, coordination that happens well before even your first RC. Transitioning through, of course, to your first, your first RC and, and very often like multiple, as you go through that kind of QA regression testing process, maybe you need to sort of... get some fixes in and spin up new, new builds, sort of managing that process.

Any distribution, pre-production, so internal distribution, alphas, betas, different kinds of ways teams are like, testing internal builds is an important stage, all the way through to, depending on the team you're on, there's a sort of sign off process potentially depending if you're in like a, a sort of regulated, vertical, whatever, collecting these sort of sign offs.

A lot of teams do go / no go meetings. There's sort of a, a stage for that.

Matt Klein: Yep.

Gabriel Savit: And of course eventually getting everything actually packaged up into the stores. Submitted, released and rolled out, and rollout is a whole domain unto itself. So, we really saw the pain point [00:29:00] as, as all these pieces put together, all these stakeholders involved, weighing in at different points in time, balls getting dropped, which slows everything down a cascade of like delays when any given part of the process isn't tight, isn't going well.

So we started quite big with the, the vision. And that made it hard early to, to get started. And we went through YC and, and sort of our, our partners there were, were a little bit at odds with us. Like their whole motto is like, do something small, get it shipped, get some folks using and paying for it.

It was hard to fill in this big vision and, and we felt there was a really high bar to delivering value 'cause a lot of the value is in connecting all these pieces.

Matt Klein: Yeah. You need all the

integrations and all the pieces. Yeah. I can totally see how it, it would be hard to find a, a scoped thing. Anyway, keep going.

Gabriel Savit: Exactly. Yeah.

Yeah. I mean, so much of the, the ROI and the value is in the orchestration across these pieces, right? So, um. Hopefully, I'm still sort of answering your question.

Matt Klein: No, you are, go ahead. Yeah.

Gabriel Savit: We started broad, we certainly picked off a few key... [00:30:00] places to start. Certain automations scheduled branch cuts, for example.

It was like small stuff. Labeling tickets in Jira. There's this whole process. For example, if you use a product management tool, like you wanna label work, so you know, like when stuff shipped,

Matt Klein: right.

Gabriel Savit: Little automation there, of course, integrations to back it up, we needed to prioritize sort of based on what we knew was being used by the majority of our kind of ICP, our, our core customer.

So we did pick off small pieces, but going back again, like compared to wireframes, we had since the beginning this like high level, end-to-end sort of framework that we wanted to, to sort of deliver teams. And we just kind of started there and, and chipped away at it. And, you know, first few... I won't even say customers.

We had like a few teams we were connected to or, or past teams of ours, that were the first to kind of test drive the, the product and, and it was a huge leap of faith for them and honestly, more... Probably more time spent than saved at that stage 'cause there was so, so little in there...

Matt Klein: yeah.

Gabriel Savit: Kind of connecting everything and there wasn't much depth.

But at the same time, they needed to get their teams bought in to try this new [00:31:00] platform. And kind of the, the process change and culture change that comes with that. So, so yeah, early days it was broad and shallow and very hard to actually

Matt Klein: Yeah.

Gabriel Savit: You know, realize the value. We, we knew we could at that point, we had conviction we could bring to the whole end-to-end... process.

Matt Klein: And then I, I, I guess there, there are particular, things that I would love to ask about more, but prior to that, I guess... I mean, how have you evolved it over time, right? Like, I would imagine that you've had to pick and choose, in terms of filling in the different pieces of the wireframe as you called it.

And I, I feel like, and again, I have my own bias in this area for sure, which I can ask about... but, I would imagine that when it comes to shipping.... there's so many components to it, I guess is what I would say, right?

Gabriel Savit: Yeah.

Matt Klein: I mean obviously there's the rollout components, there's the monitoring observability components,

there's the experimentation components.

Gabriel Savit: Yeah,

Matt Klein: I, I can imagine there are so many areas that you could get into or that customers would be driving you towards. So... again, before I start showing [00:32:00] my bias, I, I'd love to just learn a little more from you around the...

Gabriel Savit: yeah.

Matt Klein: The vision makes total sense. I'm just curious, like where are the customers driving you towards now the most?

Like where are the problems that that, that like they're trying to face?

Gabriel Savit: Yeah. Um, everything all at once. We, we work really closely with, with a lot of our customers, and these are some of the, you know, biggest and best mobile teams in the world. They come to us with great ideas. We wanna make sure those are heard.

And we've been lucky, I would say, in that like, so far a lot has aligned with how we originally saw things or... or we understand kind of where they fit in.

Matt Klein: Yep.

Gabriel Savit: So, so a lot people are always very impressed, like how much we've done with, with such a small team. Maybe to answer you more directly, I think a lot of the early focus was, well, first of all, prioritizing integrations.

Like we want to support any given kind of tool a team is using. When I say sort of shallowness or or depth in the product, we weren't thinking in terms of like, we wanna start vertically [00:33:00] integrating, we wanna start offering our own CI uh, we wanna start our offering, our own monitoring, uh, tool, for example.

Matt Klein: Yeah. Yeah. Makes sense.

Gabriel Savit: It was very, it was very much allowing teams to plug and play, bring your own tools. You shouldn't be locked in. And, and this, this coordination problem has nothing to do with specific kind of pieces of that tool chain, it's about connecting them, automating things across those pieces.

And so from that lens, a lot of our early prioritization was actually, maybe this is like reducing it sort of... or boiling it down too much, but like prioritizing integrations, 'cause the plug and play integration layers is super key. So we did a lot to kind of survey teams for any given integration point.

And these have been expanding over time. But the early ones, like version control, project management tools, of course the stores, but even then there's, there's more than just Google Play and, and uh,

Matt Klein: yeah,

Gabriel Savit: and the app store. Making sure we are prioritizing there. That's sort of a early days a big specific ask is like, okay, Runway has this, this area of the platform where you can help us manage and automate stuff around, sort of project management [00:34:00] work...

my team uses this tool. You don't support it right now, we want to get that in there. So that, that was a big part of the, the early prioritization, I guess. And then the other part of the, the sort of answer is probably more to do is like, which domains, which parts of the process we, we focused on first?

Started early with with store things. One of the early sort of North Stars was, was we don't want you having to log in apps or connect or play console all the time, or a lot of teams have folks across different apps and their switching tabs and, and logging in and out of different store accounts.

Matt Klein: Yeah, so you can pull that, pull that data out and show it and uh...

Gabriel Savit: get that out, build an abstraction layer. Yeah. So you're not having to deal with, honestly, what are pretty subpar platforms. So keep folks out of there as much as possible was an early kind of North Star, so a lot in the sort of app store related integrations, and a lot in the sort of git wrangling project management area.

We wanted to give teams a clearer picture of... what's shipping, when, aligned product and engineering. 'Cause sort of the interdisciplinary aspect is a big theme for us as well. Those were maybe some of the early focuses.

Matt Klein: Yeah.

Gabriel Savit: [00:35:00] Um. If that makes sense. Yeah.

Matt Klein: Yeah. I, I, I mean, I think it's a, it's a, it's a very real problem.

I, I'm, I'm sure you know that, that you're finding a lot of interesting customers. I, I think, again, I'll show my bias now. I think part of it that I think is really interesting is... and we hear this also when we talk to customers, is... there's obviously this intersection of you do a release, but you wanna know if it's healthy, right?

Gabriel Savit: Mm-hmm.

Matt Klein: And that is not just the release itself, but it's also experiments, features

Gabriel Savit: mm-hmm

Matt Klein: within the release. So I I, I'd love to actually learn a little more I guess maybe I have two different questions. Again, I apologize for asking multiple questions at once.

Gabriel Savit: No worries.

Matt Klein: Um, but when you talk about, you know, first the integrations for the sign-offs and all of those things.

One area which you might already support, I don't know, is I would imagine that there can be checklists for individual features. Like, like, you know, is this feature flagged? Or...

Gabriel Savit: mm-hmm.

Matt Klein: You know, do you have analytics about this? Or something along those lines. [00:36:00] So I'm just curious to learn more, like, has that come up?

Because to me, at least from what I've seen. It's a, it's a very real problem that like, stuff gets shipped and you realize after the fact, 'well, uh, oops. Like I, I don't, I don't have the right feature flag, I don't have the right analytics.'

Gabriel Savit: Yeah.

Matt Klein: And then you're back doing like another cut...

Gabriel Savit: yeah.

Matt Klein: So I'm just curious like, has that come up as something that people want help, at least like from an organizational perspective?

Gabriel Savit: Yeah. Yeah. It's an interesting question. Certainly flavors of that, I think they're kind of different altitudes. We definitely see some teams like get pretty... in the weeds with how they structure kind of checklists. And, and these aren't just like checklists for humans to run through. You can create these kind of gates in, in Runway, use our API and web hooks to interact with them programmatically.

And so teams can get pretty fancy there to build just the kind of

Matt Klein: right.

Gabriel Savit: Uh. Flow that they want based on outside signals. That's, that's maybe part of what you're asking about. But, but yeah, more broadly I think it's [00:37:00] the... this sort of, combination of these different, these different sources of data, these different sort of states that things can be in.

You talked about sort of like the feature flags, that's a whole other mess on top of just your binary release cycles.

Matt Klein: ab- absolutely. Yeah.

Gabriel Savit: Um, the effect of the relationship that has on app health, sort of like the... you need to transpose a lot of information in your day-to-day across these tools to, to make sense of it.

And I think that's where a lot of the, the friction and struggle that we saw came in. So again, a little bit more of like, can we, can we pull together this information, make it useful in, in concert for, for folks, potentially automate some stuff based on it. We have like rollout automations that look at like signals from health and, and stuff.

But more about combining these signals in one place and being smart about... helping folks understand it versus they need to go out to these different places and kind of self-serve the info, mentally, like map it. A, a key maybe to make it concrete. Like one example again, going to feature flags... you go to a feature flagging tool, it's like a [00:38:00] big long list, a mess of just like flags, super hard to understand what's on, what's off, what's up with different audiences or or rollouts of flags. And, god forbid you wanna sort of understand someone somewhere on your team has made a change and that's actually broken some checkout flow or some key path, like see... key conversion in your app.

It currently or has been really hard to just like figure that out. And so Runway has this hub where we are already looking at all these signals of health during a rollout. It could be folks instrumented stuff in bitdrift. It could be... other crash reporting tools, product analytics. This is a whole other one I want to get into

Matt Klein: right,

yeah.

Gabriel Savit: Every, everything you should be looking at to just understand and know the app is healthy. Pull that in. Teams configure metrics, thresholds, there are ways to sort of automate like the, you know, green, red sort of model of how things are going. And then layer on top of that information from feature flagging, so Runway can integrate with flagging platforms.

And we, we can run this sort of continuous diff on your flags. So any changes, not just on off, but if you've rolled a [00:39:00] flag out more, uh, change an audience, we sort of build this diff sort of mark that moment in time. And then there's some views in Runway where we overlay that on all of these health metrics.

And so if you see... key conversion or obviously crash free rate is the obvious one, but you talk a lot about how that's, you know, not enough. Any key metric you've defined, you can see sort of if there's a deviation,

Matt Klein: yeah.

Gabriel Savit: Oh, these flags were changed, this audience changed, this rollout, changed and have a quicker sort of path to, to resolution.

So, again, I hope I've answered some of the question, but it's really about, sort of like transposing a lot of information, helping teams actually make it useful versus just sitting in a dashboard that someone doesn't look at.

Matt Klein: I, i, I, I mean it makes total sense. To me. I mean, because again, from, from our customer base, obviously, uh, you know, people use bitdrift for a lot of different things, but one of the things that they are always asking about is, well, I want to compare this release to this release, and you, you have a lot of information that at least right now we don't have, it's like we don't know the rollout percentages and all of those [00:40:00] things, right?

So like, it's hard for us to know for sure when a particular release got cut, like what percentage is it rolled out. All of those things. So, um, you know, I, I mean I, I, I think there's some interesting integration points overlaps just in general. I'm not talking about bitdrift, but just between what you're doing and then the various observability providers, because I, I think like health is something obviously that people care about a lot.

I mean, like, they want to know when to pause, they wanna know, you know, when they have to completely pull it or need to do a hot fix or any of those things.

Gabriel Savit: Yeah.

Matt Klein: Um, so it's a... you know, it's a, it's a pain point that I think people face constantly.

Gabriel Savit: Yeah, no, for sure.

Matt Klein: Yeah. Um, so I, I guess, yeah, let's, let's briefly talk about the future.

We, we did talk about AI. I'm sure you are having your own conversations about AI or anything else. I - I'd love to know from you, like where, where do you see the mobile space going? Like what are the interesting, what are the things that you're most interested in right now?

Gabriel Savit: [00:41:00] Yeah.

Yeah. We have like a,

a couple thoughts on this one.

Which isn't, isn't trying to dodge the question, but it's sort of going other direction, is like, we know AI is gonna change the way that teams are writing code and building things and, and also changing like the tooling around it. But at the end of the day, well, maybe not if you take it to some dystopian extreme, but like at the end of the day, these are still teams working together.

There's gonna be these human touch points, there's gonna be coordination. And that's where, since the beginning, like we've focused, like, yeah, some of this is tooling, some of it's automation and that stuff can change with AI. But ultimately, like we see ourselves as kind of a, a collaboration tool and it's really about the people and helping them get on the same page.

Unlocking sort of... better patterns of working and ways of sort of... collaborating around stuff that shipping, releases, et cetera. So, on the one hand, a lot of the day-to-day may change, but you still have folks in a room or on Zoom together... coordinating, collaborating and, and this framework sort of across it all is still gonna be very important.

That's sort of one of the ways [00:42:00] we, we look at this. The other one, which is a little bit, sort of lower level, and, and I'm sure maybe you guys sort of think about this too. For me it's this funny like fox and and hen house situation where AI is writing more and more code, AI is gonna be responsible, or is already, for more and more of the testing of that code.

We see... and, and sort of add to that maybe the whole like, you know, the, the junior engineer problem, people are talking about that hole that's gonna exist, like we see a, a growing quality crisis or at least a risk of that, right?

Matt Klein: I agree. Yeah.

Gabriel Savit: And so we'll still need some good checks and balances in place.

Maybe it's not always human in the loop, but there might still be a little bit of that. But in general, just good, good observability. Good... yeah, good I guess tooling or people around what's happening to make sure that there's not this... quality collapse.

Matt Klein: Right. Yeah. I was gonna say, and one, one of the things that, that we've started to think about... and again, this is not fully formed, I don't know where it leads, [00:43:00] is... if we assume that in the future more and more code is gonna be written by the agents, which is probably true to some degree.

How can we help the agents write code in a way that works well with the tools? Like that - that's actually, I think, a more interesting question, right?

Gabriel Savit: Mm-hmm.

Matt Klein: Just because if the agents can generate... they, they're probably gonna be bugs, right? But how do we help release management be better with the tools? Or...

Gabriel Savit: mm-hmm.

Matt Klein: How do we help have better monitoring or observability of the code that's actually getting released?

Gabriel Savit: Yeah.

Matt Klein: So... I mean, it, it's very refreshing for me to hear you, you know, not just splash AI all over your website. I, I mean, I, I, I think we're the same in the sense that I believe this is very real technology.

Like we will certainly add AI features where it makes sense. But I, I don't think... the AI stuff is the end all, be all solution to all problems, I guess.

Gabriel Savit: Yeah, I, I think we're aligned there. There's a lot of pressure and, and especially when - when you're at this sort of [00:44:00] stage as a, a startup and a company in this kind of space, you know, there's so much, so much noise and there's a lot of pressure.

Matt Klein: Yeah.

Gabriel Savit: Uh, to do, kind of like what you said, just tack AI - AI onto, onto everything. Uh, resisting that, but not also being sort of like a, a Luddite about it. Like, we know, we know it's happening. Our own team is obviously, you know, engaging in this kind of stuff. So, yeah, like, like thinking about it sort of critically, but also thoughtfully how you want to engage, and bring it all the way back as long as humans are in the picture,

that's important. And being thoughtful about how we sort of incorporate AI in the product and how we're thinking about sort of what's gonna gonna change with AI. Humans still in the picture. Maybe even, I don't know, think of, think of these agents as a, another part of your team, right?

There's, there's interface there.

Matt Klein: Yeah, sure.

Gabriel Savit: There's surface areas that need to be managed. So yeah, our heads aren't in the sand, but, but yeah, there's a lot, a lot to... a lot of noise to it, I think as well. That's not gonna be productive long term.

Matt Klein: Cool. Fantastic. Before we end,

is there any, final hot takes that you'd like to leave us [00:45:00] with? Either about technology or scuba diving, or anything else?

Gabriel Savit: I don't know about hot takes,

i'd encourage everyone to try out scuba diving if you haven't already. Highly recommend. Um. Yeah. No,

yeah, hot takes by us... we didn't really get here, but it's related to the human stuff. I think one thing we'd like to talk about that has surprised us along the way... we had this sense of it, but then recently ran a survey and kind of measured it, is automation paradox. Maybe this ties into AI. Like there's, there's, even before AI, there's been so much focus on, on like tooling from an automation perspective, that's where the pain is.

If we could just automate all the sort of manual tasks we'll be good. And, we've had this inkling that yes, like that's some of the problem, but there's also a human layer. And we ran this survey of like 300 plus mobile teams recently, asking some questions around kind of process and, and team setup and, uh, found that teams who self-report a higher level of, of automation, more sort of maturity in the automation area, they at the same time report the same sort of average wasted [00:46:00] time, per release cycle as teams that self-report much less automation.

Matt Klein: Interesting.

Gabriel Savit: Um. Which, you know, there's various factors feeding into that, but we've, we've been calling it this automation paradox and, and sort of confirming this hunch that, you know, that is one piece of a, a bigger problem to, to solve.

Matt Klein: But then, but then what is the solution? Or you just don't know, have to figure it out.

Gabriel Savit: Solution, solution's, Runway, uh, if I'm allowed

to say that. No, no. I think it's like, it's not, it's nothing novel, right? I think it's just more understanding from, from leadership that automation, and maybe take the modern extension of that AI is not panacea, right? And so you need to...

Matt Klein: I see. Yeah.

Gabriel Savit: Put...

Matt Klein: so- sorry.

So like what you're saying is that it has to be automation done... right. Not just automation for automation's sake. Is that, is that basically what you're saying?

Gabriel Savit: It's that, and also the sense of... what we see sometimes is like you throw a bunch of automation into the mix, things speed up teams start... well, it feels like they're shipping more quickly, but you actually increase the size of the so-called [00:47:00] black box.

There's a lot of stuff happening under the hood, a lot of changes happening that's automation driven or maybe AI driven and visibility goes down.

Matt Klein: Yeah.

Gabriel Savit: The sort of transparency, understanding across a team goes down and so you have this sort of like... response on the human side of more back and forth on Slack.

More engineering/product, trying to get on the same page.

Matt Klein: Yeah.

Gabriel Savit: 'cause stuff is happening that no one knows, like what's going on. So it's sort of this, this, this reverse effect, right? Where things are speeding up, but also getting harder as they do. The solution being thoughtful automation, the right sort of process and culture around it, and the right tooling to make stuff... to, to fight the black box effect.

Matt Klein: Yeah,

Gabriel Savit: make sure stuff is clear. Communication model is sort of well set up for that. All, all that kind of thing.

Matt Klein: Cool. All right. Well, thank you so much for joining. This was a fantastic chat, also, really enjoyed it. So that's a wrap for this episode of Beyond the Noise Signals, stories, and Spicy Takes.

Huge thanks to Gabriel for joining and sharing his story. You can find this episode and all past ones on the [00:48:00] bitdrift YouTube channel. And if you had fun, drop us a review, tell your friends or yell your favorite hot take into the void and just make sure to tag us. I'm Matt Klein and I'll see you next time.