Building the Next Generation of Developer Tools

Michael Grinich (00:02):
Welcome to Crossing the Enterprise Chasm, a podcast about software startups and their journey moving upmarket to serving enterprise customers. I'm your host, Michael Grinich. I'm the founder of WorkOS, which is a platform that helps developers quickly ship common enterprise features like single sign-off. On this podcast, you'll hear directly from founders, product leaders and early stage operators who have navigated building great products for enterprise customers. In every episode you'll find strategies, tactics, and real world advice for ways to make your app enterprise ready and take your business to the next level.

Michael Grinich (00:39):
Today I'm joined by Sam Lambert, the CEO of PlanetScale. PlanetScale is a next generation database built for modern applications, including serverless apps. It currently serves hundreds of millions of queries a day across thousands of databases, and its customers include companies like Airbnb, Slack and GitHub. Along the way, this meant PlanetScale needed to become enterprise ready and build new enterprise features and functionality for I.T admins. We're going to dig into all this and more and talk about how PlanetScale is moving upmarket and crossing the enterprise chasm. Sam, welcome to the podcast.

Sam Lambert (01:12):
Thank you so much for having me.

Michael Grinich (01:14):
So let's just jump right in. Give us a quick update on PlanetScale. Where's the business and team today and what's your current focus?

Sam Lambert (01:20):
I was just thinking about this today actually. It's been a fantastic year actually over at PlanetScale. So we launched our product in GA maybe two or three weeks ago, and we launched our beta in May. Now the history of the company is that the company was founded to deliver the test, which is this open source database orchestration layer that was built at YouTube to power all of YouTube, and the founders of the company kind of broke it out and decided to build a company on top of it. And then a year ago, in fact, the first commit on the whole platform was on December first, so really just a year ago. We kind of pivoted a little bit towards building a serverless database cloud platform on top of the test and we did a lot of hiring. The company has grown 4X in size in the last year, and it's been pretty amazing.

Sam Lambert (02:07):
We've not only unlocked large websites that already run on the platform and actually ended up running on the platform during our beta, but we're still working with really huge enterprise organizations and supporting them with their database needs, and it's great. Like a new game that we are playing, which unfortunately I can't name, that was just released. A very popular video game is running on our tech and it's just been a really surreal year to kind of look back and see the name everywhere, and see our technology running on just for gigantic platforms. It's been really awesome. What's on my mind? It's like you said, going upmarket, reaching more and more companies, building for more and more folks and making sure we continue to delight what we feel is our true core audience, which is developers.

Michael Grinich (02:49):
In October of this year, you folks announced that you had this enterprise plan coming out with single sign-on auto logs in unlimited branches, many other features for larger organizations. What was the decision between prioritizing this over maybe some of those features for individuals? I'm curious if you can talk through how you balance those two in your roadmap today as you're moving upmarket.

Sam Lambert (03:08):
Yeah. So it's really interesting. And I think actually lately, it's been less of a kind of either role. It's both now for companies, I would say. We wanted to prioritize it so that we could actually just reach developers everywhere. Sometimes the developers that are most in need of modern, awesome, great tooling are inside enterprises, and so we wanted to be capable to serve that audience, and we talk about that feature set that you just mentioned as capabilities, things that are enabling so that a developer who is at the absolutely bleeding edge, hottest startup can use the same technologies as someone who is at a hundred year old company that is revolutionizing their technology.

Sam Lambert (03:49):
We really wanted to enable both groups of folks and it's about meeting those larger needs. And it came with also us releasing our managed product, which means that we have kind of this separation of our backend and our data plan, that can actually live inside a customer's cloud, and that's our kind of modern answer to on-prem. We very deliberately say we don't do on-prem managed software. We're not trying to package our PlanetScale into an on-prem device or like VM or whatever, and ship it off inside people's enterprises. So we've tried to go for this kind of, on-prem maybe cloud native way of delivering this, and it's just so that we can get in and build for more and more people.

Michael Grinich (04:31):
I'd love to hear more about that, about the managed cloud offering. I know you guys started working on this over a year ago, probably conceptualizing it even before then. Walk us through the story of how this came to be, this kind of management control plane for on-prem service. It's kind of this like cloud prem. I don't even know what to call it. It's a new architecture.

Sam Lambert (04:50):
Let's name it now. Cloud prem

Michael Grinich (04:51):
Yeah, there we go. Cloud prem. Where did that come from? What's the genesis of that at PlanetScale?

Sam Lambert (04:57):
All credit to our engineering team, we always knew that with having a technology that's powerful as ours, it's going to have large enterprise demand. We had a do over with building this new platform starting again, and they really took a step back and thought about the different ways a business would want to consume and buy the software, and they thought about how you develop this incredibly beautiful intuitive experience that you kind of really only get with SaaS software, right? This is what service is about. Service is not just about not asking someone to configure servers, which I think is a really stupid demand of any cloud platform in 2021. I just don't think it's necessary. So we wanted to be serverless and we wanted to deliver this incredible experience, but we weren't asking large enterprises to just kind of hand their data over straight away to a third party without building up an absolute massive amount of trust.

Sam Lambert (05:53):
And this it's funny, actually, I think the tide's changing a little bit on that. We speak to customers and before we go to the meeting we try and qualify the demands that they're going to have, what they're looking for, and we're often like, oh yeah, they're just going to want managed. They're just going to want the cloud prem, they're an old school company. And it's really actually surprising sometimes how often they're like, no, actually we know we're over that phase of company. We're happy to give the hosting to the experts, but not everyone is there. Right? They have different regulations, they might need to secure their infrastructure in different ways. So we really thought about how we could disconnect the data plane from the control plane. The control plane is where the real secret source is. All of the things that enable branching, enable our own line scheme of changes, all of the kind of magic, and there's a lot more magic on the roadmap for next year. And we want to keep control of that.

Sam Lambert (06:43):
Packaging software up, and we really saw this at GitHub with GitHub Enterprise. We were scaling the world's largest code host, and that takes a lot of building services that are really bespoke and specific, and then taking something that runs on thousands and thousands of servers and crunching it into a single machine to deliver that experience was hellish and we never wanted to go through that again. But we really had to think about where the parts of the stack have to break apart and what you can give the customer, and the way the team came up with it, which is absolutely fantastic. So we have our control plane. We have something called singularity, which basically can schedule work that happens from the control plane, and singularity knows how to speak to any number of backends.

Sam Lambert (07:19):
Whether that backend is through an encrypted tunnel, through into someone else's cloud or it's on our cloud. It looks exactly the same. There's no difference in the architecture. So that means our development teams can continue to build fast and the infrastructure is then just segmented and isolated specifically to the customer. So it's a very kind of elegant, modern way of building this type of service that isn't full of foot guns and it just isn't incredibly difficult to build against. And all credit goes to the engineering team. They thought we're starting from scratch. How do we figure this out with all of the criteria for the future in mind? And they laid up the architecture and it's just fantastic.

Sam Lambert (08:00):
Actually we had an incredible amount of people reach out when we started talking about it, because they were like, we were scratching our heads on how to deliver something on premise without actually doing on-prem software, and they were like, this is amazing. This clicks and they loved it.

Michael Grinich (08:13):
That sounds like a really elegant solution to a problem that's been plaguing tons of different companies. I've heard about this all the time in terms of where you store the data, data residency, who has controls over it. Do you find that the people that are concerned about having access to the data, that this is kind of a legitimate concern around data residency or something like this, some regulation or are these people who are eventually going to move to the cloud? Are they folks that are just kind of a little bit of a Luddite, not to call your customers Luddites.

Sam Lambert (08:42):
I wouldn't say it's Luddite. I'd say it's understandable. So I was the buyer once and when we were buying for GitHub, we didn't want our availability to be in anyone else's hands, and that was partly a comment on the majority of certain services or where the cloud actually is or was. But I think that is actually changing. I think SaaS software is getting better and better and I think it's going to become a generational thing. We've seen thousands of startups start on PlanetScale this year, and then you look at the stack they use. And one of my favorite things is people share on Twitter now, all the different stacks they're building their startups with, and they're piecing together all of these incredible SaaS products and building a full business straight away on this really scalable infrastructure, and that is an amazing generational shift. And I think all companies are going to want to consume software that way, because you can stand on the shoulders of giants really easily now.

Michael Grinich (09:38):
Let's talk more about that shift happening in SaaS. So PlanetScale, if you go to planetscale.com, it clearly stands out as a company that cares about design, craft, like the polish around the experience. Very different than typical, maybe, enterprise software companies or certainly other companies that are doing database hosting, right? That's been a pretty dry industry for a long time. Talk through why this is important and how you approach this, what it matters to PlanetScale?

Sam Lambert (10:04):
That's a great question and it does matter to us. It matters on various levels. You could cynically justify it by just saying that enterprise is commercializing. We look at the iPhone, right? Which is a dominant phone in the enterprise now. Blackberry won, was winning and then the iPhone showed up with absolutely zero enterprise features but it was beautiful and kind of spoke to all of us and people just snuck it in and it took over. And there's this generational thing now, there's a bunch of people in their kind of mid thirties that are buying software at enterprises that grew up in a generation that got to use actual nice dev tools like Rails and GitHub and all of these things, and people just don't want to look at ugly, terrible, hard to use enterprise software. And sadly the database market is probably the most criminal when it comes to this, databases are incredibly unintuitive, ugly and just generally not built for humans, and we wanted to change that.

Sam Lambert (11:00):
So you can just justify it as a great business case because it clearly resounds with people and we shifted the game forward very quickly, but on a deeper level, it just feels the right way for all of us emotionally to build things that actually inspire us. Like our lead designer Jason Long, who is phenomenally talented and has a very long career in product design and designed some of the earliest pieces of GitHub. I was telling him- the 30 seconds where he says, "Hang on a second, I'm just going to share my screen to show what's in Figma." I get butterflies. I get so excited just to see what he's come up with because we work so diligently to create something amazing, and I want it to feel exciting and magical and powerful to everyone that uses the product because I have finite time on earth like we all do, and I want to look at the things that we build and say, that is actually beautiful and feels great to use because why the hell not.

Sam Lambert (11:59):
Truthfully, and when we see people sharing, just even our marketing imagery and saying, wow, this is incredible. When people share our design of our CLI's and say, "Oh wow, I need to redesign my CLI that way." When you consider a database company and that that's happening and the dev tools are up in their game because of what we've put in the world, it makes me really excited. It's great for bringing adoption to the platform, and I think it's the only way I'd want to build products.

Michael Grinich (12:26):
I think for people that are drawn to that type of work, also, they almost can't work in an environment that doesn't have that. I certainly feel like that too.

Sam Lambert (12:33):
Exactly. In the beginning, when I was thinking about coming to PlanetScale and what I wanted to do, and when I was looking at all of these other databases that are out there that were just horrendous. I just thought, how do people feel excited using these tools? And then when people do get excited using your product, it just feels incredible. And I also struggled and a lot of people do struggle to use things that are unintuitive and difficult and we shouldn't put people through that.

Michael Grinich (12:57):
I think there's absolutely no reason enterprise software can't be as beautiful, seamless, polished as the consumer products that are out there. There's no reason.

Sam Lambert (13:05):
I 100% agree. And that's why we're happy WorkOS customers as well. Because when I asked the team, how was the experience? They were just like, this is incredible. It's an enterprise product, right? It's actually table stakes. If you're building for other enterprises, you have to use this tooling. And then for it to be a delightful developer experience, get everyone towards their goals quicker, everybody. Everybody wins when the management of the company says, right, now it's time to go with the enterprise requirements, right? It's not going to be the most shared thing on Twitter all of the time. It's not as cool as some of the features that we're shipping now or have shipped, but it's essential. It's so critical to unlock and enable. And when they find it surprisingly delightful and awesome to build against, just everybody wins, it's a complete win-win situation.

Michael Grinich (13:56):
We focus a lot on the speed of doing that. And I know PlanetScale does this too. The speed of getting started. Can you talk through that? How you frame that in terms of the day zero developer experience versus maybe the day 1000 developer experience?

Sam Lambert (14:10):
Yeah. So speed is so critical. And I really learned that at GitHub. When I joined there was this design of GitHub, which was this kind of set of, I wouldn't say rules, but beliefs that the company had. And one of them is that it's not shipped until it's fast. And there was this obsession with making things really quick, and people were always saying it's an amazingly quick website for what it does. And we really realized that it's a feature. And when people can dance around the product, they can just shift through what they're doing so quickly it feels like you're playing, it feels like play. It feels like excitement.

Sam Lambert (14:50):
So when we started to build PlanetScale, I said, database creation should be instant. And I think a few people thought like why? You only do it once, who cares? Let's focus on how we're going to be running this thing forever, and it's like, it's true, we should do both. Because when you have an idea for the next Stripe in your head, and you are in that moment of inspiration, you do not want to spend time waiting for a database to provision. You want to get past that because the database is a tool. It's a means to an end. And that end is your idea. And that thing you want to create, and the database is essential.

Sam Lambert (15:25):
When I talk to people about the opportunity that PlanetScale has, the fact that every company needs a database, it's just essential. You can't really be doing anything meaningful as a technical project without a way to store data. So first of all, it's a massive opportunity. It's awesome. So it's essential, and because we've kind of captured people that way, why make it painful?

Sam Lambert (15:46):
When I was looking at joining I thought I'd do some market research and the most talked about database at that time took 24 minutes to provision. And I honestly forgot what I was doing, after 24 minutes I kind of got up, made a coffee, phone rang or got stuck on Twitter or whatever, and it's gone, that moment of extreme excitement to build something just dies in that 24 minutes. It's just like if your notes app took forever to load, you'd lose information that you should be jotting down on your phone. So we wanted to make it instant so that you can move past creating the database and you can get to doing what you did and more companies need to see their database as a tool for creation. Once we did that, it became very quickly part of the culture and now we just obsess over it. People get woken up. They're paged if database creation or branch creation takes too long, because it's something that we care about. The product is in their hands at that point and it should feel high quality and speed is a real tenant of that.

Michael Grinich (16:41):
I'd love for you to talk more about how you build stuff at PlanetScale. I remember reading this tweet from you that product managers spend too much time trying to create a rigid process, and that process instead should be really creative and flexible and playful. How do you think about balancing that, getting ideas to kind of come from the bottom up or from anyone, but at the same time needing to have a roadmap and be shipping and moving towards your commercial goals?

Sam Lambert (17:07):
Great question. So it's all very different for different things. It depends with exactly what we're doing, and it's about listening to people for what their core competencies are. But also it's about creating a constant unresolvable but healthy tension within the company. So we have this degree of unreasonableness across the company. We like to be unreasonable. So I'll listen to a complete different spectrum of customers. I will listen to enterprise companies that have 10,000 developers, terabytes of data, every compliance requirement possible on planet earth and all of that stuff, and then I will listen to people that have been software developers for one year and just say, can't this be instant, and can't databases just work this way? Because that level of unreasonableness is perfect. The more you learn, you lose that naivety and some of the greatest products have been born out of just frustration and naivety.

Sam Lambert (18:06):
So I want to listen to all of it and think about it, right? And think how you deliver for people. And the question is why? Why can't it be instant? Why can't it just work that way? Why can't it just scale magically? And so those are great questions to ask ourselves, and we then try and build up a similar tension inside the company. So we have backend database engineers that worked on literally some of the largest data employments that have ever run in the history of tech.

Sam Lambert (18:32):
The people that build the backend database for YouTube that ran on tens of thousands of servers, like massive petabytes of data. We have people that were early engineers at Instagram and Amazon. All of these people that have just seen the biggest technical deployments we've ever come up with as an industry. And all of the pitfalls, all of the nasty narleyness and understand all of the rules of databases and how everything kind of is difficult and tough. And those people are phenomenally talented, they work so hard to deliver something that's so high quality and something that's so difficult to build.

Sam Lambert (19:07):
And then we have a team called the surfaces team that build all of the surfaces that are exposed to the users, to the API, the CLI and the front end of the site and we have the product designers as well. Our product designers have really very little database expertise, and that's fine. If you let the backend team tell the product designers how it should look, you end up with ugly products, right? And if you had product designers designing databases, it wouldn't work very well. So then we work really hard internally, with people that have high context and kind of an appreciation for design and knowledge of databases to create that space where that tension can just continually live, and we then try and create something that has the user in mind that meets the need of databases, because you can't really compromise when you're storing people's data.

Sam Lambert (19:54):
And so we kind of have magicians and believers and all of these people, and it all kind of comes together and it's quite chaotic. I'm sure it will drive some people utterly crazy to work at PlanetScale, but for the right type of people and the people that want to resolve the fact that they've been building software for many years and they've never have been excited by a database, have dreaded using it, and you ask them what they would look for while they were building things. It kind of just all comes together. It's really fun.

Michael Grinich (20:19):
Where does the playfulness come in? That was a very specific word I think that you had used. How is that expressed?

Sam Lambert (20:25):
It's expressed in so many different ways. First of all, we just don't try and take ourselves too seriously. I want to build a company with people that have a sense of humor because crazy shit happens, really crazy things happen. I remember we were under attack by a nation state at GitHub and the site had been down for a significant period of time and we were having all of these terrible issues. And it was stressful but the team kind of ended up turning into, wow, I mean, this is never going to happen again or has never happened before, and it's kind of weird and crazy. Let's just enjoy it and experience it, right? And it was really super stressful, but you look back at it and you think, wow, that is a memory that you are lucky enough to have. You get to do something significant and fun.

Sam Lambert (21:07):
And sometimes stuff goes crazy wrong and you just have to laugh at it. So first of all, the fun is just because we have fun. We laugh, we are a bit silly and we try and have a sense of humor. But also we make time to just play with design, play with the look and feel, break the stuff that we've used. We don't clutch on to our work as if it's the only thing that represents us. It's just a product of what we knew at the time, what we thought we could build at the time, and we kind of just make it happen that way without taking it overly too seriously, and if we need to redo stuff we just throw it away. And it's not like someone's performance review is attached to this thing living forever. And that's what happens at bad companies.

Sam Lambert (21:51):
It's like, we overly rotate on impact, so now because of a bad organization chain or a rubbish leader, we've been told to do something, shipped it. It's the wrong idea. But if I don't push it all the way through to production and just keep going, I won't have anything to show on the impact scale for the performance review process. All of those incentives are wrong. Our incentive is that we will just keep learning. We will keep growing. We will build, we will throw away. We'll redo things that we don't like, and we'll just hold a high standard and just make it like we are playing, we are experimenting. We just don't take it too seriously.

Michael Grinich (22:29):
Speaking of learning, I know that you all hired a VP of sales last year, I think last spring. What's been the most valuable thing that you've learned about building a sales team since then?

Sam Lambert (22:39):
Sales teams are fascinating. And I see a lot of leaders and founders go through learning about sales teams, and I've had to learn about how sales teams work. It's a really hard job. First of all, working in sales is really tough. It takes determination and you can get real sort of highs and lows from doing it. Unfortunately, there's a sort of a trend in tech where engineers diminish the role of sales teams. They think that they create this amazing thing and it can just be sold. And actually there's a lot that goes into building sellable software. Software that meets the needs of the people that are buying it. And it's a two way process. You have to give the sales team a seat at the table to communicate back the needs of the buyer. You learn in a lot of ways how your product is deficient and how it's wonderful, and what is really resonant with the buyer and how to double down on those things.

Sam Lambert (23:43):
You also just learn that it's very strategic, and sales teams have very well defined roles. If you get it right, a sale moves through a process and as seamlessly as you can make that process, the better experience it is for the customer and the less chance you lose the sale. And it's a very, very interesting type of organizational structure. And I would recommend, if you are listening to this and you are curious about a role that is in tech, that is completely completely essential, and you are an engineer and you want to learn about how this essential role and function works. Dissecting how a sales team works is really, really fascinating and interesting. It's a great lesson in incentive structures as well.

Michael Grinich (24:28):
I've been digging into it as well and it's just as much of a system as engineering is, which is just a fascinating thing. It's really just about incentives and structures. It's the same in design.

Sam Lambert (24:37):
I completely agree. And it's so objective in terms of how the outcomes are laid up. There's a real art to it and some of the best sales leaders just are masters at incentive structures and understanding them.

Michael Grinich (24:51):
Sam, I think we're keeping you for a little too long. Last question before we wrap up. A lot of people listening to this podcast are probably folks similar to you, but years ago, I'm curious what advice you might give to early stage entrepreneurs, founders, engineers, folks that are building a new B2B product, that haven't yet gone up market to cross the enterprise chasm, but are looking towards that?

Sam Lambert (25:13):
Stay curious about everything and learn as much as you possibly can. The one way to do that is go and talk to people who have done it before and just ask them loads of questions. People don't mind. That's the thing I realized. You don't have an abundance of time when you run a company, but I always used to assume that if I reached out to somebody that was busy and powerful in some big company, that I would be bugging them or irritating them. But now I'm kind of in the seat where I'm running a company, and someone reaches out and wants to talk, you actually like it. You kind of feel like you are helping someone out and it's complicated and it's really difficult. Just by talking to people you can really help them leap forward further. I wish I had known earlier that just taking your shot and emailing someone and saying, "Hey, can I buy your coffee in return for some advice?" How beneficial that would be.

Sam Lambert (26:07):
When I worked it out and I started just trying to connect with people that were just way above me in career and stature and everything, and asking them for advice, everything changed and got so much better, and you just realize that no matter what people have done and what they've achieved, they're human with all of the failures and foibles that we have as humans, and we are all going through a similar experience and you can learn from those people and to show up with a notebook, ask a lot of questions. Be respectful of people's time, but there's so much that can be learned out of there from so many great people.

Michael Grinich (26:40):
Well, Sam, I think that's a great place to wrap up. Thanks again for answering my questions today. This has been fantastic and super excited to see how PlanetScale continues to grow in the years to come.

Sam Lambert (26:51):
Thank you so much for having us.

Michael Grinich (26:58):
You just listened to Crossing the Enterprise Chasm, a podcast about software startups and their journey moving upmarket to serving enterprise customers. Want to learn more about becoming enterprise ready? The WorkOS blog is full of tons of articles and guides outlining best practices for adding features like single sign-on, SCIM provisioning and more to your app. Also make sure to subscribe to this podcast, so you're first to hear about new episodes with more founders and product leads of fast growing startups. I'm Michael Grinich, founder of WorkOS. Thanks so much for listening and see you next time.

‍