Daytona

Ivan Burazin proudly presents Daytona, a secure open-source development environment manager that allows developers to tailor their environments according to their specific needs. Ivan looks back on how their company started as an enterprise product before becoming an open-source project for individual developers. He breaks down its most important features, particularly those that assure ease of use,clarity, and simplicity. Ivan also explains how automation and AI tools could further improve Daytona, all while balancing user privacy and accurate data collection.

---

I have a super interesting guest with me, Ivan from Daytona. I canceled on you twice. One is due to family logistics and one is due to an impromptu Taylor Swift concert with my daughter. First of all, thanks for that, and thanks for being here.

No worries at all. I hope you enjoyed it with your daughter. I’m also a brand new father, so I’m starting to understand as well. Thanks, for having me and I appreciate it. No problem at all.

My first question is, are you a massive car guy?

I am.

Is that where the project name Daytona comes from?

It does. It’s very much inspired by racing and circular tracks and whatnot. We’re not allowed to identify too much with racing because of the name, but I still wanted to keep it in there for people who do ride.

It’s a town as well.

It’s a town and a beach.It’s a watch and a car. It’s a bunch of things.

I was talking to a couple of friends. I went to Le Mans several years ago. I amin any way interested in car racing. That is something that I would 100%recommend you do. It’s an amazing experience for people who don’t know. It’s a24-hour car race. It’s crazy and super beautiful. I was talking about doing atrip to Daytona. Maybe this is a sign.

I was at Le Mans as wellin 2023 or a year before. That was great. I’ve never been to Daytona, but I have driven in those old arcade machines in the ‘90s, the Daytona 500 by Sega.I’ve driven it so many times. That’s why it’s probably that itch in my mind.

Career Background

Funnily enough, the arcade I go to with my kids has a Daytona cabinet there. They’re getting harder to find because they’re breaking down and very old. There are moving parts in them. By pure chance, the local video arcade in London is one.I do play that quite a lot. I’m not very good at it. I don’t know why we’re talking about cars all of a sudden. Do you want to tell us a little bit about your background? You have an interesting development environment background. You started a pretty big conference as well. It’s a very interesting CV. How did you get into technology, engineering, and all that stuff?

Engineering as a kid,that’s always how it goes. Not the first company, but the first startup, let’s call. It was called or is called Codeanywhere. As far as I know, it was the very first browser-based IDE out there. It was started officially in2009. My cofounder started it and he brought me on. I had this company, which was a services company. He was a CIO at a bank and he was tinkering around. He had a big laptop at the time. He was like, “Why can’t I bring my laptop with me and log in from anything and do client work or whatever I did?”

He created this browser-based black-and-white, FTP connector on the left-hand side in the notepad on the right hand. I tell this story so many times. At that time,Google acquired rightly to become Google Docs. I was reading that on TechCrunch. This is so old.TechCrunch was the side hustle of Michael Arrington. It’s a side hustle for hisHIO or whatever his start-up is. I’m like, “No one has built this. Everything is going into the browser.”

Salesforce had already started. I forgot what the slogan was, but everything was moving to cloud. I’m like, “No one has created this for dev environments. It was in development back then. We were probably like fifteen years too early for that one. We started all that. In 2013, we founded a company because then there was Web 2.0 so you could do all the syntax highlighting in the browser. You couldn’t do that in2009.

We were contemplating using ActionScript in Flash. Who remembers those things? Luckily, we didn’t use that, and Web 2.0 came out. You can do all these things in Ajax. I was like,“We can do this now.” We founded a company and we started. We had a bit of success. We did that full-time probably four years after that and raised a bit of money. Over the years, I think two and a half million people signed up for Code anywhere. I had a call with someone very interesting. I told them, “We did Code anywhere, this company.” He’s like, “I used that.” There’s so many times people say they’ve used that.

We worked on that hard.It generated revenue. We never ran out of money. We only raised probably shy of$1 million but there was no enterprise expansion. There was no market pull to that. It was for hobbyists. Those people churn a lot coming in and out. We decided to leave one person running that. We bought all the shares back from our investors. It stayed as this internet thing on the side. It has made a bit of money for the last couple of years and then I went off and created a developer conference because I know what else to do.

That’s a story on its own. The conference is now I think four and a half thousand people in two locations. One is on the Mediterranean Coast in Croatia. The other one is in Miami. Both are tropical palm beach-type places. I sold that to this large private communications platform as a service company called Infobip, a Twilio competitor. I was on their executive board and ran their developer experience team. I built it out and ran it for two and a half years, and then started Daytona. Sorry, long intro to that.

The reason we started Daytona and I left Infobip early was because Codeanywhere still existed, and people still reached out to us. At one point in time, we got so much in bound from analysts, that Gartner Forrester, IDC, and CB Insights wanted to be briefed. We got a customer call that they want this on-prem dev environment manager thing. We even started getting acquisition talks. We got an offer to sell Codeanywhere. At that point, we did start building out an MVP for what ended up being Daytona.

For the people tuning in,Daytona is what we learned in orchestrating these remote dev environments for a browser-based IDE and rebuilt that as a DevOps platform engineering tool that can be run on the infrastructure of very large enterprise companies. We found that companies like tech companies, so Google, Twitter, and Slack all have it built internally.

I’m not naming any customers. I’m naming companies people would know. Stuff like a Boeing, Airbus,Walmart, or Walgreens. They have large developer teams and they don’t have anyone to build this internally and they’re looking for a vendor off the shelf to do that. That is where the original idea for Daytona came up and that’s how we started the company. We started as an enterprise first. Not as a nopen-source bottom-up motion first. That’s how we kicked it off.

That’s interesting the parallels with Flagsmith. We’re the opposite way around. We were open source first but then it was obvious that on-premise enterprise was where the business was going to make revenue. We’ve also found that probably the best feature flagging stroke,multivariate, and testing platform that exists on the planet is internal to Facebook. If you’ve worked at Facebook, you’re extremely familiar with it.

When you leave your job at Facebook, you desperately miss it to the point where we had ex-Facebook folks reaching out to us. There were like PhD level stats and stuff in their platform to do this like crazy A/B testing. It’s interesting the choice of tooling that the companies at that scale make in terms of things that they do internally. They also don’t open source. They open source Kubernetes but there’ll be a bunch of internal developer stuff that they see as it’s like critical IP, which is interesting. If I were to clear my brain and write five things that they’d be interested in protecting from an IP point of view,feature flagging, A/B testing, and remote development environments wouldn’t be on that list.

You haven’t thought about it in this way, but one thing I have thought about and to your point is, first,I know a lot of dev tool companies that are now creating services or platforms that these large companies have created for themselves, so the Facebooks andGoogles of the world. It’s like, “Let’s give this to everyone else.” That seems to be very popular and happening lately.

I also wouldn’t agreethat remote dev environments are something that they’d think of as IP. On the other hand, especially at that time, it’s like, if our devs are more productive and faster. For us specifically, we found that developers waste 56% of their productive time. That’s more than half of their productive time is gone because they’re working around dev environment issues. If I’m Facebook, maybe I’m faster than whoever I’m competing against in this sense. Is it Google or Apple?Whatever it is, because I have it and then they have to go and make it themselves. It could be that as well or it could be they don’t care. Not sure.

Developers waste around 56% of their productive time working around dev environment issues.

Perfect Timing

I remember that time around 2008 and 2009. There was that rush to put everything on the browser, wasn’t there? Google, I can’t remember the company they bought.They acquired all the companies that all the products that are now part of whatever it’s called nowadays Google Workspace or G-Suite or whatever the hell it is. Google is in trouble because they start renaming things every three years like Microsoft does. That was a super exciting time to be in web development. Gmail was the first platform that made people realize what was possible.

It’s interesting when you talk about timing for a product like Daytona because it feels like there’s a whole bunch of stuff that came out. Especially around the docker and containers, the capabilities of browsers, and things like wrapping web engines in applications, which is how VS code works. You mentioned you were fifteen years too early when you were talking about Codeanywhere.

Were you conscious of these pieces falling into place over time? We tried tons of projects. As an agency, we start loads of side projects and Flagsmith picked up. It did feel like timing but we didn’t know that the timing was right when we started. It feels like with your experience, where you in a bit in the matrix.

We had done something similar before so in the same space. We can say it’s the same space. We startedit early and then there’s a company called Replete, which is now worth $1.2billion, which is the same thing. Execution is there and all these things.They’ve done and great work to get where they are, but they started probably five years after we did. It was a closer time to get to that. When we started to think about Daytona and got this pull from the market, we’re like, “Let’s double-check why now.”

Investors will ask you why now and we ask ourselves why now. We wanted to make sure because you have this chip on your shoulder from last time. At a VC meeting, people said,“You’re idiots. This is not happening. This is useless.” You don’t want to got hrough that again. You look at things like, “We had to build our own IDEs.”We’ve built four versions of an IDE. I don’t want to do that again. Now, VSCode has support for remote development. JetBrains has support for road development.

Cursor, which is all the rave, has support for remote development. That’s something like, “We don’t have to do that. That’s there.” To your point, Docker is fairly standardized everywhere. It’s like super easy to use. Also, one thing that wasn’t an issue back then is the dev environment. If you recall, back in the day, you used either Visual Studio, the old one, JetBrains, or Eclipse. These were actual IDEs.

You spun them up and everything was there. When you typed code and you hit run, it worked because you had all the dependencies that you needed there. That’s not the issue today.The most popular editor is a text editor because it’s so complex that you have to set up all the dependencies. Now that has become so complex, you’re like,“Can I make this easy again? Can I standardize all this complexity so that I only have to work with it?” Now the pain problem is an actual problem. Plus the neighbors in the technological aspect are now there. We’re like, “Now this is the actual time where this makes sense to do.”

VS Code

I can’t imagine Microsoft starting VS Code. Maybe they did have that in mind at the time that they created or started working on VS Code. It’s like you could do some fiddling around and then suddenly it would appear in the browser. I’m curious to know if that was on the early roadmap for that project.

I know some people who worked on that fairly early, but as far as I know, VS Code wasn’t intended to work in the browser originally. It was originally supposed to be a desktop editor, but they used web technologies to build it. The first person to make it work is a company called Coder. They made code servers. They used VS Code and put it inside a browser. It wasn’t Microsoft itself originally. It’s like, “We might use it.” When they created their Codespaces, which is essentially a typeof Codeanywhere, Replete type or SAS product with the dev environment in the cloud and a browser-based editor, then they decided to package that all together.

Daytona

For those of you who don’t know, an end user is like an engineer’s experience. What are the before and after differences for a large enterprise organization using the platform?

All of these standardized development platforms or remote development platforms. Whatever people callthem. We call them development environment managers. Whatever may be. We specifically try to make a zero-learning curve or near-zero learning curve. Asa software engineer using it, either through the terminal or through agraphical user interface, a dashboard, you have one button which says create oryou type Daytona create. You point to the repository which you’re working on.Your project is going to be on GitHub or GitLab, whatever.

You point to that. You define what IDE you want to use, so your local VS code, IntelliJ, PhpStorm,Cursor, or whatever you may be using. You define where it’s going to run. The enterprise can define for you, “This is running in this geolocation or this geolocation. I need this size machine.” On the open source, you can define where we want to run it, on your local host or somewhere else. That’s it.

There are three dropdowns or three things you have to define in the CLI. You hit enter, and then Daytona goes out and does everything for you. What it does is it will spin up a dev environment for you. A pod or a container, depending on what technology is used. It changes, depending on where it’s installed. You don’t have to think about that. It will read through a repository, so it’ll check it out. It’ll see if there’s any configuration files. Is there a Docker file? Is there a dev container file?

Upcoming will be Nix as well and dev file. If there is, it executes against that right away. It does that for you. What it does after that is it will then open the editor that you wanted. It’ll do it all automatically. All the credentials will be done. It opens up for you. If it’s on a remote machine, it’ll be a proxy to local hosts.As a developer, you don’t have to think about, “Where is this running right now?” It’s all local hosts to you even if it’s running somewhere else.

You also get a fully qualified domain name so you can share your dev environment as a staging environment. You’re still going to have to use staging for everyone but let’s say you’re my manager or my colleague. It’s like, “Is this what you wanted?Here’s a link to what’s running on the machine. You can see that right away.”What Daytona does essentially is it will spin up an environment. It will create a secure connection to that environment. It will open your editor. It will proxy it to local hosts.

It’ll execute against all the configurations so you don’t have to do that. Everything is up and running as soon as it’s done spitting up. It takes anywhere from twenty seconds to a minute. If it takes longer, there’s a pre-build option so it’ll pre-build it after the first time and then it’ll never take longer than a minute to spend ina dev environment. Use as a developer.

The idea is that you never have to worry about working some of my machine issues or going through the readme. There’s something not right. It automates everything for you and it doesn’t obfuscate. It’s not a competitor, but it’s like a Vercel. It makes things super easy but obfuscates a lot of things for you. If you’re a superhardcore person, you can’t get into the nuts and bolts of that.

Daytona lets you seeeverything and you can get into everything, but it automates that for you soyou don’t have to think about that. That’s what it does. Anywhere it runs, it’salways local hosts. Anywhere it runs, even if it’s local remote. Anyone in theworld can share it if you share the URL. It automates all the installationsteps for your dev environment.

There’sa configuration file within the repository that explains what stuff. Just sopeople are clear when you say it’s running on a remote server, where’s theseparation of concerns in terms of what the IDE is doing, what your localmachine is doing, and what’s running on this piece of VM somewhere?

To be clear, theenterprise product itself is a cluster that you install on. The companyinstalls it on their cloud provider, on-prem, or whatever it may be. Yourentire dev environment, for the most part, is running over there. On your localmachine, there’s nothing running, except for your editor. Your editor has athin client and everything else is running somewhere else.

This is also interestingfor large enterprises because it is very much a better version of the virtualdesktop interface. Users that have ever used a remote desktop or a Citrix orwhatever. A lot of very compliant companies ask their developers to use thesetypes of remote development environments because everything has to be centralas a server. That’s an awful experience. You’re streaming everything. It’sterrible.

With a Daytona-likeproduct, it’s the same security but you as a developer feel like it’s on yourlocal machine. You don’t care, “My company is very stringent and compliant forthese things. I can still work as I would normally.” Those are the things thathappen. Also, Daytona does let you run as an option on your local host.

Thatwas going to be one of my questions. If I’m on an airplane and not in theairplanes in Europe have WiFi for those Americans using.

They’re getting there.Ryanair doesn’t and the other one.

Inever understand why Ryanair doesn’t do WiFi because they try and get money outof you in every possible other direction. I never quite understood that. Ifyou’re on an airplane, you can point it at your own machine and it runs onDocker, does it?

It runs on Docker. To beclear, if it’s part of the enterprise package and your company says you cannotrun this on a local host. You’ll not be able to do that. There’s also the partthat we open source for the individual developer. They can say, “I want to runthis on my local host or I’m going to run it on DigitalOcean,” or whatever itis. They can say, “Run it on a local host.”

If your computer orlaptop has enough CPU or RAM, it’ll run it from your local Docker with the sameexperience. Why this is very important to me specifically, before we decided tostart it officially, I interviewed most of the managers of these largecompanies that we mentioned, so Google, Facebook, Spotify, Shopify, andwhatnot, and how they did it.

Shopify was the mostinteresting to me because of the way they had built. First, they created alocal standardized environment. For every new employee, first the laptop. TheMac would go into the engineering team, and then they create everything insideso that when you got it, you could off to work. This is a quote. They said,“The cables coming out of my laptop were melting.” The size of the environmentand the computer that was needed. They went off to build this remote. It’slike, “Now we have to give people remote machines because it’s bigger.”

They created a simple CLItool for the user. I don’t know the commands but it’s something along the linesof create dev environment-cloud or -local. The cloud one will spin up. Thelocal will spin up local, but you as a developer don’t care. The entire interfacingfield is always the same, but you’ll use the resources of whatever machine youwant. If you need cloud compute, you can use it on cloud.

If you are on an airplaneor you’re doing a front-end app or whatever, you’ll do it locally because it’sa fraction of a second faster and you like that more. That was the guidingprinciple when we were creating Daytona. It’s like, “Can we do this? The userdoesn’t care what is running.” It’s completely the same local or remote. If theenterprise by a manager says you can’t run a local host, you can’t.

Convenient And Streamlined

Ilike filling around with these tools. I’m obsessed with it. I find it interesting the way engineers are so particular about the tools that they uselike not their IDE. Python is a good example because of the packagingenvironment. Python is such a shish but how they install Python, how theyinstall that, and how they manage their virtual environments. All this stuff.An engineer, especially on Flagsmith.

We’vegot Emacs and Vim. I was not using Zed but I used VS Code and the JetBrains’PyCharm. Everyone is completely different. If you were to look at the laptops,we were all sat around the table. Everything would be completely different. Howdo you deal with that? Maybe if you work for mega-corporations, you’re more usedto being told, “You will use this IDE,” which for most engineers, I know wouldbe like, “I’m not working in that company then.”

Some companies say thatby decree, you have to use something. We didn’t want to get into that. That wasone of the learnings with Codeanywhere as well. The only way that will work ata time is our IDE, which people are like, “No, you guys don’t have theseplugins. This is not it. I’m not doing this.” That’s why very much call Daytonaasl an infrastructure tool. We connect to everyone on all sides.

What we standardize on isthe runtime. The runtime no matter what operating system you’re running on ifyou take for example, the open source product that runs on your machine andyour local Docker, if it’s running on your local Docker. You can connect to anyIDE you want. It could be running on Windows, Linux, or Mac, but the Dockercontainer is a Linux container on all three options of that. Whatever IDEconnects, it’ll connect. It’s not an issue for us.

The same thing if you’rerunning it on a centralized server. If it’s an enterprise product, you’rerunning on AWS or GCP or wherever else. The container itself, that runtime isexactly the same all the time. The user can use PyCharm, VS Code, Zed, orCursor. We don’t care because we’re taking care of the underlyinginfrastructure and making sure that we’re serving always that same computepackage to the person. That is where it is super beneficial and makes thingssuper easy and why there’s no work on my machine problem anymore. Daytonaremoves all that differentiation from it.

Touching on Docker for asecond. Docker did a great job of getting us where we are but the vision withthem was like, if you create it inside of Docker, it can run anywhere. That’sit. You’re done but it never solved it. We still have issues every single day.Someone told me that we’re a rapper for Docker, which is not true specifically,but you can think about it.

Isee what they’re saying.

You have that corestandard environment and then you have all these layers on top of that thatDocker didn’t put in there that we did. That finally gets you to the point thatit always works.

Youtalked about how the enterprise motion came first and you had followed by theopen source. How did you go about it? Because that’s the wrong way around howit normally happens. I wouldn’t say it’s the wrong way around, but it’sunusual.

We started because we hadinterest from the market. We went out and started serving the market. That wasgreat and that’s going great. We were always thinking about how we serve therest of the people in the world. Google and Facebook have this and now theBoeings of the world have it. Great. What about the engineers who are hackingaway on a weekend at home in London, Croatia, Singapore, Accra, or wherever?They still don’t have the access. They don’t have this thing.

Codespaces are similar,but they still cost you money. If you can afford it or whatever, it’s still anissue. We were thinking about how to create something that we give to thecommunity to solve this. I’d say this is sometimes in presentations, and Itruly believe this. Everyone in tech knows this, but I was in San Francisco twomonths ago and I drove a Waymo. It was awesome. I thought that was awesome. Ihit a button on my phone and sat in a car. Without a driver, it drove mearound.

That works by itself,mostly but dev environments still need people. What’s going on with the world?Where are we? How is that even possible? I was like, how do we give this to therest of the world? We thought about it for a long time and we also want to bevery cognizant of issues happening in the open-source space. We didn’t want topull a hashy corp or something like what Sentry did. All these people arepulling back slightly on their licenses.

We want to be verydeliberate what we give to the market so that we don’t end up pulling it away.When we’re thinking about the enterprise product in the sense of value, thereare three levels of value. One is the security level, the visibility, and theaudit logs where it can run and all these things. The second is the clusterthat spins up and down all these dev environments on a centralized server.That’s mostly for DevOps and platform engineering.

The actual developerdoesn’t care about all that. It’s like, “Can I have one button, and theneverything runs?” We carved out that bottom and that is what we open-sourced.You can think about it as open core but we split up the repositoriescompletely. The repository DaytonaIO/Daytonaon GitHub, you will purely see that. It’s a very small repository. You candownload it as a binary as well. It works out of the box.

We want it to be neat andsmall mostly because it’s easier for people to use. If you have an open core,especially for a lot of contributors like, “This folder has this license.” It’sso much more complex. We ended up carving out the simple part. The output ofthat is a CLI tool where you type in Daytona crate and point to a GitHubrepository. It does everything for you, either on your local host. Now we’velaunched providers, which means you can add your tokens from AWS, DigitalOcean,and Fly.io. We’re adding GCP in Azure.

If you’re like, “I wantto run it somewhere else because of size and compute,” you say, “Do it overthere.” It goes out and spins up the EC2 or the droplet or whatever and it doeseverything for you automatically. That ease of use and that removal of allthese issues is what we open-sourced. We did that in March of 2023. Back toyour question, it took us a good nine months to think about how to give valueto the individual user without hurting the business and have a clean cut withthat.

We thought a lot abouthow to do that. Although, it’s the opposite. Now when I talk to a lot of opensource founders, I see that a lot haven’t thought about it. They like, “Let’sput it out in the open and then figure it out.” What happens is that later on,you end up having to pull back the license, cut out things, or do other things.We want it to be very deliberate so that we don’t put ourselves or our users inthat situation.

Wehave an almost identical approach. We didn’t do a huge amount of upfront thinking,but when the activity around a repository on GitHub started to pick up, andinbound leads started to pick up, we had to start thinking about how would webuild a sustainable business. At that point, I started thinking about it a lot.For those tuning in, if you Google Flagsmith's open-source philosophy, there’sa deck that I put together that’s like five pages. It boils down to the samething.

It’slike if the engineer cares about it, then it’s open source. If the engineerdoesn’t care about it, then you have to pay me money for it. It’s been greatbecause there’s almost never anything. It’s very unusual for us to come up withsomething that we’re working on and say, “This is a gray area.” There’s almostnever any gray area around that definition. Especially if you are an engineer,you know. They don’t give a crap about that like OIDC. They don’t care. I auditall this, they don’t care. All that stuff.

Thathas worked well for us. It’s interesting as well because even though they’reradically different projects and products, we’re speaking to exactly the samepeople. You can do that. I feel fortunate that we’re in this position wherethere are never those difficult decisions. I can imagine that you’re working ona business where there are many more arguments within the product team about,“This should be open. It should be closed.” You open-source it and then youstart losing deals because mega core pay has gone off.

Providing Value To Individual User

Weonly cared about that thing. You’ve made that open-source. We’ll use that. Thatdoes happen with us from time to time. Not that often, but sometimes they’relike, “We don’t care about that. We’ll run the open source.” We’re like, “Fine.Hopefully, you’ll call us back in a year’s time and say, ‘Now we care aboutit.’” Sometimes they don’t and that’s fine. I started working with someAmerican folks who didn’t have as much experience with open-source as me.

Venture Growth

Whenwe started losing those deals, they were so upset. I’m like, “You have to seethe big picture. You’re going to lose some, but then we’re going to win a bunchmore because of our stance around that. You have to be at peace with it butsometimes it’s painful.” Interms of usage and stuff, can you talk a little bit about the growth of theopen-source projects? How long has it been available? How did you go aboutmarketing it or that stuff?

I wanted to touch on andthat will lead into this. I did more on the licensing and when we launched intohow we did. To your point, in a lot of things that we’ve done, it is very easyto figure out what is open-source and what is in the proprietary productbecause it’s a very clean line. Even if an engineer is like at one point, “Thisshould go here.” The same as you. What is the first principle? Is this usefulfor an engineer, for the platform team, or the CISO? You know where it goes.It’s very clean cut.

On the other hand, as toyour point, when other companies, there are examples. I won’t name names, butcompanies that’s like, “We don’t know.” It’s because some companies are notthat fortunate that they can make that clean cut and then they open-source thewhole product. They work around licenses then you have licenses like AGPL 3,which is like, “I’m going to stop Amazon from hosting this and killing me.” Thesame as Lasttix issues. On the other hand, large enterprise companies are goingto host on-prem because they can and you lose huge clients because everythingis open-source essentially.

You don’t want to dothat. Not everyone has that but when you’re thinking about open-sourcing aproduct, you very much should. Most of the service products aredeveloped-oriented, I believe anyway. It very much should be what is for theindividual user. That’s open source and what’s not closed source. Pay me moneyin your words and whatnot. If you can do that upfront or think about it longenough, you’re on a good path to success, but it won’t hinder your successlater on in life.

That also shows a pointfor us. We open-sourced Daytona on March 6th. It’s 5 or 6 months since then.It’s still a very short amount of time. We hit 8,000 stars. Looking atcompetitive or comparable companies. We’ve passed two of them. There’s one morethat we have to get to and these are companies that have been around 2 years, 3years, or 7 years, so a long time, which goes to show maybe a bit of how wemarket in our go-to-market and whatnot.

Also, because we’ve opensourced a specifically such a small project for these specific users, there’sless complexity. I believe that’s one of the things that helped it propel somuch faster because people are like, “This is very clear. This is all that’s inthere. I understand this. I can use this right.” That’s my thought around thatas well, which leads us to when we launched it and how it’s going.

That’svery interesting that you say that. Something that drives me crazy is I seethis quite a lot. Again, I’m not going to name any names and this is verycommon around enterprise stuff, especially commercial open-source or open-coreenterprise stuff. I see this a lot. You’ll find a GitHub project with 10,000stars or 6,000 stars or whatever. You look on their marketing website. You lookat their GitHub read me and you’re like, “I don’t understand what it does.”It’s massive.

Onething I always do is like, “Is there a docker-composed file? How many servicesdoes that start?” It starts with sixteen services. Three of them you’ve nevereven heard of. There’s some weird database or whatever. I’m like, I’m not goingto make the investment of figuring out what the hell it even is. How does ithelp me? I don’t even know. It’s this complex thing or maybe it’s been opensource from some large enterprise organization. I’m trying to attack a singleproblem and show a solution to a single problem rather than trying to solveseven. I wonder how much. I’ve been thinking about Vagrant a lot while you’vebeen talking because you can explain it in one sentence. That’s amazing growthyou’ve shown to get to that point.

There’s one underlyingprinciple that we’re trying to work on and it’s not perfect. For people who aretuning in, there are still things to do. From a user perspective, it’s like,“Don’t make me think. Can I make it so that the person doesn’t have to think?It’s very understandable.” Even the repository, if you open it up, it’s all ingo. When you see people looking at it, they look at it. If they’re not clear,they open up the repository and take a look at a couple of files. “It doesthis. It does this. Why does it do it?” It’s fairly easy for you to pick upwhat’s going on in there. We try to do that in the interface and whatnot.There’s still polishing to be done. How much stuff can we remove to make itsimpler? What we can’t remove, can we make it so that people understand itinherently with minimal cognitive load?

Telemetry

Wetalked briefly before we started about telemetry and VS Code telemetry as well.Telemetry is something that we’ve not ethically struggled with but we’ve neverbeen quite sure what the right balance is. I’m not interested in a largeorganization or engineering person who pulls out our images and runs themlocally. It’s not like I need their phone number or anything but it would begreat for us to know other than the GitHub or the Docker image download metriclike how many instances there are out in the world. You’ve introduced telemetry into theplatform.

For us, it was a similarstruggle. The team is very small. The company is fifteen months old, so we’restill fairly small. We have an enterprise product and an open-source product.One of our investors also said, “If you’re going to do two of those things.Everyone told us not to do it. Lightning has to strike twice. You have to havea great open-source product and a great enterprise.” How can we build this asfast as we can?

Telemetry was one of thelast things to add at all. The reason was what we were talking about earlier.Our engineering team was like, “We can’t do this. People are going to getturned off. They’re going to say we’re spying on them.” It was a very bigargument in how we do it. I’m like, “We have more important things to do thantelemetry. We have Stars, Docker download, and all these things.

There are people, whichis great but what are they doing? Where are the issues? I would love to knowthese things. To your point, we don’t care what the person’s name is or whatcompany, whatever. Generally, how many workspaces do they have running? How arethey up and running? The programming language is inside.

WhatIDEs are they using?

Is everyone using cursoror is it like no one? Is it like hype? It’s code. We’d love to know thesethings. We very delicately launched it. It’s very explicit in the documentationlike no personal information. These are the only things we’re doing. This ishow you can disable it. There’s all these things around that. No one disabledit. No one cares. We were talking about this as well. When we did our researchlike VS Code. Open that up, they take a look at every single thing you do. It’sscary when you look at it.

I’m not going to say noone disables it. I’m sure people do but the vast majority of people don’t. I’msure, if you look at your VS Code, it’s on. Mine is on. You don’t essentiallycare. It is one of those things that I think, us being engineers buildingengineering products, you’re more sensitive to than it is. That’s my take onit.

Daytona In The Future

Iknow some people are religious about this and where religion is concerned,people like to tread carefully. There’s been a bunch of stuff we’ve done withFlagsmith where we’ve been way too timid. Not timid, but I don’t know. You canget stuck into a situation where you’re overthinking these things as well. Youend up arguing it to death and then no one’s happy with the outcome. In termsof the future of the platform and the business, you could see finishing theproduct, other than IDE support. It’s one of those things that could almost belike I’m done. Do you ever think about it now or not?

I think it’s a trickbecause I listened to it. Myself and my cofounder talk a lot. It’s like, “Whenare we going to finish this?” We’re like, “Done.” Six months pass and then youhave more things to do. There are a lot of things. Also, on the other hand,being an engineer, this is something I’ve learned with the third company that Ifounded. Even let’s say the product is done, which is never going to be done,the company isn’t done.

Evena finished product will never be completely done. It is always a work inprogress.

That’s when the actualgo-to-market sales or whatever motions have to get started. The thing that weall make the mistake as engineers is we’re excited about building the thing.You build it. You figure it out. Can it be done? Can it be done on time orwhatever are all these complications with it? When you get to not even done,but a usable product, where’s my sales team? Where’s my support team? Where’sthe building?

There are so many thingsthat you have to take care of and do, which is as exciting as building it. Maybenot to an engineer, but to a person. That’s why a lot of founders if they sellthat early, they end up doing another one and another one. They like thezero-to-one part. Whereas the 1 to 100 or whatever the one number is equally ifnot even more interesting, but it’s different.

If you have a look at ourenterprise product, our enterprise product has been done for six months, butcompliance, applications, GDPR, and SOC 2. Now we have to implement this. Auditlogs, logging this, and logging that. There are so many different things thatyou have to do and they’re all tasks and work but they’re not the excitingthing. Is it a magical experience to set up a dev environment? It has been forsix months. Now it has to be a measurable experience inside a data center orthe Pentagon. I’m making this up, but somewhere. That’s the strengths that youhave to build out.

Interms of speaking to engineers who are using the open source products, arethere things that you’re excited about that are coming down the plane?

The open-source productis not even a stable version. We haven’t changed it yet. Everything is still0.28 or whatever. It’s enabling us to move super fast. Some of the things thatwe’re missing, at least from my perspective, is when it ran on the local host,it was slow to get up because of the way we piped the data through a VPN tomake sure it’s secure and all these things.

We wanted it to get tonative speed. It’s as fast to set it up like a machine not using the Daytonaand using the Daytona. We wanted that to be there. We wanted all the IDEsupport. We need the support of all the other configuration files. Also, thisis something very personal to me. One of the things missing in the entire flowis the automated creation of configuration files. Let’s say Daytona is amagical experience if you have a great configuration file. If you don’t, it’sstill better than what you were doing but if you have one, it’s magic.

We’re in the world ofwhere we are and I’m not trying to get into AI hype. Having a service that looksthrough your repository, “There’s no configuration. Do you want us to make itfor you?” You get it done and then you have that managerial experience. Youalways have that experience. I think that a Daytona service, hopefully, Daytonaitself will become ubiquitous as much as even more than Docker. It’s like Idon’t need to think about this stuff anymore. Pick any repo and every repo, assoon as it spins up, it opens an IDE and you can run and it always works.That’s what I’m super excited about to get to. Some of them are smallermilestones. Some of them are larger, but that is all in the foreseeable future.

Isthat work being done or thinking around deterministically figuring like,“That’s a mono repository and it’s got a Django application and a Reactapplication. I’m going to magically figure all that stuff out.”

There’s not an exact datefor launching that. We have a beta version that we’re going to launch as astandalone project that anyone can use for any repository. That will come outin two weeks but there’s still a lot of work to do to say that this isdeterministically always going to work. It’s not there yet, but the way wethink about it is the project, that service should be part of Daytona as well.Read the readme file instead of me, the human so I know what ports I have toopen, what I have to install, and where I have to go. Go through therepository, read through the code, and figure out what are the dependencies andwhatnot.

The readme will usuallyhave most dependencies in there, but sometimes people forget. Can you scanthrough that in a reasonable amount of time and then create this configurationfile so that I, as a human don’t have to use that? I would prefer to wait anextra 30 seconds for the LLM to go out and do that than having to do it formyself because I don’t want to figure that stuff out.

If we can do thatautomatically, that would be great. Also, what we think is if you already havea configuration file, it should check, so you commit some changes. That servicewill also go, “I didn’t see any changes to the configuration file. Do you wantme to go check your diff? Did you mess something up?” That should be there.That’s super easy and super fast. That’s something that will be usable foranyone who uses Daytona or not, but having a Daytona service, having that meanspure magic for anyone on any project at any time. Where we are now, that worksbut if you try to spin up, let’s say backstage, which is super complex, it’llbreak. It’s not there yet.

That’san unwinnable battle at the end of the day. You could only strive for 90% X ofcoverage.

If we get to 80%, I’msuper happy.

Maybethings are getting a little bit more esoteric, but the vast majority of peopleare running on frameworks and platforms that you could probably reason aboutfairly well.

You can figure that out,especially if you’re looking at or starting up a new project for most people.It’s like whatever is hot on GitHub right now, you could probably figure that out fairly straightforward. It’s not that super complex. Also, you could say,“We’re probably not sure that this works.” It warns you and then go figure it out. It’s like, “It doesn’t work on my machine. Sorry, go figure it out.”Otherwise, you could say, “Here you go.” That’s something I’m super interested in and that’s the last technical issue of why now that’s missing to be supermagical always for everyone.

Episode Wrap-up

Ivan,thanks so much for your time. It’s been fascinating. I’m going to show in a couple of AWS keys and give it a spin. I wish you all the best for the future.

Thanks so much. Thanks for having me, Ben.

Important Links

• Daytona
• Codeanywhere
• DaytonaIO/Daytona
• Daytona.IO
• GitHub/DaytonaIO
• @DaytonaIO– X
• Daytona LinkedIn