November 24, 2014
Units of Value: A Framework for Scaling
As a Partner at Greylock, Jerry invests in new enterprise apps, and cloud and application infrastructure including distributed app platform ...
In episode 29 of EnterpriseReady, Grant speaks with Mitchell Hashimoto of HashiCorp. They discuss Hashicorp’s founding story, fundraising tactics and pricing models, and monetizing open source projects.
About the Guests
Grant Miller: Alright, Mitchell, thank you so much for joining.
Mitchell Hashimoto: Thank you for having me.
Grant: Cool, let's dive right in. Tell us a little bit about your background and kind of how you got into enterprise software.
Mitchell: Sure, I got into it by accident, I'm sure a lot of people say that.
I started building open source software, build some software called Vagrant got reasonably popular.
I did not build it for enterprises I never worked at and still never worked at an enterprise, that wasn't our customer at the time, the company was working on.
I really built it for developers and SMB and things like that.
And just over time, building a company realized sort of based on open source software, and the type of software we're building Enterprise was a really good way to monetize that.
Both as a business but also as a way to not cause much friction with our open source communities.
And I got into it back then and had to learn along the way.
Grant: Okay, so you went to school for computer science
Grant: And you studied computer science at Washington?
Mitchell: University of Washington.
Grant: Great and then you had a couple of jobs, it sounds like more consumer companies or what were they focused on?
Mitchell: Yep, I worked at a consultancy to sort of spin out websites every three months type of consultancy and then a worked at an ad tech startup in San Francisco.
Grant: Okay, cool and you built Vagrant, then sort of to scratch your own itch like a nights and weekends project, how did it start?
Mitchell: Nights and weekends while I was in college, it was a mix between challenges I was seeing at the consultant job as well as challenges I was seeing during undergraduate research at UW and just things I was getting on my own side projects so I scrapped on my side projects are working on this as a side project, but it's totally a nights and weekend thing.
Grant: Okay, and so you've kind of always been doing development work and you told me some stories about building things in college that were like apps and other pieces to sort of help automate workflows.
And so you're kind of very familiar with the developer workflow kind of still sees that kind of where this idea kind of sprung from?
Mitchell: Yeah, I like to think that I still am, I still do a lot of stuff for fun, but especially then, all day every day, I was either building websites or building desktop software, or just building whatever I was working on the side.
So I had this very large lens of the dev lifecycle, I guess.
Grant: And so talk about Vagrant, what was the problem that was solving initially, what made it so interesting to you?
Mitchell: So with Vagrant, the problem we saw initially and this will get funny later, because we'll see later that we got really into Cloud, Clouds everything that we do, but at the time, that I built Vagrant, Vagrant had nothing to do with Cloud, it just happened to be a very powerful Cloud tool later.
But the problem I was trying to solve was for projects, how do I spin up all the dependencies they need, and get working as quickly as possible, I was running into an issue at the time, I don't know if this is still the case, I don't think that's still the case, thanks to Vagrant and Docker and so on.
But at the time, you would install like Apache and PHP and Ruby and stuff just directly on your laptop and run websites but--
Grant: But the LAMP stack or something?
Mitchell: Yeah, and if you're working on a project that was doing maintenance on one that was two years old versus one that's new, you'd have different versions of things, and those versions are fairly significant and things wouldn't work.
And it was very frustrating because to switch projects took you hours just uninstalled things, reinstalled things, save your config files is just the total chaos, so I was really frustrated.
What I saw myself was I want a way to--
I was a Windows user at the time, I was like, I want a way to double click an .exe and start the project.
Like that's what I want, the development environment for the project. And that was sort of when I went off with.
When I started Vagrant, what's interesting is I had no idea what a virtual machine was.
So that ended up being the solution that I chose, when I decided this was a problem to solve I didn't know about that.
Grant: And what year was it? - 2009
Grant: Okay, so the VMs have been around, right?
Mitchell: Oh, yeah I was just like, 18 or 19 I know what that was.
Grant: Funny, okay so then eventually you integrated pretty closely with virtual machines.
Mitchell: Yeah, the first version, it was like virtual machines was the solution, I knew I want to solve this problem.
And I was like, Okay, how do I sandbox things? And it was actually Armon, who's now my co-founder for this company that told me "Oh, you should look at virtual machines."
I had no idea what that was went home googled virtual machines, found VirtualBox.
Virtual Box is actually critical 'cause being a college student, having very little money, VirtualBox is basically the only free virtualization that you could get that was point and click easy to use.
And so VirtualBox was critical I downloaded that and I was like, okay, this could work, I just don't know if I could automate this, then went on digging and found a way to do it.
And that's what Vagrant was built around.
Grant: So VirtualBox was an open source?
Mitchell: open source Sun project at the time, eventually became an Oracle project but at the time of the Sun still.
Grant: And so if I remember correctly, the initial monetization of Vagrant was, you would pay if you wanted to use VMware,
Mitchell: Right, so that was two years later.
So for two years, I grew Vagrant, just on its own. It was never a full time thing.
I always had a job I graduated, moved, got another job, like all during this time, and just let Vagrant steadily grow.
And over that time, it grew to over a million downloads a year, which is tiny compared to today, but at the time it was a million dollars a year.
And my first stab when I quit my job and started HashiCorp was I'm a pretty financially conservative person.
So I was like, Oh my god, I need to make money as soon as possible otherwise, how am I going to eat even though I would have been fine for a little while.
And so I thought, okay, people have been wanting-- VirtualBoxes is kind of subpar to put it politely in a variety of ways.
And VMware is quite good in those same weaknesses VirtualBox has.
So I looked at supporting multiple, I call them providers or virtualization providers and VMware, I felt VMware's a commercial product, all-in.
This is maybe a good opportunity to charge people it seems to make sense that I wouldn't give away an integration for free that integrates with a commercial product.
People seem to agree with that so I started charging for that.
That was really important because it actually paid multiple salaries for the first couple years of our company.
Grant: Okay, so when did Armon come into picture here you guys went to school together, so, you mentioned he kind of helped give you some of these ideas, but he wasn't working on Vagrant at the very beginning.
Mitchell: He did not work on Vagrant.
So he was my best friend in college so he was aware of Vagrant the whole time.
Actually really funny story is I was so excited about Vagrant, we had lunch every day, I went to lunch one time, and I pulled out my laptop, so I got to show you something really cool and I showed a Vagrant and he was like, "Ah I don't know."
He was not into it initially, within a couple of years he was using it for everything.
He loved it, but it's just his initial reaction he did not like it.
But, so we would talk every day, the key sort of part of our relationship at the time that was important was that he was on the same undergraduate research project as me, and that research project was focused around --
You wouldn't call it Cloud at the time it was just server automation, the servers in the university were donated and because servers with an API were a new thing that were coming up, and we're in Seattle, Amazon gave us AWS credits, and that was very foreign to us, but we had those.
Microsoft had something at the time called "data center in a box," or something.
They put a literal container on the lawn of our computer science building, and had a bunch of servers in it and we could access those, and then Google also donated some resources and then we had some really terrible physical machines in the basement of the computer science building.
And my job along with Armon was to figure out a way to deploy secure network.
We were the ops people, I guess, for all those servers, while other people were writing Python apps, Rails apps, like C++, like desktop software, all this stuff, like how do we use these servers to enable those people?
We totally failed but what we came out of that was, understanding the problem.
Mitchell: And loving it to be honest
Grant: I know it was hard.
Mitchell: It's a weird thing to love, but like we just loved solving those problems and so when we went into industry, that's sort of the direction we ended up going, and that's where the overlap starts happening between Vagrant, coincidentally helping the DevOps community, the fledgling DevOps community.
Armon realizing this, what would be called Cloud problem or multi Cloud problem years later, and then me sort of loving the challenge so that's where all the overlap happened within those two years.
Grant: Okay, cool, so you have this project, it's getting adopted, you're not going to make any money off of it yet. You're working nights and weekends, you decided to start HashiCorp with Armon at the same time that you started to charge for it or how did that?
Mitchell: Yeah, the timings a little strange there but we talked about it. But the problem was we were sort of the only two senior engineers at the company, and we didn't want to just leave, we cared about this startup--
Grant: Or at the startup you're at.
Mitchell: Yeah and we didn't keep we didn't want to leave them. I think it would have been fine with struggle, and we didn't want to leave them with that struggle.
Grant: And you we're both working there together.
Grant: Like the two most senior engineers, and you're both somewhat young, but your senior in terms of skill set, and you're leading a lot of projects and you're like, well, we could leave and start our own thing.
But then we're going to leave these people we care about in a lurch.
Mitchell: So we decided I would leave, start working on some stuff, we would always talk about it the background, and then Armon would notify that he plan to leave but it would be a very long notice like six months, in order to hire and help people get started.
And also that even though I quit, I would keep working out of that office so that I could answer any questions that the new hires that took over my stuff had.
So I wasn't doing anything day to day on Keep's stack, but if the person across the desk from me had a question they could ask me, that was sort of part of the deal.
Also, the co-founder CTO of Keep was one of our first angels he was super supportive of what we're doing, he actually convinced me to quit, but in convincing me he's like, I'm kind of shooting myself in the foot can you help.
I was very much open to doing that, and that's how that timing worked out. So during that six month period is when I came out with the VMware thing and start charging for it, during the six month period is also when I started the project called Packer.
Mitchell: So that all happened without Armon directly hands on keyboard, but we were again, still having dinner every day and things like that.
Grant: And so Packer was the second project.
Grant: And Packer was sort of package up things into a VM is that the--
Mitchell: It was... The idea was one workflow to create any image the kind of the idea.
It was-- creating AMIs was hard, creating VirtualBox and VMware images for Vagrant was really hard.
And those were the only three initial targets, and then now it supports 40, 50 output targets, but everything had its own special snowflake workflow and API calls and inputs and all sorts of stuff.
And I wanted to try to unify that as much as possible.
Grant: And so you had at least one angel investor, maybe a couple others came in or how did you take funding?
Mitchell: Yes, I didn't take funding actually, for a while, so I think it's a cool story.
The day that I told my boss at Keep that I was going to quit, he actually had a check already that he had filled out for a certain amount of money and he was really happy that I was doing this and he handed me the check when I noticed and he's like, "You don't need to cash it."
But if you want to cash it, whatever valuation you're at, "I want to put in money."
I didn't end up cashing that check, because checks expire after like 90 days or something. And I didn't fund raise within 90 days, I wasn't sure if I wanted to fund raise.
Mitchell: So again, didn't know, this was going to be an enterprise company, we didn't even know if this was going to be a lifestyle business or a venture backed startup, we weren't sure which direction we were going to go.
And it got even more complicated when we built the VMware thing, because within a couple months, the VMware thing was making about $400-- $500,000 a year.
And so there was definitely a question of, do me and Armon, just sit on that and just do whatever the hell we want, from a program perspective, not like go sit on the beach.To just build projects and just have fun, and just sit on this lifestyle, and maybe we could make that 500,000 into a million into 1.5, is this lifestyle we're going for, and that was all these existential questions and ultimately, we decided, the venture backed route, world domination, etc.
Grant: And so the venture background was that it all informed maybe Vagrant isn't going to be the solution for an eternity, it probably has a finite life, shelf life, and maybe we should try to kind of get ahead of the market and build where the market should go?
What were the other impacts that sort of helped you decide?
Mitchell: I think there's a few inputs on it, I think one of them was, what you pointed out was, we wanted to do a lot, and we didn't see how we could alone do all that stuff that we want to do.
We didn't think that it was five people that we needed either we thought it was more like 50 people that we needed, which is also a joke now because we're much larger than that.
But we thought it was like 50 people and we just didn't see out, we did some back-of-the-napkin math, which was way off at the time was like 50 people, maybe we need $10 million a year or something and it's not going to work out and we didn't see how that was going to work.
The other part of it was, we did want to have an impact, and we weren't sure how big an impact individuals can have with products that are great, we hope to be great, but we knew that companies would need help and marketing and they're not just going to adopt it 'cause it's the right thing, so we had some fears there.
And then I think the last part was, why not we were at the time 22 and it was sort of--
I'm very risk averse but it was like, there's a time to take a risk being risk averse, this is the time to do that risk, so let's just go big on this one and if it doesn't work out, I'm definitely never doing it again because I'm risk averse, is like let's do it.
And my parents were actually for me personally, were a big deal because my parents, unsurprisingly, are very risk averse that's why I'm risk averse now that works, and I went and talked to my parents during a holiday and said "I'm thinking about quitting my job and starting this company."
And I was really sure, 100% sure that they would be like, "No you have a great stable job, you have health benefits, "you're on a good path, you just keep doing it."
To my surprise, I don't think I even got to finish saying what I was thinking of maybe doing my dad was like "You need to quit your job tomorrow."
Mitchell: And I was like, what? And so I went home, and I was like, if they think I should quit over this, then I probably should, and so that was a big part.
Grant: That's awesome, and you knew that you wanted Armon as your co-founder--
Mitchell: 100%, it's hard to convey over voice or without meeting him but I think anyone who meets him realizes really quickly, he's a force. And he complements me in very different ways, and it's very important that he was involved.
Grant: So you started this company, it's going well, and I guess we think about it from the enterprise perspective, I'm guessing you had customers that were at big companies, but you're selling individual licenses or something?
Mitchell: And, I would contend, going well, I don't think it was going well, for a long time.
I don't think it's going poorly, in hindsight now talk to our investors and stuff.
And our investors, said to that for about three or four years of the company, every board meeting, they were like, "what's this company ever going to do?"
They're very concerned that would be nothing more than just the money pit where we built some great things and R&D came out of it, but it wasn't a venture success and there was a lot of fear.
From a non venture standpoint, making half million dollars a year on a VMware thing is great, because we could pay three people, four people with benefits poorly, I guess, you could do that but from a venture standpoint, if that 500 is not turning into a million, turning into 2 million, and so on, like, it gets stressful on that side, and we didn't do that for a very long time.
Grant: Okay, so you start to sell this thing, but you're selling it to like individual licenses, you're not really selling it like as an enterprise, here's some management piece, here's something else, it's just like, you're one license at a time.
Mitchell: Yeah, so here's where us being naive and young and ignorant and not knowing what we're trying to do, comes into play as our initial business plan was we'll create this Saas management thing, and our investors being much more wise than us asked us, "Who's your customer, who are you selling to?"
And our answer was "Everybody, anybody who will buy this, not realizing that building a company that sells to small businesses versus enterprises versus different segments is totally different.
And so we built this thing and what we ended up building was something that nobody really wanted because it checks some of the boxes for enterprises, checks on the boxes for SMB, pricing was wrong here or there, all sorts of stuff. And also the features weren't very good, so--
Grant: What was it called?
Grant: Okay, Atlas.
Mitchell: Yeah, so did some things really well, and people wanted to buy those things, but you had to buy the whole product.
And so they're like, okay, we want to buy this feature for let's just make up numbers for $5,000, but you're trying to sell us this whole platform for $100,000.
And Enterprise was concerned a lot of other ways the price was fine, and their startup was, I'm not paying you $100,000 for $5,000 feature.
So there's all sorts of problems intermix there, and is actually a very singular moment when we decided to be an enterprise company.
We just had probably our worst board meeting we ever had in our history currently.
But as a Friday board meeting is late, I remember driving back to San Francisco it was very quiet drive back with Armon back to the city, we went to our office to talk about it and they didn't have to say anything in the board meeting, there was no outward negativity, but you could just feel that things weren't going well.
And it just feels bad. And so we went to our office to talk about it, and we sort of challenged ourselves that maybe we have a bit of a sunk cost fallacy going on that we feel like we built this thing and we have to make it work.
We just decided, what if Atlas didn't exist, what would we build today, taking back what we learned, not just technically, but as a business.
And I think we came up with a few things, which was we need smaller, more focused products is what we build rather than this big platform.
I think there was a little bit too much contention with the open source community on the Alas side, there was stuff that they wanted in there, it's like, how do we avoid that?
I think the way we thought to avoid that was the features for the largest companies in the world, the people adopting open source generally aren't doing that.
And so, that sort of guided us towards okay, let's build individual products for enterprise, that's what we're going to do it.
We were small enough at the time that was literally the decision, I think that would take a year now.
But that was Friday, I think we call our investors over the weekend, we called an all hands with the company on Monday, and we said, we're going to be an enterprise company now, all employees were actually really hyped because it was sort of the most specific direction we've ever had as a business.
We were afraid that they'd be like, No I came here to work on cool flashy things, and maybe that's not that although, I think it's pretty cool.
But no, everyone was supportive, the investors I think, again, same thing, they just were like, awesome direction and also direction that they could be helpful with, now they have a direction, we could tell you, who we think you should hire, and what you should be thinking about as a company like that--
It's just impossible to do when your business is anything for anybody.
Grant: And so, at that point you had--
Mitchell: It was five years ago,
Grant: It was five years ago, you had a couple of products at that point, not just--
Mitchell: open source?
Mitchell: We had Vagrant, Packer, Serf we had almost all of them. We had all of them except Vault, I don't know, I'm sure people on the podcast won't know what Terraform is.
But we have a couple of open source projects now that you might call de facto industry standards.
Mitchell: And they look like overnight successes, but Terraform actually was a very long burn so Terraform was not doing very well at the time.
Grant: Okay, and so it just described what Terraform is.
Mitchell: Terraform is a way to manage infrastructures code, so automate the creation of servers, creation of networking equipment, things like that.
But on any platform, so we're created to Amazon works for Microsoft's VMware, On-Prem Cloud, everything.
But, it was not successful and honestly not a great product for about two years.
Grant: You actually believe that it was the product's problem, not just like a market readiness problem?
Grant: Okay, so an Atlas kind of like wrapped up Terraform plus Vagrant end suite for developing what, how do you describe what Atlas was?
Mitchell: The way we would describe it was we built all these little component pieces, and this puts it together into a platform so that you could just deploy software.
In reality, it wasn't that easy, wasn't that good, and we're missing a bunch of stuff that you needed anyway.
Grant: Okay, and the other thing context I think is probably important is realistically, like HashiCorp is kind of one of these very early developer focused companies, and like GitHub started around the same time that you've created Vagrant, I think maybe they were--
Mitchell: They did, actually, I think I was on GitHub fairly early, not as early as some other people, but I think I was still user less than 1000.
Grant: So like, it wasn't quite obvious then that developer tooling, and all of this stuff that we sort of see as these like big companies, and this big opportunity, this big ecosystem, wasn't very obvious that that's going to be a big thing.
There's a lot of belief in the ecosystem that you couldn't make money from developers, they wouldn't pay for stuff, they didn't have access to big capital.
And so it was not necessarily like an obvious play, right?
Mitchell: And I would say the environment at the time was almost actively hostile to both open source which was a challenge and to DevOps as a category because it's easier to see now that it's a huge successful category.
But at the time, our competition when we went to raise or pitch this thing was they compared us to Chef or Puppet and I love Chef and Puppet as products, but from a VC standpoint, we heard this from multiple VCs, it was like they're doing okay, as a business, they're just not venture successes.
I'm quoting somebody else, that's the way they put it.
And so people were giving us really bad valuations and really bad terms, they're like, "Okay, well we're putting money, give us 40% of your company, or something."
Just really terrible terms, so we weren't taking those and it was a bad time.
And the thing that saved it honestly, is actually Docker, there's various parts in our history where we've been contentious with that company.
We're not anymore it's a great technology but there's various parts where we were contentious, but for all the negative things, the positive thing that they did, sort of ironically, if they viewed us as a competitor at the time, was they made it so much easier for us to raise and gave us great valuations just by being adjacent to them.
Grant: Oh, really?
Mitchell: Yeah, it was huge, it was like a overnight, over six month change, we're trying to raise before and trying to raise after the nuclear explosion of Dockers launch was so much easier.
Grant: In theory--
Mitchell: Although annoying because every pitch was like, "Are you a Docker competitor in a positive way, "we want you to kill Docker."
And we're like, "no, no, no, no, no, that's not what we're trying to do, but, helped us raise."
Grant: It also theoretically, one of the core things, particularly about Terraform and sort of, infrastructures code, I think that containerization sort of created this movement of more declarative, sort of infrastructure, which for the audience, convergence is sort of the other possibility here, and that's kind of what Chef and Puppet did?
Mitchell: Exactly as I said, it is like, you could say a lot of things and analyzing those companies, but they had a lot of the concepts e xactly right. And their core concepts are what power the industry now.
Grant: Great, automation and everything else.
Mitchell: Like convergences are the core infrastructure idea like that is Terraform that is Kubernetes, they got that right declarative same thing.
Grant: I guess my thought is, it sort of helped, there was a shift with Docker and for the way that the infrastructure was described and applications were described, and I think, ultimately, the infrastructure is code, I think of that as this really important trend, or sort of concept that you really led and became that reproducibility and this idea, I think, like Docker helped it kind of push that along a bit as well.
Grant: And your company, just like, it really benefited from the attention in DevOps, everything else to a point, right?
Mitchell: Exactly. That's exactly what happened.
I think what we did poorly, we still don't do as we're getting a lot better in the past couple years.
But, we're such a heads down, quiet working company that we were never viewed except it's very specific circles, we were never viewed as the exciting flashy company, to our detriment.
Even though I think the reason Docker had a fire sale, basically now, most of the startups that we came up with don't exist anymore.
And the reason we still exists and we're doing much better today I think is due to that, is because of when everyone came up out of various hype things they're like, this company is just been quietly executing, for a decade, basically.
But being part of it just helped us a lot.
Grant: And, I will liken this back to one of your core values, which I think is like pragmatism as well?
Grant: What do you call it the like, what's the HashiCorp like?
Grant: The Tao
Mitchell: Pragmatism is, I mean, I self described myself as like painfully pragmatic.
I definitely get in friendly arguments with a lot of people just because I'm overly pragmatic, there's a Japanese word, phrase, I guess "shoganai." Which is basically, I think it literally translates to, "it cannot be helped."
And I use that word all time drives my wife crazy, drives everyone crazy 'cause it's like, something terrible will happen, and people will be so emotional or something.
And I'll just be like, "shogani ."I'm not going to get emotional about this is just, and but taking that to products and stuff it's like we just got to do this.
Like, this isn't the way to do it, like there's a very pure, theoretically great way to do something, and there's just let's just solve a problem and always choose to solve the problem side.
Grant: Yeah, and I think that's what industry needs. They need solutions, you do have this, great academic background and Armon you guys, you're super, very smart academically, but you also just like opt towards like, okay, well, that's good theory, let's actually get something executed in the market that like people can use.
Mitchell: Yeah, I don't know where that comes from, but we try to solve real problems, I guess is how I put it.
Grant: Yeah, and I think it kind of we talked about heads down like that kind of like, well, let's just go solve the next problem.
Grant: Move on to next solve, the next solve the next one.
Mitchell: Yeah, and I think like a very real example I could give is today is like service mesh, service mesh is the I'm sure everyone's heard that word.
It's listen to this, it's a very hyped word, and there's a lot of companies out there trying to sell service mesh, and we have a service mesh too, and we do the same thing.
But when we go into companies actually sell this thing we realized that in the fortune 500, offices of CTO, experimental teams, super hyped about service mesh, or the team that's super focused on like Kubernetes adoption, hyped about service mesh, that's still even today, and I think it'll get much bigger.
But today is like, optimistically 10% of the workloads, so it gets optimism, I think it's more closer to five.
And so the rest of the company is just annoyed that people are saying service mesh, because they have real problems that that's just noise to them, and so we actually sell Consul, the biggest reason we sell Consul today is service catalog.
It's just going to a CIO and saying, can you even tell us what are all the services and servers that exist globally in your company?
And, he's like no, she's like no, it's like, if you can't do that, how are you going to route traffic to them?
If you don't know, they even exist, you can't talk them, and so that's step one, and so we sell deals that way.
So I could say undoubtedly, like Consul is probably the highest deployed and highest revenue service mesh that exists but it's probably like one of the lower utilized for service mesh it's all the other features.
But the thing is they're heading towards that, there this is incremental adoption and I think incremental adoption is really not sexy.
And it's very, not flashy so we pitch this idea of, we actually call it crawl, walk, run, we have sales plays around it, we have marketing plays around it, service message run.
So like crawl, walk, what are the steps to get there, and it's basically service catalog, service discovery, that's crawl, walk, and people think they could get a service mesh without having service discovery.
It's like, you got to get there first and if you're in Kubernetes you have that but if you're anywhere else, you don't have that for free.
And so we're selling these sales plays and sort of paving the way there and doing well financially because of it, but, you see in the service mesh market, that there's a lot more noise for other people.
That's a concrete example, that you could extend to basically every one of our products that has happened at a certain point
Grant: Interesting, just for context, let's just go over the kind of primary products that you have.
Grant: It's probably just helpful, so Vagrant sort of the first one. Like we kind of talked about that one already. Is that still like a fairly popular product?
Mitchell: It's actually huge, it's still gets a-- I just figured this out yesterday, actually I think it still gets like four or 5 million downloads a year.
Grant: Oh, wow, okay, great. So from Vagrant then you mentioned Packer.
Grant: And then sort of-- - The next one's--.
Mitchell: I'll just mention a point about Packer is we don't monetize Packer at all, we don't really plan to in the short term, but it's the backbone of many Cloud services.
And Microsoft's the only one that's been very public about it, but like Microsoft as an image building service, and it's for pay Azure service, and they were very clear on the blog post, like this is Packer, this is all just Packer.
And there's a lot of other stuff that isn't public at other Clouds that it's just Packer and you just don't realize it.
Grant: Oh funny.
Mitchell: Or like any Azure service, you spin up databases service, load balancer service, those images for that service.
They build internally with Packer for their own platform, so Packer's like this quiet thing that does a lot for the industry, but we don't we try to monetize it at all.
And then, so Vagrant and Packer, and then there's Consul.
So Consul is the one I was just talking about, but it's a system for cataloguing, discovering and connecting all your services no matter what they run on.
And that's going to be a consistent pattern with all our products is we're not 100% containers, we're not 100% VMs.
It's you could bring bare metal, you could bring VMware, you could bring Cloud systems, Containers, Serverless, and we will work with all of that, that's very core to our philosophy.
The next one is Terraform. Terraform we talked about a bit earlier, the same idea bring everything up infrastructures code, no matter what it is container or server or bare metal.
Then there's Vault. Vault is our security tool so stores secrets, does certificates, does identity management, and same deal is you can get consistent API's for all that no matter where your applications are.
Then there's Nomad. And Nomad as our application scheduler and compared to something like Kubernetes is our big selling point there is, " very simple to operate," and also it's very workload agnostic.
So we could run direct executables, we could run jars directly, we could run Windows, I use the word containerized generally, we containerize it using, protection mechanisms and windows, just like CLR applications, things like that.
So that's the Nomad and that's all them I think.
Grant: Do you monetize Nomad as well.
Mitchell: Yeah, so Nomad, Vault, Terraform, and Consul are the four that we monetize.
Grant: Got it. Are they all used by like a similar end user?
Mitchell: No, and so it's actually a huge benefit for us.
So I won't assign them right now, but there's security people, there's ops people, networking people, there's developers, and all our products are sort of first adopted and bought by different groups.
So the benefit as a business for us is that, say we land a deal and it's $100,000, our expansion is in $10,000 next year, we could get a different product for another $100,000.
So we're looking at 100% growth in an account year over year rather than, 10% being successful. And I think that's where a lot of our business excitement comes from.
And we've seen that happen, we now been selling long enough where that happens, like we've had accounts go from like 1 million to like 3.5 in a year.
And it's a totally different product and that's why it went up so much.
Grant: But like, that also has some like, unique challenges.
Beause I mean, most I think the standard advice is, do one product stay focused, do one thing, do one thing, and HashiCorp would be counter, you'd be the counter example to that advice.
Mitchell: And it's hurt us in a lot of ways.
Grant: So talk about where do you think that like, what are the challenges been around that and what's allowed you to do it successfully?
Mitchell: I mean, the challenges from a marketing standpoint, just a huge challenge is to try to make noise about this.
Like if we could focus 100% on analyst relationships, press like DevRel, conferences on one product, it'd be way easier for us, for example, we have to split budget between four to six products, depending how you count it plus new stuff R&D.
So two years ago, we raised like 100 million dollars or thereabout, and that says a lot of money. But we joke that we're four companies in one.
So we basically raised 25 million per product at a series D level, which is like, I guess is going to be tough, right?
It's not a lot of money to work with. And so, I mean, just training people. So sales support, engineers across the board, it's overwhelming.
Grant: Do you do train everyone on all the products, should you kind of have people that are focused?
Mitchell: We're starting to specialize now, but it's only recently when we were 100 people, everyone kind of slots to do everything, which is really hard, like if you are support for the longest time had to just know everything pretty well.
And finding those people's really hard, training is really hard, they tend to get paid a lot because they're very high caliber people.
And we're now starting to specialize in this specific products. And it's in the past six months, starting at, like 600 employees it took us a while to get there,
Grant: I mean like, the go to market side, like you think about-- You've always done great design, but you have like separate websites, you have like so many web properties for these things, right?
Mitchell: It depends who you are, I mean, it definitely drives a lot of traditional marketing people crazy.
So all of our projects have dedicated domains and websites. And then our corporate website is dedicated, so there's hashicorp.com, there's terraform.io, there's consul.io, etc.
And so you can imagine from a SEO perspective, from a just user journey tracking perspective, it's very hard to make that work.
We've now transitioned the people that we have now and over the past year, we've transitioned into loving that.
Our marketing team-- actually maybe loves to strong word, but they're very supportive of it just because it gives us a really clear spot for us to have developer centric information, versus what we call technical decision maker, centric information.
So all our project sites, we assume the person going there is the user, the practitioner, and we could focus on that, and dot com and all the pages hanging off there, we could assume that you're a director or a CIO, you're a buyer, you're looking for a solution, you're not hands on keyboard directly.
And we could tailor to that, and it's made for some really clean journeys. Versus I mean, I can't think of a good example but if you go to another website that is both open source project and enterprise software, you kind of get messed up like on the homepage, you both have a download now and have a download a white paper.
It's like this person has download now is not the person that download white paper it's split up.
Grant: I mean, I think it helps to, if you think about who's going to be going to those different properties, you can then sort of target that like these are developers, these are like more decision makers and let's target because there's just different information, different things you want to put those you want to surface to those folks.
Mitchell: And I think, I've now later as HashiCorp had some unique things happen, spent more time with like traditional business school people, both that we hire, of course, but academically spend some time like a Stanford, USC, things like that.
And it's funny 'cause one person actually showed me that there is a pretty standard path that big, big companies follow and their growth and there is a step actually in that path where you diversify the portfolio.
T hat's a phase of a company, but it's usually at like the, like 500 million to billion plus revenue post IPO mark.
That's like, you're public. To continue your public gains, you need to diversify the portfolio, hedge against downturns, things like that.
And it's usually a very late stage if you put HashiCorp on there, we diversified super early, and it's starting to show benefits now, but it's definitely had some real challenges.
Grant: I mean, that's what I was actually thinking, if you look at like Cisco, or IBM, or any of these huge companies, they all have multiple product lines, and they're organized around that.
But to do that as a actual startup, and you've been doing that, even before you had series D or whatever funding you would, had all these sort of different products.
Mitchell: Yeah, and the way we sort of hack that early on was, frankly, at some products just got neglected for a very long period of time.
We just didn't have the resources so we just wouldn't, we maybe have one person looking at it.
I mean, sometimes we had zero when we're earlier like you could, see early on in our release cycles or commits, you could see like, okay, this products getting a lot of commits, than the rest are zero, okay, this one isn't the rest are zero and it was like very phased.
B ut over time, we got to dedicate more resources and now that's a benefit, because like I said Nomad, Nomad has a very clear distant number two compared to Kubernetes in terms of community, and hype and things like that.
But it does really well as a business, it's an eight figure annual business and it's like, okay, even though we're engineering is 200 plus people now, we could afford to put, six or seven people on Nomad.
It would be a venture failure if we were Nomad corp because we have this portfolio, it's actually just a really solid line item on our PNL and it's awesome.
Grant: And then give me the other advantage, right is like, you have lots of different touch points to get into a company, so it's like that I mean, the biggest challenge--
I can imagine what that and I guess this would probably go back to how you sell this stuff would be you because you're open source, you're probably waiting for people to kind of raise their hand to talk to them, rather than doing a bunch of outbound, and trying to go top down, and figure out who should be using your stuff, all right?
Mitchell: Yes, so it has that challenge.
We both don't know the users, you could download it without, we don't have your email, we don't have your name nothing, and so we don't know who they are.
We do some like clever tricks there, we do reverse IP searches on downloads and stuff to be like, okay, somebody inside Nike downloaded this, so maybe Nike is using it.
And we actually use those metrics to give our articles sales reps that have are named account sales reps will be like, you're going to focus on Nike, the Nike's a total example, you're going to focus on Nike this quarter, because they've had 200 downloads from their IP space, they're doing something you need to find it.
So we do stuff like that, but you just really can't, so we work on inbound, named accounts, which is a big one, our software, we feel confident is highly adopted enough now that even if we never seen a download, we could say, go talk to Comcast or something, we've never seen a download from them, they've never talked to us, but they're so big.
They have to be using our software somewhere in there. Go find it.
Mitchell: So that's sort of what we do.
The other challenge, though is, if you close a deal and you onboard people, is they could be coming from any state or version of the open source that's out there.
So, it's great in our contracts that we say you have to be running a version at least two years old, that's the most we'll have support for.
That's fine, but this happens every quarter, you bring a customer in and they're using a four year old version, and we need to get them from the four year version, to the enterprise version, and as you know, it's getting a little more into the go to market weeds, but you can't recognize that revenue until they're deployed and running your software.
And so it's actually very stressful to be like, okay, can we close the deal, we can't put that on our statement until they're running it.
And we got to get this four year old version to work without-- Also, if that goes down, they don't do business anymore.
So we got to do in a way that it stays up. We've never done this path before and get them going, it's tough.
Grant: But I mean, it's the same general process, it's not like every time is a bespoke sort of solution.
It's like, well, we kind of know generally how to pull people up from an older version, just use like--
Mitchell: Generally it's sort of like a one by one, we just go exactly one version at a time, we don't try anything tricky.
It's like, okay, you're 70s versions behind, counting all the releases, we're going to do 70 upgrades over the next week, is how we're going to do and that's sort of how it goes.
Grant: And then watch it every time right?
Mitchell: Yeah, it's pretty rough and we have pretty good documentation internally now of knowing like, okay, we know the upgrade between this specific version, and this one is a rough one, we need to do these precautions, but for the most part, 10 are easy and then it's like you have a rough one, and then the next 20 are easy.
So it's not too bad, but it's just the thing that makes it hard is that nine out of 10 are pretty standard, the one out of 10 that's non standard is just very tough for that customer success rep to get them going.
Grant: Okay, great, so you have all these products, and you just decide, kind of going back that Atlas isn't the right way to deliver this. But you're going to go enterprise, how did you make that transition? What does it look like?
Mitchell: So we felt we knew what it meant from a product perspective, at least enough to get started, we're like, we're going to break this down into instead of Atlas, we're going to have enterprise versions of our product. So Vault enterprise, Terraform enterprise. That was the approach we're going to take.
Grant: What does that entail? What was that going to mean?
Mitchell: So our viewpoint was we wanted to focus on sort of organizational challenges and that would be governance or role-based access control, auditing, monitoring, those sorts of things is the categories of stuff we're looking at.
Grant: Generally, like the EnterpriseReady featureset, a standard set of stuff that comes along some kind of Single sign-on maybe.
Mitchell: And so we knew from a product perspective, a rough direction we're heading, but for a business perspective we have no idea.
So the first thing we actually ended up doing, again, tied into multiple things, but over the first six months was we approached the board and asked to hire a CEO.
And that was, again due to a lot of things I generally just disliked that role. Armon didn't particularly like it either.
We thought we weren't very good at it. It was all sort of things.
So that was the first thing we did took us about a year to find the CEO, at the same time as looking for the CEO, we started a VP of sales search, we actually ended up getting them as a pair.
So Dave is still our CEO, we hired Dave, and then Rob is VP of sales, and we got him at the same time and they'd worked together in the past, and it was coincidental, they didn't realize till a little bit later in the process. "Oh you're talking about the same company," and then they decided to join.
They were like if you join all join, and it worked great because they were both people that we thought were really high caliber.
And we didn't think we'd be able to get them. And they sort of joined as a pair and they were really high caliber and so.
Grant: And where'd they work together?
Mitchell: They were together at Hortonworks
Grant: At Hortonworks
Mitchell: Maybe other places, too, but Dave was VP marketing at Hortonworks, and Rob, I don't know if he was VP of sales, but he was a sales leader at Hortonworks.
Grant: And then Dave, after that was CMO or Chief Data marketing at GitHub as well, right?
Mitchell: Yeah, he was CMO GitHub, I think he got promoted at Hortonworks just before he left, but he's a marketing leader.
But the thing that helped is that he knew how well, he didn't know, but he had a much better lens of how to build an enterprise company having worked at Hortonworks and Microsoft and VMware in the past.
And so he came in paired with our investors and said, "This is the playbook," you don't need to innovate here is-- hire VP of sales, this is what they're going to do, look at customer success, look at marketing, look at finance.
There's now a business school playbook, you could run on enterprise, the things that are wildcards are open source, DevOps, like those are the wildcards and that's what you should focus on, you should not stress about building the company, that's playbook and so that's what we started running with.
And, I think the most important things about that is just get good people in the seats.
I think, one thing I've always said is, I've never regretted paying too much to bring in someone too senior, if he or she is the right person, sort of like, personality wise, gets the mission, things like that, but you're a year too early, and the team's going to be too small.
If they understand that, and they understand the budgeting constraints and things like that, then, they can make it work, and we've never regretted that.
And so that's sort of what we did, we brought in a bunch of senior people, we brought in a whole suite of VPs that had zero or one or two reports, like there's no VP thing about it, except for the fact that they were going to build an org in the future.
And so that's what we empowered them to do, and we said "go."
It was very scary as a founder, because things were about to move so fast that we knew that we would be a bottleneck if we were in the middle of decision.
So we had the VP of sales who said, we need to hire 10 sales people, and we're not going to interview them, you have the green fly, you just get them.
And so that's what happened and suddenly 10 sales people, and suddenly there's five marketing people, that's when the company first started growing into people that we didn't interview.
Grant: There's a lot here, so from a business perspective, you basically handed over the reigns to Dave, and sort of this VP of sales and said, look like we trust you to run the playbook and go on the like on this side.
We're going to focus over here on product, and maybe some open source and some other pieces.
And then you're obviously still very involved in the business but you're going to kind of let their expertise guide you, in the enterprise go to market.
Mitchell: Right, and I think it's about hiring the right people because you alluded to our principles earlier.
One of our principles is humility, and I think that was a key thing we looked for in all these executives and I know they had it now looking back, they all had it, because they joined--
We were willing to yield all this power in a sense, autonomy, give them all this autonomy, but they had the humility to, understand the current culture, the current vision, and then very sort of kindly respectfully keep us involved in the right way say, "we're proposing these changes for these reasons" and us say, "look, we trust you and it's very helpful to bring us into this, let's go."
And so that was helpful. We didn't bring in anybody who had really strong convictions of what they were going to do.
I think sales is a really good example, a lot of the VP of sales that we interviewed, just in the interview, were like, what I'm going to do is I'm going to come in, I'm going to hire these three people, we're going to do this.
It's like, how can you know that without knowing our products?
And knowing the open source movement, we don't even know the open source movement, business movement, things like that.
And so it was very helpful to talk to Rob and say, "what I would do is this, but I really need to understand how you are."
And one thing I really liked about Rob, for example, is he asked me, "I could build a sales team that feels like anything, "but what do you want it to feel like when a HashiCorp "salesperson walks in the door."
It just demonstrated empathy in a way that was very helpful, I think, we just hired a CRO who started a month ago and one of the first things he did was, I think somebody will come in and be like, we got to hit our quarter and do all that.
He's doing a lot of that but one of the first things he did was reach out to our VP of engineering and say, "I know there's always a lot of tension between product "and sales sometimes, and I want to understand those better."
It's like, having that empathy and understanding that's important is a key part to I think success. And so that's what I would recommend.
Grant: Find someone that the answer is it depends, right?
Mitchell: Yeah, they're confident and they could execute together, but they realize different circumstances require different places and they could adapt.
Grant: Okay, so there's a couple pieces here. Number one, I think that that Dave is this really interesting person to come in, and you were kind of recruiting him for a while, you were spending six months.
Mitchell: So it was actually 18, because we talked, he actually came up before in a VP marketing search that we abandoned, we actually ended up not hiring anybody.
We had a VP marketing search and he came up and we were way too early and way too small, that's when he took the GitHub position instead.
So you can clearly see where like, the comparison is very different, we were never established, we're very young, but we loved him.
Like he was the only candidate in that whole search that we're like, we really want to hire him and during the CEO search, he popped up again, this coincidental because it's a different search firm.
They didn't know we talked him before, and he popped up again.
And we're like, "Oh, that's interesting." And it was like, I call it dating it almost felt like dating.
We're having dinners with him every two weeks, and I was getting frustrated because we're repeating a lot of same stuff but I think in hindsight, I really respect how big of a decision it was for him and I didn't understand how big of a decision it was because, he was viewing this as this is going to be very hard work, but also I'm staying here five plus years like I'm not leaving.
And so it was just a huge decision and so he really wanted to understand me and Armons personalities, but also the business, everything and so a lot of the dinners, were just repeating ourselves, he would ask us the same questions and I'd think "we said this twice, right, like, what do you need to hear?"
But he just wanted to make sure that was true and that was our conviction. So, it took 18 months, and he joined, but he's been critical.
Grant: Yeah, I mean, he's great. And I think what one comment is I always think of myself as a human YouTube video where I just repeat the same shit over and over again.
So it's like I the same, the same thing here, Dave asked the same questions, you got to get this.
Mitchell: Actually Armon and I joked about that as like, should we just record videos, and when he asked them, we just start playing them and like, eat our food.
Grant: I've thought about that many times, I could just send a YouTube video.
Grant: And then you can watch it at 2x speed like, anyway, right?
Grant: Okay, but the other thing that I think is interesting about Dave is I think enterprise marketing is very hard.
Grant: And Dave is really great at it, but I actually think that HashiCorp, the one thing that I think about, even before Dave joined, you always have like this great brand and like a really good design and presence and I think a lot of people recognized that about the company early on. So who kind of brought that to the table?
Mitchell: That's definitely me not Armon, he's just throw it up on a-- We didn't even need HTML, just throw some text on a page and they'll be fine type of person says definitely me.
I don't know where that comes from directly. I attribute a lot of that to, I worked in an apple retail store and even as a retail employee, Apple instilled quite a lot of aesthetics mattering into it.
I attribute it a little to my dad, he's a dentist, which is a weird thing but I remember growing up, my mom asked him, "Why do you care so much "if your teeth are straight," things like that, and he's a dentist and orthodontist.
But, why do you care so much, your teeth are straight?
And he said that, "It shouldn't matter, but unfortunately, "the world tends to see Your teeth first, "and if they're straight, they have a different initial perception of you."
Shouldn't be how it is, but it is how it is.
And I sort of took that to tech, which was like, what's the emotional response when you see a project?
You haven't used it yet, what's your first impression?
How does that frame you for how you're going to feel?
Are you going to go into it being like, this probably going to suck, or are you going to go into it being, this is awesome.
And then if it's a little bad, you're like, okay, it's still pretty awesome versus this is going to suck, it didn't work it does suck, it sets a different base.
And so, I was always spent too much money on design, well I don't think it's too much now, but it was a lot of money on design, and caring about things like that.
So we had a really good developer brand and aesthetic for sure. We had terrible corporate one, but a really good developer one.
Grant: And why do you think the corporate one was bad?
Mitchell: In terms of just awareness, I think visually, it was pleasing, that's what people say.
It's like, that's a pretty website, but I have no idea what you do, I have no idea why I should be interested in this, and I don't know anything about your company.
Grant: So if they went to hashicorp.com, that's what the response will be, but if they went to terraform.io and it's like they got it.
Mitchell: Well, depending who it was.
Mitchell: So actually one of our investors did this thing where, he messages a lot of people, he sent 10 CIOs or five CIOs, hashicorp.com and ask them what they thought.
And all of 100% came back and said, I have no idea what this is, and they're CIOs, they're in our space, they're there.
Grant: I mean, are they really in your space?
Grant: But like, - You know,
Mitchell: They should understand, I think, to a certain extent, if we're trying to sell to them. There are people we would try to sell to.
Grant: But you would probably never be successful with like a top down sale like that to a CIO. And so, there's this thing that you're basically bringing, you're trying to bring developers and give them the tools so that they can describe it, like up to the CIO eventually.
Mitchell: So that that was the challenge.
So we had this great bottom up movement, and the bottom up movement would get to a director, VP, CIO level and the problem was, they were saying, "I don't know why I would spend money on this "because I don't know need it."
That was where the disconnect was happening, and understandably, the practitioners didn't know what to say to their boss besides, we need this.
They would say like, "This is the problem it solves." Okay, they'll say let's say, "Vault, it solves storing secrets."
And if you say that, then the director say, "Oh goes to Gartner says what can I solve the secrets?"
And they say, well, they're cyber ark. There's this, there's that and they'll look at that and be like, this looks way more established.
This makes way more sense, we should just buy this, why are using that?
And that's where the practitioners had hard time making the case, so we needed to help as a company to do that and we weren't.
Grant: That's interesting, I mean, it's hard, that's a very complicated sort of--
Grant: Yeah, to make to say, okay, we know how to sell and get people to use this thing, that's a very hard thing to do first for people to use a thing you built.
And then to give that person the ability to describe it up the chain and help people understand why and then when they compare it to the other things. It's a hard problem.
Mitchell: Yeah, so I think our short term was, we just need to improve the landing page a little bit, which we did, it didn't need to like be the solution.
It just needed to be better. But the short term solution was exactly solved, the real problem, which is current problem, which is practitioners explaining it their CIO.
So we worked on that first and then the long term thing, though, is that it'd be great if when the practitioner said, "We need to pay for Vault," the CIO said, "Absolutely, I know what that is, and we need to pay for it."
And so that's an ongoing journey, but in the past two years have gotten a lot better of just, analyst relationships, describing Vault the right way, for better or worse, getting into the right magic quadrants.
Gartner, things like that and so that's a multipronged sort of long journey but that was where someone like Dave really helps because he came in and framed the marketing challenge we had very clearly, he's very good at frameworks, and he put up a bunch of boxes on the board and said, "HashiCorp doesn't have an answer in any of these boxes and we need a corporate mission, and this is why we need a corporate mission. Not just to have one, we need a one line, 'what is the solution that this product addresses.' And then within the solution, we need to use cases."
Like he had this framework that he just threw out. And it was, for me as a programmer, I thought that I would be annoyed by this but it was actually beautiful because he showed how this lead to product roadmaps but also mapped perfectly to the PowerPoints you'd make for sales pitches, which match perfectly to conference talks you'd give even about to developers, like this is your use case, just dive into the tech.
And it was like this grand unifying theory that was really nice, and again, the empty thing, he put it up there and said, this thing that that aligns the whole company, but nothing will ever changes as an engineering lead company.
This doesn't mean it's marketing lead, this just gives us a framework, but what the engineering group is building and what they're thinking is always going to drive the framework rather than the framework driving them. And I respected that a lot.
Grant: Yeah, I think that sort of framework helps.
Number one half the time it's just like you know the answers to these questions like in the you could check these boxes, you just need to like do a brain dump into that dock and create that narrative or create that information for somebody else to consume and turn into a thing.
And then the other part sometimes, when I've done this, like, reveals some kind of some hole.
Mitchell: Oh, yeah,
Grant: You're like, yeah, we don't really have a thing for that, we don't have a story for that and then you kind of build more product to help solve that.
Mitchell: Exactly, that definitely did that.
Grant: Okay, so that's really great insight on the product and go to market side and obviously, Dave and Rob, these guys have done a great job to sort of really build that and bring the company along on the go to market side.
But the product side I think is actually really interesting, well, it's sort of you said, like you kind of knew what to do in terms of making these enterprise, but like, how did you know was it just a bunch of customer feedback or what gave you that insight?
Mitchell: It was a mix of guess and early customer feedback, but the problem is, early customer feedback, the sample sizes like n equals three or something.
And so it's, you don't know if it's correct or not, high probability and so we looked at other software, what other software did and we'd probably going to need this.
So let's do the obvious things first. It became a lot harder later and what we actually came out of it was sort of like a framework of what goes into enterprise?
What goes in open source? What are we selling?
I think, important things that came out of that was realizing we need to sell use cases and not features, if you sell features, it gets really contentious, because you'll say, this feature should be enterprise or we need to move this feature open source.
And it just causes arguments between a bunch of people in the company.
Grant: Dive into that a bit more like how do you like define feature in that world?
Mitchell: Yeah, so one of the use cases we sell, let's start with the use cases, it's easier.
One of the use cases we sell involves, for example, is governance, governance includes policy, frameworks, things like that.
So underneath governance, there are features that are basic policies, our back, MFA, Sentinel is our policies code and more advanced policy, things like that but what we're actually selling is the ability to have stronger policy for your organization.
That's what we tell, the field teams is we're selling stronger policy, don't get caught up on the features, what you're selling that it could do this, the open source cannot do this at a use case level, it doesn't solve this use case.
And that gave us a little bit more freedom to not necessarily move features wholesale, but to let us say, like, this knob of this feature, is an enterprise never mind it's an open source and sort of dig those around.
But we're not compromising the fact that without enterprise, they wouldn't have solved that use case.
All the stuff is a little nuanced, but it just got a little bit better from an engineering side because we were proposing features and it was just falling into this argument every single time of Enterprisers was open source, and it got messy.
Grant: So is there a definitive framework or is it like, okay, only if you're using this with multiple users or 20 users or some number is that?
Mitchell: Yeah, so we do have a separate framework, which is, it's three categories, basically, we split it between individual, team, and organizational challenges.
And so what we always say is we do not make technical challenges enterprise, the technical challenge should always be solved for free and in open source.
So a very concrete example I give you is, you should not have to pay us to be secure with Vault.
Vaults as a security tool, you will get security, like world class fortune five levels security with the open source product, you could do it, but we sell organizational solutions.
So you could do that, but you might not be able to do it in a way that satisfies GDPR requirements, satisfies data sovereignty laws, things like that.
Those things we will sell to you because those are not, they're solved by technical solutions, but that's has nothing to do with making you more secure, you're just adhering to laws and things and so that's sort of the line we make between the two ends.
The middle line, that collaboration is, you should be successful for free with you and a group of buddies basically can use our products forever, never going to charge you anything.
But if you're working with multiple teams, two teams usually draw the boundary, if you're working with two teams that need to coordinate, you can try to make it work, you probably going to have to duct tape a lot of things, but we will sell you a lower cost thing that solves those challenges associated with teamwork and collaboration.
So it's an organizational challenge, but it's a small organizational challenge.
Grant: Interesting, that sort of allows you to differentiate sort of this good, better, best, free, good, and then best basically.
Grant: In terms of like best in terms or just like most complex, like middle complex.
And then the other advantage is, you don't have to expose the most complex features to your free users, because it would kind of they don't need those things.
It just kind of muddy up the overall experience.
Mitchell: It obviously gets a little more complicated with an organization, challenge, we sell multiple modules which match to use cases and we do that so that there are different price points.
And there's all sorts of stuff there but the big picture is what I just told you and it also makes it really nice because it doesn't get too contentious, there's always the out layer, but 90% of time, what's an open source, what's an enterprise no one argues about it at the community level, like the open source user never gets pissed at us.
The challenge there, and this is something cultural I'v h ad to try to instill is that as times change, so do our beliefs in what constitutes open source for enterprise, the example I bring up is sort of Elastic here.
Elastic made this decision very early in the company to make sort of security, broadly an enterprise feature.
Grant: TLS or something?
Mitchell: TLS AUTH, I think off was a big one, but it's basically like you could run a search cluster, but if you want AUTH and TLS, and that sort of things you got to pay us.
And I think that was totally fine, the time that we did that but I think as an industry have gotten so much more security aware that security has to be a given now.
And so that caused some issues, with Amazon and very public issues and challenges, FOX, things like that and so I've done the same thing with our teams just like times are going to change and I've sort of red flag like, I think this won't last in Enterprise because as a use case, because it's just going to have to be open source.
And so, one of the things I gave an example of was things like multi factor authentication and stuff.
Historically MFA is one of those things and then LDAP and Single sign-on, one of those things that you would withhold in Enterprise.
And I think we're still not there yet, but we're just getting closer and closer as an industry to expecting MFA on everything and expecting to have a single identity for everything, having multiple identities kind of sucks.
And so that's on the fence of I think the whole industry is going to tip over in the next couple years.
Grant: Yeah, there's like some GitHub repo that like shames Saas products for like charging more for SSO or single sign-on.
Mitchell: Because it makes you safer and people are believing that now.
Grant: Right and but five, six years ago, it was obvious that if you were paying for Okta and single sign-on, you could pay more for this product because like that single sign-on solution was expensive.
Grant: And the reality is, I think that the world of password managers actually makes password based accounts a bit more secure.
Grant: So like--
Mitchell: The blast radius is way smaller.
Grant: Yeah, there's an argument you can make and if you have support for 2FA, I still think like single sign-on because of the account provisioning--
That's really the organizational problem you're solving how do you automate the process around account creation and restriction in a large organization with 1000 users it's like--
You don't want to have to go into an admin console and remove and add people every time.
Mitchell: Okay so this is actually a really good example of the feature versus use case thing.
You could still sell the use case of sort of governance and put single sign-on in that bucket, but we might be able to say, certain single sign-on LDAP is now open source but Okta and things like that they might work with, they work fine with LDAP, and native integration things are more enterprisey because you're paying for that it's commercial entity, they have these more advanced features to get that you have to pay us and that lets us slide these things around but the use case is still pretty solid.
Grant: And I think like, ultimately, it's always important to be able to differentiate different pricing plans for the same sort of core open source value, you want to be able to have different pricing plans for enterprises and mid market in open source.
Because ultimately, the amount of value that an enterprise is gaining from using your product, it could be orders and orders and orders of magnitude more than like the amount of value that a small startup is really gaining from it, right?
Grant: So they should pay more?
Mitchell: It takes time to build all that out, we've been monetizing for about three and a half, four years now, and up until last year, we only had top-end enterprise pricing.
It was if you had 100 or say if you had $75,000, and you wanted to pay us, we didn't have anything for you, there was nothing you could really want to buy something for that much and there was just nothing we could slice and dice and give you.
And in the past year, that's been a focus to go more down market but it's really hard to do that, because it's not just about having the product and slicing it up, I think that's the easy part actually is slicing some features and packaging it up and giving it to you.
The hard part is what is the right boundaries, what is the right price point, and also all the support sales you need to hire to support that.
So that's been where we've held back is because it's, we could sell you just replication in Vault for, let's call it $20,000, but if you have a support ticket, we're going to be so overwhelmed by our people paying $500,000 that I don't think you're going to have a good experience, so we're not going to do that.
It's not a product issue, it's an organizational issue for us.
And so we're getting there but I think my learnings there is you can't dive in too quickly into selling to everybody that kind of goes back to not just selling to any segment, but also any price point, I think anchoring at one is a really good idea.
And I would say the rough anchors are SMB well, small, mid and enterprise.
Grant: And you anchor Enterprise, right?
Mitchell: We're going to enterprise and we're moving down.
Grant: It's interesting, because you're open source. So you're already kind of downstream.
Grant: Right? But like you use that open source really, as a marketing channel to get into the top and sell these new enterprise deals.
And now you're saying, okay, well, we're going to also pair sort of an up market or like an up, we're going to move up from open source independently, and maybe let some developers pay for things online and do things.
Mitchell: Yeah, and you can see that, Conway's Law is a thing and you can see that and just our organizational structure.
The number of people whose primary focus is the success of open source is probably 100 people in the company 'cause it's probably like engineering, predominantly, and then DevRel.
And then there's another 800, 900 people that are 100% focused on making Enterprise successful.
It's like customer success, all you do is Enterprise customers, for customer support, 90% of what you do is Enterprise customer sales, named accounts, all Enterprise customers, finance, handling, hundreds of millions of dollars that are from Enterprise customers, the rest of the company is all focused on that.
That doesn't mean we don't care about open source, it just shows we haven't invested in the frameworks and systems that we could support them to be successful beyond their initial adoption curve.
Grant: Yeah and I think it wasn't that long ago, maybe you remember, always talking about "we'll monetize the top 2000 companies" kind of thing.
But now maybe like, there's some new offering even like, Terraform file that say can we target that sort of smaller user, right?
Mitchell: Yeah, and it's mostly our confidence, again, like all things with multiple factors, but our confidence that we've built a system and machines that needs to grow but it's all just they're actually in playbook now.
We built a system machine that could satisfy the top 2000. And we could get there we have, I think we recently passed, I think like 250 out of 2000 of the companies are paying customers now.
And we have like clear line of sight to how to get to 2000, so let's focus our system building on this mid market, we do not have the systems in place, we do not have a playbook.
So let's get in there, so that two, three, four years from now, we are just executing.
Grant: You both.
Grant: That makes sense. Yeah and then we've been working together for the last just over two years, like so for context Replicated, basically powers distribution of Terraform Enterprise. And so that's like a true Enterprise--
Mitchell: Only two years?
Grant: Just over two years.
Mitchell: It feels like an eternity
Grant: Feels really long. Feels like we've been your customers forever.
Mitchell: I mean, you've been a great customer that entire time and your team one thing that I think for any startup, when you find kind of early customers we are still pretty early, two and a half years old, three years old, and we had some great customers along the way, like Travis CI and other folks to help kind of maybe pave the way for you guys to become customers.
Grant: Yep, for sure.
Mitchell: But ultimately, when you find these customers that really understand the problem as much as you do, and they decide, I think, your team had built a couple different versions of like do Enterprise
Mitchell: Replicated was what we call V3.
Grant: Yeah, exactly, so V3-- but you understood the problem.
But you worked with us to make sure that the thing that we were delivering together was the right solution, and it made our product, much better product, and we learned a lot from that, and then for us, we just made sure that we prioritized and we're super responsive and like made sure that your team knew that their requests matter.
But I think it's a really great way to think about building a great enterprise product, is fine that early customer that gets the problem and then work with them as like a almost a design partner as you're kind of building success together.
Mitchell: Yep, if you become a customer HashiCorp now, assuming you're paying us an average amount, you're part of a system and you'll get good service.
But it's going to be standard service, that we give every customer, whereas our first, I mean, 10, 20 customers even like me, Armon, and Dave would go to their office every quarter listen to their problems, Armon would just literally go and debug problems with their engineers just to learn more.
And so you get that early access, and they definitely guided our products in the same way as it sounds like we guided yours a bit.
Grant: Yeah, and it's just a really powerful, it's so important to have that customer that really also gets the problem.
Grant: If they don't get it, then you end up building stuff that like nobody else really wants.
Mitchell: So I don't know who said this. I did not come up with this but someone described that segment as Hipster Enterprise.
And I think it's someone we know, but is Hipster Enterprises, and it's super accurate, which is that your first customers are going to be Hipster Enterprises, but they're actually really bad long term customers, but they're fantastic early customers.
Because Hipster Enterprises, basically to define it, is they are enterprise companies that are highly technical, that get the latest trends and they're excited to adopt it and so they're really good at driving your product to be better, really good at evangelizing, 'cause you could say, so far us a Hipster Enterprises was Twitch.
For the first two years, the templet customer would use was like Twitch uses our stuff on all their servers.
And it's great they're really smart, technically, but the problem is because they're so smart, technically, they never pay you that much and they don't need a lot of the facilities that a standard enterprise will need.
So they do a great job of driving your product and your initial business, they don't do a great job of driving the organization, you need to build, we didn't realize we needed enterprise blueprints that level solution engineer because they'll get it but when you're selling to another company we might talk to is Revlon a makeup company.
It's like, they're smart, but they're much lower staffed, in a lot of these companies, not them particularly, but a lot of these companies.
You know, IT is viewed as a cost center, and so it's like they're really strapped for cash, the caliber of person they get is generally lower than like a Facebook.
And so the type of things you need is very different. So I talk to a lot of new enterprise tech companies, and I actually categorize them as looking at their customers and like you're in the Hipster tech phase, so this is this is different.
Grant: Realistically, it's like, you don't want... And I love the folks at KPMG and BFA, but like, you don't want to go to them in the beginning, I think they can't be can't be your earliest customer. Because take all--
Mitchell: They are all my earliest customers.
Grant: Yeah, but the process is so long and so if you can find a few companies like Twitch for these others, sort of more.
They're just a little bit faster moving in decision making, to be that early customer and get you into production, that's going to be really, really useful to like help you cycle through things faster.
Mitchell: Yeah, banks are another, finance in general is another one that when I'm working with early enterprise companies, it's like you should get them as customers.
But don't get too excited, because they buy everything and it's all perspective, it's like, one of the banks was our earliest customer and they spent I think $150,000 on us.
And it was our biggest deal of time, we saw it as a big deal but then you realize that their annual IT budget is something outrageous.
Grant: Like billions.
Mitchell: Yeah 10s of billions, I think it was like $20 billion, and they're spending 100,000 on you and that puts it in perspective like, you're not important.
Like this is somebody like buying In-and-Out for lunch, like, that's what was just there, and they just have it to have it to try it out and so the next four year cycle, I think, with that bank, it took us four years, we just like in the past six months, got to the point where it's like a seven figure deal where you could actually say, still committed billions is small, but you could say like, every developer uses this now in the company.
But it's hard, money is green, so take that money, help the company, but understand that you need to build the machines still.
Grant: But like that kind of early customer development is so important and you get that feedback, you get to build things.
Mitchell: So we're sitting here in LA, I think that is a challenge,
Mitchell: I live in LA, I've always worked remote from LA, but we base our headquarters in San Francisco.
And that was a big reason and I don't know if that's still the case but six years ago is the case where these Hipster Enterprises that are your earliest customers, and fastest feedback loops are just in San Francisco. And so it was so beneficial to us that, when Twitch called we were next door to Twitch.
And so you could walk over there and I think, at our scale now, our headquarters doesn't need to be in San Francisco. It is established there so it'll stay there.
But it doesn't necessarily need to be we could do anywhere. But early on it was so important, so I don't know if that's still true, but it kind of sucks.
Grant: Yeah, I think it is largely still true. Like even though we're based in LA I go up all the time in the Bay Area just has such a high concentration.
And it's not just like the fact that it's like partially because they're next door, there's just this level of serendipity, where like you're walking down the street, you someone who's like, you should use this new thing that we launched and just like all these little side conversations but just don't happen.
Like when you're in a meeting, there's a big difference between seeing someone at blue bottle and talking for five minutes three times a month, then the hour long meeting over a zoom or even in person, it's just like a little bit, you're able to have less serendipity you're less top of mind.
And so being there is very helpful. There's why you see a lot of companies that will open a sales and marketing office in the Bay Area and have development remote or development and some different city.
Mitchell: Yeah, I think it's fairly off topic, I guess, Enterprise Tech, but I think that is going to be the struggle with these alternate silicon, silicon cities because this is like the teeth thing, it's like it's a societal unfortunate thing but the people you're around and have immediate access to the easiest, really make a difference.
And one example we had was when I lived in San Francisco, I didn't really talk to my roommates and then I found out one of my roommates is an associate to VC firm, and he got me a meeting and like that we didn't raise from them, but that like helped a lot and it's like, I've lived in LA now for like, six, seven years.
But, in LA there wasn't any tech people my entire street, I want to say like zero and so it's like, having--
When I go to a coffee shop, I'm the only person with an editor open everyone else's screen writing stuff open.
So like, I think that sort of stuff makes it harder and it's fairly cultural.
But that definitely exists for again, I think it was an early stage problem.
Grant: Yeah and there's areas where I think it matters most to enterprise software company.
Mitchell: Yeah, right, I call it "tech for tech." I call it tech -- if you're a consumer, LA is actually really great.
Grant: 'Cause you have lots of things but eventually--
Grant: I think what will likely happen over time is like as consumer companies kind of get started anywhere because that's the world is technology that eventually like Twitch could be in LA like you're in so your customers can kind of be anywhere here.
Mitchell: Snapchat is here.
Grant: Snapchat is here, SpaceX is here as always. And all these different companies so you'll start to see more high tech quote unquote companies starting all over the place, and eventually it'll distribute.
But Enterprise will be last 'cause we're selling to that Hipster Enterprises. Today that's probably in the Bay Area, I mean, you just together piece I always say, like, in kind of in defense of LA, is just, I just love being around different people.
Mitchell: That's why I moved here, that was actually I was so tired.
It felt like I was never not working, because I would go to a coffee shop and I'd be stressed out, because everyone just about tech or I'm in like, if it's something you recognize, you can't--
You're not trying to listen, but you just do. And I would just get stressed out and that's a big part of why I moved.
Grant: Yeah, like I sort of being, at a dinner up there, just sitting across from maybe Mark my co-founder and like the two tables next to us here, it's all two tops, and they're both talking about infrastructures like Docker and Kubernetes are like five four years ago and oh my god like this is--
Mitchell: You got to be able to shut it off.
Grant: Yeah, in LA is great 'cause you get to be around creative people and do different things and you that's--
Mitchell: Industry wise it's super diverse, right?
It's like okay like Hollywood's obviously huge but so is Aerospace and so is media and so is to a smaller extent finance.
And so you go to like a bar and you meet people, and you can be for people that are for different industries, and it's awesome.
Grant: And like you and I probably spend less time in West Hollywood so
Mitchell: My wife's an actress.
Grant: Oh, yeah.
Mitchell: Lot of actors out there.
Grant: I'm probably more removed from being, just in the west side, so I'm rarely I've friends like producers (murmurs) I'm not I'm not in West Hollywood.
Mitchell: I use my street as an example, there's no tech people, but all my street, my three immediately like forward, left, right neighbors are all entertainment. So, you get that it's a high concentration.
Grant: Cool and so I mean, now, like HashiCorp, the company's been around for a long time. You're obviously doing more and more like you're getting 1000 people, right.
Mitchell: Yeah, we're any week now, it will be 1000 people.
Grant: That's crazy, and you have these core products, you mentioned some new products as you kind of thinking like next products? Is the intention to kind of continue to expand the--
Mitchell: Yeah, so we're doing a few different things, I think we're actually--
So it's funny is we're still four companies in one company and our four companies are at different maturity stages. So Vault is probably our most mature.
It's been our revenue leader for a long time, although that's rapidly changing now, it's very diverse now, but it's just the oldest business that we have.
And so we've sort of reached a point where we're starting to look at adjacencies that are not core Vault like new products, but still security, things like that.
So we're doing stuff like that. And then we haven't built any new open source, new category projects and four years, four or five years, and that was because we needed to build this business.
I think, as we've reached this maturity and this size of company, I've shifted my personal attention back towards there is some new stuff we're working on finally.
So it's been a while, but I'm excited about it so we are we are doing stuff like that.
Grant: Awesome and then I'm guessing when you try to monetize right away, or when you try to do more adoption first is that the--
Mitchell: We're still trying to figure it out but--
I'm very clear with all our executives that everything seems like an overnight success once it's successful.
But if you look back at Vault, Terraform, is that our Vault was our fastest success.
And even if you look back and Vault, it wasn't super successful for about six months to a year. And that was like extreme speed, I mean, that's crazy.
Vagrant took two years, Terraform took two years, Consul took three years or so.
So it's like, I think it's very important and I've been very level with our executives that we need to figure out a way to just focus on the community and work on the free open source project.
And it might fail, but just to give it that six month time where we're not splitting concerns or polluting anything with monetization, we just need to focus on the core solution.
So we'll see it 'cause it's the first product we'll come out with as a sort of successful, quote unquote, enterprise business, so we'll see how it goes.
Grant: Yeah and I guess the other part there is, it'll give you a chance to really make sure you're built the right thing.
Grant: And so like, that's it, if you start charging people right away, you kind of are locked into some amount of maintenance, or at least migrated them off of it.
So if it's an open source thing, you can kind of iterate it and keep it at version zero dot one for a while.
Mitchell: That's the plan
Grant: Until you find Okay, here's the real thing that we know people want and it works. And then and then add those you kind of have the game plan for taking it to enterprise.
Mitchell: Right, exactly.
Grant: It's cool, hat's kind of next in your mind for HashiCorp, where do we see the impact? What's the vision of the future?
Mitchell: I mean, I think our mission still stands, which is we want to be that important piece of infrastructure that every company has.
And I think that's a long term, decade long, sort of goal. So we're still working towards that, I have always said the audacious goal we had when we started the company, and it's less and less audacious as time goes on, but it's so crazy.
We want Terraform, Vault, whatever projects to not be choices so much, it's like, there are choices, obviously, but it's the obvious choice it's sort of like get today is like, you could use still subversion, you could use bizarre there's a bunch of stuff that's like, cold popular and stuff, but get is just fantastic.
When you start a project, no one says, okay, what version control are we in, no, when you start a new project, I don't want anyone to say how we're going to build our infrastructure.
It's Terraform skip that whole discussion, You know, that's where we sort of want to get to for every company, and that's working towards.
Mitchell: Yeah and it's ubiquity with knowing that, and we're very clear with our executives, knowing that maybe only if you reach that level, ubiquity, maybe only 5% of the people actually pay us.
Mitchell: I think that's the power of open sources that ubiquity and a pure commercial product means like monopoly.
And I think for us, ubiquity means you have choice, but it's just the best product and they're choosing to use it for free, that's what we're hoping.
Grant: Yeah, that makes sense, I'm interested too, because there's one piece about ubiquity, which is, if you compare HashiCorp approach to open source, versus like sort of the foundation approach open source are very different.
So you sort of are the primary company behind these projects, whereas if you look at other projects they have sort of a foundation yet in that community, and there's all this governance and all these things. How do you think about the advantages and disadvantages of that, are you interested in--?
Mitchell: That'd be like a whole podcast.
I mean, I think the basics of it is that, I've described it as iPhone versus Android, or iOS versus Android, is that I'm an iOS person, I think Android's fine, but I think that iOS, it's a hard thing to say 'cause you can't like quantify it, but I think iOS is more beautiful, I don't mean that aesthetically.
I just think that when you have a single entity that helps drive the big picture vision, you just get something that's more designed.
It's more designed and I don't think foundation's have ever done a very good job of that.
I think foundations are great at a certain stage of product, like for example, I think Java, having a foundation or Patchy having a foundation like for the actual Apache web server, like is excellent, because they're sort of done.
I mean, they're evolving a lot, but the vision is done. It's very clear, everyone can get behind the vision.
I think projects that enter the foundation super early or have governance, open governance a little bit too early, it's messy because people bring their own vision.
And that's shouldn't be the point, the point should be that everyone gets a fair say, in how to achieve the vision.
But that's not clear yet, and that's the struggle that I sort of have with it. I think that the fears the foundations address are lack of control, change your control in general of like companies, things like that.
And I think that we address that via very liberal open source licensing, as well as being very open to contributors, open to feedback.
We've never really taken strong combative stances on things, we were open to these discussions. I think being a good player as you build that reputation, that's what we focused on.
Grant: Cool Mitchell, thank you so much for all of your time today.
Mitchell: Thank you, that's great.