Peter Wang - Anaconda

Anaconda: From Bootstrapped Startup to 40M Users – with Peter Wang [418]

Anaconda: From Bootstrapped Startup to 40M Users

Peter Wang is the co-founder and Chief AI & Innovation Officer of Anaconda, a platform that offers essential open-source Python packages for AI, data science, and machine learning.

In 2011, Peter and his co-founder, Travis, saw an opportunity to make Python mainstream in data science and analytics, but they faced a tough road ahead.

At the time, Python wasn't widely accepted in enterprise, with most companies heavily invested in Java-based tools like Hadoop. Convincing them to switch to Python for big data analysis was a big challenge.

The founders bootstrapped Anaconda by offering consulting and training services while investing heavily in building an open-source community.

They also had to take on well-established competitors in industries that had been relying on the same outdated tools for decades, and prove that their modern, open-source solution could deliver better results.

In 2015, they launched their first enterprise product, offering corporate clients a more secure, controlled version of their open-source tools. This was great for monetization and revenue growth.

However, it also led to internal challenges, as employees struggled to balance the needs of open-source users and enterprise clients, causing confusion around resource prioritization and aligning the company's vision with its business model.

Today, Anaconda serves over 40 million users, generates 8-figure ARR from its enterprise solutions, and employs over 350 people. The company has also raised $45 million in funding.

In this episode, you'll learn:

  • What strategies the founders used to overcome skepticism and get businesses to adopt Python in a market dominated by Java and Hadoop.
  • How Anaconda competed with long-established incumbents in key industries.
  • How the founders balanced the needs of their vast open-source community with the demands of enterprise clients.
  • How Peter and his team turned customer feedback into a profitable enterprise product while scaling with limited resources.
  • What lessons Peter learned about staying adaptable and focused as Anaconda grew from a bootstrapped startup to an 8-figure ARR SaaS company.

I hope you enjoy it!

Transcript

Click to view transcript

This is a machine-generated transcript.

[00:00:00] Omer: Peter, welcome to the show.

[00:00:01] Peter: Thank you. Thank you. So glad to be here. Thank you for having me.

[00:00:04] Omer: My pleasure. Do you have a favorite quote, something that inspires or motivates you that you can share with us?

[00:00:10] Peter: Yeah. Well, there's I think one of my favorite quotes when it comes to entrepreneurship in business is a quote from Antoine de Saint Perry, the author of The Little Prince, and he had this fabulous code.

[00:00:21] He said, if you want to build a ship, don't drum up the men to gather wood, divide the work and give orders, but instead teach them to yearn for the vast and endless sea. And it's just one of those like beautiful quotes that, especially when you're early stage, I. You seem to be doing a possible task.

[00:00:40] It's a vast to endless sea, and the thing that the founder can do is to create that inspiration. Now, if all you ever do is teach them to year, you'll not get a ship built. At some point you have to go gather wood and divvy up the orders. But, but I think it's important. It's important when we're building new things to put that dent in the universe, kind of that mentality.

[00:01:01] Founders need give themselves some credit of. Of having the, the chutzpah, having the inspiration to do that. And that is a, a unique founder thing to own. Always.

[00:01:09] Omer: Yeah. I love that. That's a great quote. So tell us about Anaconda. What is the pro, what should I say? What do the products do? You know, who do you, you know, who do you help and, and what are, what are the main problems that you're solving?

[00:01:24] Peter: Yeah. So Anaconda is, it is a very popular distribution of software that people use to do data science, machine learning, python, numerical computing, engineering, and all that kind of stuff. So it is a collection of open source software. Primarily this is what people know us for. And you go to our website, you download the software, and now you're, you're, you have all of the different pieces that you need.

[00:01:49] And if you don't have all of them, you can use our software package installer update tool to go get other bits and pieces. 'cause the open source ecosystem around Python is very, very large. And some of those pieces you want to get are very complex. Even though it's all open source software, it's extremely hard to build those things correctly, make them work right, for whatever hardware you have, et cetera.

[00:02:07] So we've taken care of that problem for people and we, we've been doing that for a very long time, 12 years now. We've done that and really we were. At the center of the movement to bring Python to numerical computing, python data science into the mainstream. So, the reason why Python is the most popular tool for data and AI today is because of what was launched at Anaconda.

[00:02:27] So we both build that open source software installation and and, and update mechanism. But we also do a, we invest a lot into the open source ecosystem, incubating new things, maintaining old things that people don't wanna maintain anymore. But that are critical to the ecosystem, all of that. But the really, I think the important piece for the people listening to this podcast is the monetization of that.

[00:02:49] And so that is then the enterprise products and services that we sell to our thousands and thousands of enterprise customers, enterprise accounts, right? There's millions of users of those places, but thousands of accounts. So, so we do all of those things, but I think we're best known and probably most known for our software download.

[00:03:06] That people use to run Python data science, ai, ML stuff.

[00:03:10] Omer: Great. So open source product used by millions of people around the world and then a few thousand enterprise accounts, which are, how, you know, basically is how you, you monetize?

[00:03:26] Peter: A few thousand paying enterprise accounts. There's hundreds of thousands of enterprises that use us, but only, only a few thousand that are paying us right now.

[00:03:35] Omer: Yeah. And give us a sense of the size of the business. Where are you in terms of revenue, size of team?

[00:03:42] Peter: I, we're, we're, we're a very, very healthy eight figures.

[00:03:46] Omer: Eight figures in ARR. And how big is the team today?

[00:03:50] Peter: North of three 50. We've been

[00:03:51] Omer: hiring pretty rapidly. And you've raised just around, we were talking about this earlier, it's about 80 million.

[00:03:58] That's right. Great. Okay. So the business was founded in 2012. It was you and your co-founder Travis. He, he, so I did, I, this was a little factor that, that I sort of discovered that he was the, was he the creator of mpi?

[00:04:17] Peter: Yes, Travis is actually the creator of MPA in SciPi. And for those on the call who are not familiar with those libraries,

[00:04:24] Omer: who know what is

[00:04:25] Peter: mpi?

[00:04:25] Yeah, yeah. I mean it, I don't assume anything. It's a big world out there of software. So NumPy is the numerical python library that basically creates, you know, allows you to do vector and array. Matrix Math with, with Python, and it's a fundamental, very fundamental piece of, of numerical. I mean, it is sort of the most fundamental piece of numerical computing, sort of like what the Linux ker is to the Linux ecosystem, right?

[00:04:48] And then SciPi is scientific python, and that is the various additional functions and like nerdy math calculations that you want to do. You know, and so it's used very heavily in engineering. It's used very heavily in, in finance and places like that. And then all of the other tools in the Python ecosystem really build upon these things.

[00:05:05] So there are two really I would say, like one is maybe, one is ker, the other is GCC or something. They're so fundamental to what makes Python work for data. But yes, Travis is the founder, so he is the creator of those two tools. And, I guess he's a co-creator of SciPi, but he is the sole original creator of NumPy.

[00:05:23] These are, of course, now maintained by teams much bigger and that than, you know, came later. But yeah, that's, we, we created the company because we started doing a lot of consulting around this stuff. So before we started the company, we were consultants on top of numerical computing in Python and Python for engineering simulation, financial simulation analysis, all of that stuff.

[00:05:42] Omer: So let, let's go back to 2011. You, you guys, I mean, we say the company was founded in 2012, but you bootstrapped for the first three years mostly your revenue was coming from consulting and training services. So just walk us through that, like how you guys got started here and then. At what point did this become a product business?

[00:06:15] Peter: So the company was founded on a thesis that the Python, numerical python and, and, and the python data stack as we called it, that that ecosystem was ready to really be taken mainstream into bus, into transforming and disrupting business data analysis. And disrupting in a good way, right?

[00:06:36] Transformatively innovatively disrupting that, that software ecosystem. That was our bet. And we knew that there were many deficiencies, many things that needed to made be, made better. Usability bits, scalability, you know, data connectivity, all these kinds of things. But based on our consultancy, we could see that this was starting to organically get adopted into.

[00:06:54] Big banks, investment banks, hedge funds, people with a lot of money to throw in it and software, but they were opting to use this, still emerging, somewhat open source Python stuff. So that was the, the bet. And so when we started the company, we knew that a lot of our work was gonna be community building.

[00:07:11] A lot of our, of our work was gonna be advocacy and marketing. So very unique among startups. We founded the startup and a conduct, actually, we used to be called continuing analytics. So we started continuing analytics as a company. To, to build products around this open source ecosystem and sustainably fund additional continued innovation there.

[00:07:30] We also started a nonprofit, so not many founding teams start a 5 0 1 C3 and the Delaware C Corp at the same time. As, as multiple components of the same underlying business thesis. But we did that. And so as part of the community efforts, we then created a foundation to house a lot of these projects, raise grant money for them.

[00:07:48] We created a meet set of meetups and conferences. So I, I personally created the concept of PI data and the PI data, you know, you know, meetups and, and conferences that are now global. So all of those things were efforts to. Drive adoption of Python. So those first three years we had a couple million dollars of seed funding from friends and family that we raised.

[00:08:07] We won some DARPA contracts, which was amazing. DARPA was a great partner to work with. All that money that came in either went directly into seeding and building new kinds of open source projects or creating more adoption and creating more marketing materials to advocate for the use of Python, all these kinds of things.

[00:08:24] Some that went into. Building things that would eventually become our core product, right? And that is the Anaconda package installation system, the Conda package manager some of our various websites and hosting sites that host, and the packages that people build, that we build. All those things were then, you know, those were built on the, on the margins coming from the consulting service that we did in those days.

[00:08:45] Omer: Can you explain the rationale? I mean, in, in the early stages when you're, you're trying to get your business off the ground. There are so many things to do. You just, you know, I found it just doesn't have time to do everything. And then for you to also say, okay, we're gonna have this nonprofit side, you know, we, we are gonna kind of focus on that piece as well.

[00:09:11] Can you just explain the rationale, like why you guys. Felt that was so important for you to take on at that point?

[00:09:18] Peter: You know, that's actually a question no one's ever asked me. That's a really, really good question. I've told the astounding story many times, many podcasts I've told about like street event, I've told on other podcasts and whatnot.

[00:09:27] But no one's ever asked me that question, like, why did you waste your time with this distracting activity around this other stuff? But I think in a beautiful way, it connects back to that original quote, right? Year for the vast and endless sea, there was a mission to what we were doing. We really believed in the quality and the amazing power, the, the, the amazing power of the open source community to create high quality innovative things.

[00:09:52] And so for us, you know, building that community and supporting and pushing, that was not an optional thing. That was what we're trying to do. There's a related quote. That I, that almost selected, but I'll give it to you now. And it's from Peter Norvig and he said the best businesses are are built by those who wanted to achieve a goal and had to start a business to do so.

[00:10:13] And in our case, it was exactly that. We wanted to achieve a goal, which is to make Python, the language and the technology for data analysis and numerical computing in the world. And to do that, we had to create community support things and to really make that sustainable. We had done enough consulting to know that just trying to piggyback this on the basis on the back of consulting margins was not gonna be enough.

[00:10:35] If we really want to create a sustainable ecosystem, like we need to actually show we could build products around this. So we started a company around that as well. So for us, the company is, is, is, is sort of integral to the community motion that we're doing, but, but from a very tactical and pragmatic business standpoint and prove to be a great business strategy because the bigger, the bigger you make the pie and the more people you get to grow that pie.

[00:10:58] The bigger your share naturally becomes is an expanding universe. Right? So it was really important that we did that.

[00:11:04] Omer: Yeah. I I think it's, it's turned out to be a very smart strategy. I was just curious whether it was intentional at the time or an accidental outcome of, of you, you know, doing, doing it for a different reason.

[00:11:18] Peter: Oh, and I, I, I didn't, I didn't take it in a defensively at all. I mean, I think it's a great question, right? But, but it, it's really, yeah, it was, it, it, it was something that we've, we were, we were really motivated, you know, we did have a vision behind, and I think this has helped in terms of general credibility of what we do.

[00:11:33] We've certainly made a lot of mistakes along the way and whatnot, but all along so many people in that community have, you know, they've gotten a start. They, many others have started have gotten funded startups, built companies, built their careers out of this ecosystem and, and they would not have had that opportunity.

[00:11:47] We don't have such a vibrant ecosystem. If we hadn't had, if we were always just doing it from a, you know, just a, a, a bean counting perspective, like how much are we getting out of it and all that stuff. So in the end, it really did pay it forward for us. Right. As a business,

[00:12:01] Omer: one of the things that's interesting to me is that personally, you know, if I think about data science or data scientists today, it's hard to imagine.

[00:12:15] About those terms without the word python coming up at some point. And so it was really surprising to me when, you know, earlier we were chatting, you said in those early days. We were struggling to convince people to use Python. Right. Just tell us like what was going on, like what were people using at the time and what kind of objections were you getting?

[00:12:38] Peter: Yeah, so again, 2011 timeframe, 2010, that was when big data was the thing. People, people really were starting to turn the corner on cloud. So cloud is gonna be much more populated, adopted and no sql, big data, Hadoop. Right. These are the big things people are talking about. Data science was even still sort of a term that people were looking at with suspicion.

[00:12:58] 20 12, 20 13, 14. You know, lots of jokes about data scientists, just a statistician who lives in San Francisco, you know, these kinds of things. Those are new terms at that time. And so everyone, all the focus was on Hadoop and big data. So, you know, when you're, when you're then talking about this Python stuff on the side, it's like, well that's not Java.

[00:13:17] And you know, big data, big Iron, big systems, Oop resulted written in Java. Spark was Scala, which at least ran on the, the JVM and there many people thought, well, Scala and Spark, that's gonna be the, the, the, the future of all this stuff. We're gonna rebuild everything, everything that the r people and the Python people are doing.

[00:13:33] You could do all of that inside Spark and inside the JVM, we, we certainly didn't feel that that was the case, but you know, that was where the argument well. Ultimately resolved itself in the market. Right. But, but there were certainly perspectives there. And from a statistics standpoint you know, there are these there are a lot of big statistical tools and big analytical software that were around.

[00:13:52] And people did see that open source tools like r would be a competitor to that. So I think r was seen as the rightful air to the throne in that space. And then in comes like a bunch of these like apply physics, engineering, whatever kinds of like nerds from the Python land. And, you know, Python at that time was defending itself on all, all fronts, right?

[00:14:13] There were, you know there was competitive competition with Ruby for web frameworks. And and then on, on the data side, numerical side, we were just a weird little sideshow. So the first first or second strata conference I went to, I think Strata was the big data conference that Riley put on.

[00:14:27] I was getting. I would say, not roundly mocked, but people were definitely like, wait, why are, why are you talking about Python? I mean, this is a big data conference, right? And, and so definitely there was a lot of that. But you know, we, we had a bet and I think our bet played out pretty well. And, and my, my fundamental thesis was that what we've seen in our consulting was that end users, people who are not traditional programmers.

[00:14:52] They have analytical needs. They could be, they could be a mechanical engineer or electrical engineer. They could be a quant finance person. They could be someone doing marketing analytics. They have a lot of data. They want to go and ask really interesting questions of beyond what they can do in Excel.

[00:15:09] Certainly beyond what you can ask in sql. And they're not programmers, so they don't wanna learn 15 different languages. The ho, you know, hodgepodge of stuff and Python was something that would fit in their heads, let them do. The data stuff they wanted to do, maybe do a few other things. And it was all, it was good enough.

[00:15:24] And so our bet was that that was going to drive a lot of adoption and that we could then plumb other kinds of connectivity into that, that ecosystem, that Python ecosystem, if you could just be the first mile. So that that, or, or I guess in the telecom sense, they call it the last mile, right when it actually goes to the house.

[00:15:41] So when you actually have a person with a brain with interesting questions of data, fingers on keyboard. What are they thinking? What are they typing? And our bet was from what's seen in our consultancy for what's, what we had seen in our open source ecosystem. Who, who, who were the people that were showing up in it?

[00:15:55] There are people that didn't have a traditional programming background. They're not software devs. They don't have 12 years of c plus plus experience. They have all these other domain expertise and we saw those people reaching for Python as a preferred tool. So we bet on the end user, we bet on this tool that will fit in their heads.

[00:16:11] And then we try to bring other things to make that ecosystem more powerful. But we plumb it through to, to the end user. And now this has really been validated. I dunno if you guys saw the, the announcement just recently, the Excel integration of Python. So now anyone who's using Excel can reach for these Python tools.

[00:16:27] And it's right there in the formula bar. So the, so our thesis has proven out brilliantly beyond what we possibly could have imagined when we started 12 years ago.

[00:16:34] Omer: So you, you were facing a lot of skepticism about using Python rather than. Sort of pushing a rock uphill and trying to convince all these people.

[00:16:46] What I'm hearing is you were like, well, we, we did that to some degree. But you know, what was more impactful was focusing on the end user. People who, you know, weren't sort of traditional software developers, but needed to be able to use a tool like this, and. Make sure that you were, you were speaking to them, serving their needs and so on, how you talked about community building, but how were you reaching these people?

[00:17:17] How, or was it all inbound? Like how did this, this sort of community effort, I. Really, really take off. I mean, today, like how, how many, how many developers would you say, or, or let's not say developers, how many end users you know, are using Anaconda today?

[00:17:35] Peter: It's in the tens of millions. You know, it's, it's, it's weird because we don't have a traditional multi-tenant, SaaS.

[00:17:41] You know, you don't go to website to run Python. You download Python, you run it locally. We don't put creepy trackers and everything in it. Like the software, you download it and if you choose not to give us your email address, then you know you just have your stuff. Now, when you go and you update packages, of course that reaches our.

[00:17:56] Our our, our, you know, package hosting repositories, and we can sort of, we, we run a lot of analytics and we do a lot of this kind of stuff to sort of see even when people don't log in or don't create an account with us you know, what, what is the actual number of uniques and, and all that kind of stuff.

[00:18:11] And based on the best analytics that we can tell it's, it's north of 40 million. Users, which is shocking. But then if you think about the fact that we have you know, millions and millions of unique downloads of our bulk installer every single week, I mean, it adds up a lot. And you think about how many people in schools, universities, their learning python.

[00:18:31] The first thing they do is they, they download an a condom. They start playing with this stuff. So, and it's international, you know, it's China, it's India, it's South Korea, it's Australia, it's everyone in India. Everyone is doing this stuff 'cause everyone needs data and they wanna do machine learning on it.

[00:18:43] So yeah, it's, it's millions and millions of people. But to your question, the, the, the, like, how do we get it going? It was all organic and that's why the community building was so important, right. That we. We by creating these PI data conferences and having even that term, I mean, you know, GitHub was part of it, Twitter was part of it.

[00:19:00] There was definitely a socialization element in the virtual spaces. But having these real conferences and meetups create a venue for these practitioners who wouldn't go to, like, they don't go to traditional programming conference. 'cause that's where people talk about language things and all these like.

[00:19:16] Very, very sort of programmer dev centric stuff. And, and, and so when you have like a PI data conference where people go there and they learn tips and tricks, how to apply these different libraries to the, the data problems they have, well they feel like they found a tribe, right? And the people start running books and they're, you know, just all these things, YouTube videos, tutorials.

[00:19:34] We do webinars. We invest a lot in that community building and really. Validating that there is a Python data science thing, that that actually is a thing and not just an off-label use. I think that's a lot of sweat equity and sort of, I won't call it emotional labor, but it's a lot of invisible investment.

[00:19:51] I mean, Anaconda Inc. Underwrote a lot of those conferences. I. You gotta sign venues, contracts, food vendors, stuff, all these things. And the nonprofit, no focus did not have much money at the time. So we underwrote a lot of those conferences. And once those became self-funding, we handed all of that option on focus so it became a revenue generator for them and they could take those revenues and turn 'em around to being grants to the open source projects.

[00:20:13] So that was kind of how we got that separate engine going in the nonprofit open source space.

[00:20:19] Omer: It's just mind blowing that number of like 40 million. Plus end users. And in many ways, I think a lot of people would be like just freaked out by this idea of, wait a minute, you've got 40 million plus free users of your product.

[00:20:37] You don't ask for an email.

[00:20:40] Peter: We ask for it, but we don't require it. We do ask for it, but you can click the X so you don't have to give it to us. Right.

[00:20:45] Omer: You don't require it. And. You don't have a multi-tenant solution. People have to either, the way they can, you know, get Anaconda, as you said, is either download locally on their computer or for enterprise customers, which we'll we'll talk about in a while.

[00:21:01] They can have some kind of on-prem solution. So in many ways you're, you're doing things so many things which are so, you know, contrarian, I guess in terms of how you are supposed to run, you know, a, a subscription business like this. Let, let's talk about like how that on-prem product. You know, it sort of started to come, you know, come about.

[00:21:28] So we talked about like the first three years of consulting, training services, and then year 4, 20 15, 20 16 is when you started to provide some clients is on-prem solution. How did that happen?

[00:21:45] Peter: Well, I, I do, I do wanna clarify. We do have an a cloud-based. Multi-tenant, sort of traditional SaaS application for hosted notebooks in the cloud.

[00:21:54] You go to ana cloud.cloud, I'm encourage people to, if they don't wanna install something, they can use it there, but it's not our primary thing, right? We have, we have, you know, a healthy number of users on there. It's growing but it's not like our primary thing. So the on-prem, the on-prem bit and your question specifically was which aspect of it was like, how did the bus, how did that business emerge?

[00:22:13] Omer: Yeah. So, you know, you're, you're kind of going down this path of consulting training. You're building this community. You're starting to get some, some traction. You've sort of figured out who the, who the right ICP is, and at least in terms of the, the community side and the open source side. Mm-Hmm. Did you guys just wake up one day and say, Hey, let's add an on-prem, you know, offering?

[00:22:35] Peter: No, no. Actually this was very lean startups. 'cause this was me and Travis' first. First startup, right? So we're like, well, we read a bunch of books. We're a couple of nerds, so we run, read some books. The we were both in Austin, which did not have as much of, there was like, no, like venture ecosystem. It's not like if we were in the Bay Area, every, every, every night there's a meetup, drink up, something where you're just swimming in other founders and like, you just learn so much from them through osmosis.

[00:23:02] We had very little of that. So we were reading books, we would go to some of these things, but it wasn't, I would say it was not rigorous and it was not in the groundwater right in, in Austin. So, so we really took like four, four steps to the epiphany lean startup, a lot of these kinds of things, very much to heart.

[00:23:16] So we got inbound. So, so our software download you, you download this thing and then like so many pieces of open source software, there's always updates. You wanna get the newest version with some feature or maybe there's a bug fix. So people are getting package updates all the time, and it's a very large ecosystem.

[00:23:30] Hundreds of thousands of different libraries and packages. Okay, great. So we get emails from companies. The first was actually from a law enforcement agency, and they said, Hey, we love what you guys are doing with Python and this is great, but we cannot have our, our, our folks here internally just installing random packages off the internet.

[00:23:49] Do you have something you could sell us? Now we would have our own behind the firewall package repository that we could then manage, right? That we would be able to exile certain packages that we don't want people to use, that we could force certain kinds of updates, if there's security updates, et cetera, et cetera.

[00:24:04] And we said, you know, we don't, but we could put one together. And we put that together. So we put that together, sold it to them, and there it is. That was your first product sale. And Lean startup just says, when customers come to you with money in hand, say We want this product. You should at least listen to that, right?

[00:24:20] So we listened to that, we built that, and then we started talking to other people. You know, again, inbounds, but we also did some webinars, which real, we did invest in a little bit of outbound marketing and positioning of this stuff. And it turns out that a lot of people have this need, especially when it comes to running, you know, when people do Python stuff for data, they're running it on data.

[00:24:38] And so, so there's sensitivity around it. Oftentimes, you know, there's, there's corporate environments where the data cannot leave a particular. Network. In some cases, data cannot lead a particular machine, right? So they really need to have all the stuff running. They really wanna have an internal mirror and a way to manage their open source.

[00:24:54] Consumption internally. So that became, then this, we just added more and more features. People wanted the ability to filter out GPL code. They didn't want their junior data scientists to absolutely put to, to accidentally put some GPL web service into production. Right? And so all these kinds of things.

[00:25:09] And then we started doing things like CVE scoring and, and vulnerability alerts and stuff like that. We had, you know, custom package build services. We had ways for people internally to create additional channels and make it so that only the dev the dev cluster only sees some, some packages, the production cluster sees other packages.

[00:25:25] We just added more and more bells and whistles and, and things like that. And I mean, that's kind of, that product is still a mainstay of, of what we sell today.

[00:25:34] Omer: What, what percentage of, of your revenue roughly comes from. That, that product,

[00:25:42] Peter: so that product is in the I'd say probably about 30% of revenue.

[00:25:46] Maybe more than that. So it's, it's a little bit difficult because there's the on-prem, but then we also can run that in the cloud for people. So people do want their own customized package server, but they want it, they're all cloud, you know, they might be cloud native and have a Kubernetes thing spun up on Amazon or you know, wherever.

[00:26:02] And so we do that. And then we also have a secure commercial package repo that we sell subscriptions to. And that's, that's another good chunk of our revenue.

[00:26:12] Omer: Let's talk a little bit about, you sort of mentioned the, the incumbents that y you know, were, were around at the time, and so you're making a bet that Python and Anaconda is the right solution for these, these end users.

[00:26:33] But I, you, you also told me earlier that, you know, one of the struggles that you had was that you just felt that. You know, we weren't as, as good or as polished as a lot of those other alternatives. Can you, can you give an example about that? Like, you know, I can say not as good as polished, but, you know, make it real for us.

[00:26:54] Give us an example like so people can understand.

[00:26:56] Peter: You know, some of these other analytical tools, they have a lot of specialized analysis functions that have been developed over decades. I mean, some, these, these tools, these incumbents are decades old, right? And so if you are doing a particular kind of financial analysis, I.

[00:27:10] You'll literally have every possible thing in there because they've been accumulating customer use cases for decades, and they put functions in there and they, they would just, you know, that's what they've got in the open source world. You get whatever you get, whatever somebody decided they needed and they took the time to go write it and put it into the thing.

[00:27:26] So when you're first starting in an open source ecosystem, that crowdsourcing contribution dynamic is not so large. Your, your, your baseline of how many people you have is not that big. Now as it gets bigger and bigger, you then you have more and more people, more pull requests and then maintainers merge those in and it becomes more and more capable.

[00:27:43] And that's great, but when you're early on, you don't have as much coverage of basic functionality maybe. Right. But then there's other kinds of things where I. Yeah, where like there may be some visualization, some special visualization. They want some particular format. I, we saw this in Geophys, like, I'll give you an example for like oil and gas, geoscience and geophysics.

[00:28:01] There are a lot of very expensive tools for doing analysis that have just been there forever. And so when we're first doing even consulting with some of the stuff. People like using the Python as an alternative maybe to Matlab. They like using, they, they, they, you know, certain kinds of things. Were nicer to write a python, but then certain kinds of visualizations I.

[00:28:21] Now we're very custom. You imagine looking down a well, like if they have a well and they do some analysis of all, all the measurements down a well, that's not a standard plot type that most people will run into, right? It's kinda a weird kind of chart, but they've got a very custom way of looking at this kind of data in the geosciences and all of the big software suites for.

[00:28:41] That industry have very custom, very nice and polished UIs for this. They've been developing over decades and in the open source world, it's like whenever I hack up together in like space a couple of months, that's what you get, right? So there's that kind of a thing. So there's just like a lot of that kind of stuff.

[00:28:58] That makes any sense.

[00:28:59] Omer: Yeah, that totally makes sense. You know, I, it kind of make, I was thinking about that, that, you know, with many founders. They, they have this dilemma that they have to deal with, that on the one hand, they know. All the flaws in their product. They know all the stuff that's kind of works, but is kind of, you know, just, you know, glued together and could fall apart.

[00:29:29] They know about things that they want to do that could be 10 times better, but they just haven't been able to invest the time and, and the resources to do that, and yet at the same time, they have to go out there. And tell the world and tell customers how great their, their product is. And it sounds like you went through sort of a similar process.

[00:29:55] Was that, was that like a difficult thing to do? At the time?

[00:29:59] Peter: It wasn't a difficult thing because I was such a believer in the community and the amazing people in it. And. I really believed that the idea of empowering these subject matter experts or domain experts, empowering them was fundamentally the right thing to do.

[00:30:17] So this is where, you know, if you're the founder, you're creating a religion. You, you have to be the center of sort of inner subjective belief that there's a meaningful mission that. That there's a thing to do. So for me, it wasn't difficult to manifest that future visionary positive thing, but also for me, like if you talk to other people who work with me and whatnot, that is sort of my natural mode is I'm very forward looking, optimistic.

[00:30:39] So that was also a little bit easier for me. And I recognize that for many people, depending on their. Psychological and sort of neurological thing that might be a really difficult thing to do. Right. And I know that a lot of technical people do struggle with that kind of thing. I'm naturally very extroverted, so it's easy.

[00:30:53] So I don't wanna make it sound like it's always gonna, it should be easy. 'cause you're absolutely right. That is an enormously hard thing to do. It is enormously hard because you have to hold that vision. You have to manifest the vision and the fact that you could do it. You have to fundamentally be existentially.

[00:31:08] Positive and, and, and optimistic if you're going to push any kind of organization to, you know, the promised land. But I think the way you get through that is developing an internal capacity to sort of manifest two things at the same time. So you have to manifest for yourself that, that honesty and like, oh, here's what we need to do, but I really believe we can do it.

[00:31:31] But, but the other thing is that it's sort of a, it's a customer subjectivity. So you're gonna be your own. You better be your own harshest critic, or you're not gonna make it right? So you're gonna be your own harshest critic. Your customer isn't. Your customer is looking at you for the same reason why it's hard to sell stuff.

[00:31:50] You also actually probably have a bit more leeway than you think, right? Which is the customer. If you use the Jobs to Red done framework for product development, your customer is hiring you to do or hire your product to do something, and all you have to convince them is that it does that one thing better.

[00:32:05] Whatever they've got currently going on that it's worth them making, you know, taking, taking the leap and making the switch. You don't have to solve, like when you are building the product, you, you feel like you have to solve the world's problems, make the best possible thing. But every single customer only has some subset of, of those things that are value propositions for them.

[00:32:22] So if you just handle it for them, well that's great If that, if some prospect customer comes in and says, this is not good enough for me. Great. Maybe in the future you'll go back to them and it will be good, so you move on to the next prospect. And so manifesting that like other center of perspective, it's very, it can be very liberating in the sense that you don't have to be such a harsh critic for yourself.

[00:32:44] You can just say, well, what does that customer really need? Out of all the things we've got, can we cobble something together? Can we put this in a way that it solves that problem for them? And if it does, they'll be happy to pay you for it. And that's great. And then you come back later when you've got other, all this other stuff that you think is gonna be even better.

[00:33:00] Like in the, it's also tied to like don't sell past the close. It's tied to all these kinds of things. Where as the founder, as a visionary is all these things. You could see all things. It could be. But you don't have to sell all of those things to every customer, right? And, and that in that, in that space, you can find a lot of forgiveness for yourself and find a lot of peace, hopefully.

[00:33:21] Does that make sense? That's like a lot of hand waving, but hopefully that

[00:33:24] Omer: Totally, yeah. The hand waving is fine. Totally makes sense. And I think the way you put it, it's, it's. It's such an, it's an obvious thing, but I think the way you described it, I, I, I, maybe I haven't heard that before, but often I think founders when they're, they're building a product, they look at, you know, the, the competitors or the incumbents.

[00:33:46] They see something that has a hundred features and they're thinking, oh my God, I, I just got this one or two things here. There's just a huge feature, you know, feature gap between what I can offer and what these products have. Number one, often it's taken decades to get to that point, but what you pointed out, hey, how many customers are using all those features in those products?

[00:34:12] And if you can figure out, I. You know, which subset of customers are using a small percentage of those features and just focus on them and help them do a particular task better. Like, you don't have to go and build like, you know, all those features and spend a decade before you launch the product. Right?

[00:34:32] Which kind of is, is often how you feel like you, you know what you wanna do. Let, let, let's talk a little bit about the. You know, we, we, we talked about the, the open source side, the, the enterprise business. So now that's getting traction. You know, you, you've got more and more these on-prem solutions in place.

[00:34:51] The open source piece is growing and then, you know, as, as you know, you told me, as you start to hire people, it becomes like really. Confusing for a lot of people coming in. Like, I, I guess the question I asked you at the beginning, like, why are we doing these things? Why, why are we, you know, like, and, and you had like two opposite kind of perspectives depending on where somebody was in the organization right now.

[00:35:18] Peter: Yeah. I, there's for us it was particularly hard for two reasons, not just one, right On the one hand and the first one is that we, we did, we have a very broad, it's a, it's a weird product if you think about what we have. 'cause we, although we write some of the open source that's in what we ship.

[00:35:36] We don't write most of the open source. We're really a distributor. We we're, we compile and build. And that can be very, very nuanced and tricky. So there's value that we add there when we do that. But nonetheless, the upstream source code comes from other people. So it doesn't look like a traditional software product.

[00:35:50] I mean, if you think about it, if you're a non-technical marketer or salesperson, someone who's not technical, but you go into a tech company and you ask them, well, what's your product? Okay, it's this thing, it's a box of other people's bits. And you're like, well. Like, how am I gonna, what, what? It's a box of other people's bits and, and so like that is kind of a weird thing to try to sell in any case.

[00:36:11] But then, then additionally to that, we have this weird thing where, because we pushed and sort of broke new ground and created this Python data science, you know, that job category didn't exist when we started the company because we made the tools available. That job category became a thing and people started using our tools.

[00:36:33] And so there's grassroots adoption inside businesses. But those people are, are oftentimes new. They don't know how to procure. They know how to buy. They know how to navigate budgets inside enterprises. And meanwhile, enterprises, they have a love hey, relationship with new technologies, right? They know they need to be innovative.

[00:36:50] They know they need to stay on top of this stuff, but they also just feel like they're constant, the technology treadmill. So there's always internal corporate inertia and, and pushback headwinds against new things coming in. So you have to convince those buyers, the economic buyers. The IT people, the the compliance safety, CIO people procurement.

[00:37:10] You have to convince them that this new tech is good, and by the way, you should probably buy this thing to support these people using the free thing. So we didn't have a simple like, oh, here's, you know, a screwdriver. If you want more screwdriver heads, buy the plus plus subscription. Or here's a database.

[00:37:24] If you want more rows, buy the subscription. Our problem is a very weird thing. Here's all this free, amazing stuff. It works great. None of it is limited in any way. Use the crap out of it and we start using it a lot inside your business, the the managers are gonna wanna have some way to govern it. They'll wanna have some of the stuff running OnPrem.

[00:37:41] They want better collaboration tools, and we sell through that way. So then internally, what that means is when we hire people in, we'll hire someone who's doing marketing or we hire someone who's doing sales. They have to, in their heads, wrap their heads around. Number one, the product is a box of other people's parts.

[00:37:55] Number two, you're not upselling. The straightforward upsells of these things. You're not saying, oh, you could do a million rows. We wanna do 10 million rows. You gotta pay us. You're not upselling in that traditional way. Your upsell is actually not even going to the people who are using it, your upsells to this other route, which is these other stakeholders.

[00:38:13] So it was very complicated, and if you get someone who understands the practitioner aspect, they sometimes don't understand how to do top-down marketing to the B2B economic buyer. But vice versa, you buy, so you get someone who's experienced, who has a Rolodex salesperson, they know how to approach the B2B buyer, but they don't actually, for them, any investment in the open source, the community, the practitioner tools is wasted because you're not selling that stuff like that stuff, you don't make any money.

[00:38:40] So why are you putting your developer resources onto making the enterprise product better? Right. And so it's a really, it was, it was just so many dimensions of hard. If I had known that that was how hard it was gonna be, I would've tried to engineer other kinds of products, right. That would not have this con confusing and conflation that happened with this stuff.

[00:38:59] Omer: It it, yeah. It's fascinating because, I mean, it's like, as, as you were telling me earlier, you've, you've, you, you were in a situation where some people are saying, we've got this massive open source. Community, why are we getting distracted with this enterprise thing? And then people on the enterprise side saying, we are, we are generating revenue from this.

[00:39:20] Why do we need this opensourcing? But really that was how the flywheel was, was spinning, right? Because you, you had this massive distribution and then people are starting to use this in their organization. They don't need to pay for anything. They don't necessarily need permission as long as they can.

[00:39:39] Install Python and, and you know, Anaconda and get, get started. And, and in many cases you don't have the email address, so you can't even, you know, even if you wanted to do some kind of upsell or marketing, you can't do that. So how, how did that actually happen? Like, so you've, you've got, let's say like a million people out there who have downloaded Anaconda, who are using the product.

[00:40:06] Some of them are in enterprise you know, organizations. Is it all inbound? Like eventually when some manager says there has to be a better way to do this, and, and them coming to you and saying, you know, give, give us, you know, like, like the first customer saying, can you give us an on-prem solution and we'll pay for it?

[00:40:25] Or will your, will your team finding ways to. Do outbound as well in these, these,

[00:40:33] Peter: it's a little of both. It's a little of both. I mean, over time you know, we don't have to advocate for the use of Python data science anymore. People do that, right? It, it python's a language of ai. So that's what it's, but we do have to still drive awareness for people about the vulnerabilities that might have in adopting just purely open source approach.

[00:40:51] Not it's always open source, but just using. Things from the, the free and community side of things. There are, there are, there's real value in them securing their open source supply chain. That's a term that is now, you know, it's actually in executive orders. There's NIST standards around this kind of thing, and, and people can't ignore it anymore.

[00:41:09] So that is certainly one of those like market shifts that has been a, a tailwind for us and it's helped us. You know, find receptive ears when we talk to people about this kind of security stuff. So there's definitely an outbound aspect of the marketing around this. And we could do a better job of.

[00:41:24] Of fixing up or cleaning up some of the UI bits for what we do install that people know about other value added services, even as a practitioner or single practitioner that they can upgrade and get from us. We're, we're sort of, we, we've split our focus in many different ways over the years, but we definitely, like, that's an area that we are trying to invest more in.

[00:41:46] It's just getting that direct. Practitioner consumption of our cloud services or using us to share more of the things they build with other people, share how easy it can be to do that. There's a, there's a lot of that kind of stuff that we you know, everything takes money. And so doing that kind of outbound marketing takes money.

[00:42:02] All these things take money. And so we've had to sort of, sort of peanut butter some of our investments in this way. I mean, the funny thing is we go to a conference, we get very bimodal sort of responses. Some people will say, oh, Manon, I use you, you guys are great. And then it's like, whoa, you guys have enterprise products?

[00:42:18] I always thought you were a nonprofit. Like people literally, you know, and that's, that's not great given that I'm like, you know, looking at the p and ls all the time, being like, okay, how do we like balance the needs for more engineering resources versus this and all that? And so it's like, oh, we thought you were a nonprofit, but now that we know you got an enterprise product.

[00:42:34] This is great, we can actually use that. So that tells me I need invest more in some of that outbound messaging. Right. So these are the kinds of things that we sort of, you know, even now are still trying to figure how to dial in.

[00:42:43] Omer: One. One last question before we go into the lightning round. How does how does the world of chat GPT and people now suddenly using prompts to, you know, query all kinds of different data sources?

[00:42:56] Like how does, how does. How do you see that impacting what you're doing?

[00:43:00] Peter: I think that it will, so, you know, there's definitely this, this sentiment that chat, GBT will just write all the code for us. So why do we need to learn Python or learn any of these other things? My thesis is that right now it is humans writing code and, and then, you know, we get it to where computers can be writing code, but.

[00:43:21] But it's important to have a language that the computer's writing the code and that we can verify. So, so we might not be writing a lot more pi, more of the Python code or code being written in the world in general will be done by, by computers and machines. But we're still gonna wanna have an audit trail.

[00:43:37] And so that case, it's not about the language that a person can write as much as the language that people can read and maybe tweak and, and verify on their own. That's still Python. It's an easy language to read. It's an easy language to write. It fits in most people's brains. You know, it's actually, so I think even, even if we buy into this idea that the LMS will do all this coding for us in a short, short, in short order having a language that we can read and audit and run independently to get this to verify the results is super critical.

[00:44:08] I mean, the last thing you want is a machine that says, well, I generated the code, spun up all these docker containers. I ran all this code, and here's your result. The answer's 42. And you're like, what? Let me go try that. And you go look at the Dogger container and it's spewed out. It's, it's put out 10,000 lines of c plus plus and then it's got a weird make file of another thousand lines.

[00:44:26] You're like, oh my God, there's no way I can verify the answers. 42. Right. So I think it's really important, even if we move to mode where they're programming now, we have auditable things that, that are human readable.

[00:44:36] Omer: Yeah. That's a great answer. I love that. Okay, let's let's wrap up. Let's get into the lightning round.

[00:44:41] I've got seven quick fire questions for you. Alright. What's one of the best pieces of business advice you've received?

[00:44:48] Peter: One of the best pieces of business advice is that it's, in general, it's all about meeting other people's needs. So whenever you're thinking through a problem, when you're thinking about what do I do?

[00:44:59] Or Why am I struggling with this, or what's wrong, I mean, whenever we're in a pain fear mode, we are very, you know. We're trying to meet our own needs, but in general, our problems come from not meeting other people's needs. So it's really important that you develop the conscientiousness and the consciousness to be able to have a state of mind that that outside of your own subjectivity frames and holds the other person's subjectivity and understands, well, what are their, what are their needs?

[00:45:25] How do I meet their needs? And one other one's he asked for one. But I wanna give you another one that's really important. That I wish I had learned earlier, which is in general in life, but it's true in business. Boundaries are loving. It's not, it's not wrong to say, here's my boundary. This is all I can do for you.

[00:45:42] As a founder, you know, you, you, you, you're sort of the person who does everything of last year's award. You care, you catch all the balls as they drop. But in relationships in general, in life, but also in in client relationships. You know, it's okay to fire a bad customer sometimes. You have to with employees.

[00:45:58] I, you know, I, I actually had sort of my, my, my emotional, mental breakdown at one point was because I was trying to be the counselor and the camp coach for a hundred people, and that was just too much. Like my internal emotional thermal sink was completely deregulated and dysregulated. And so I. So I, I, that's a, that's an important lesson that I hadn't learned, but I've learned now, and I wanna make sure everyone else, as many people as possible understand that it is, the boundaries are loving.

[00:46:27] Omer: Right. What book would you recommend to our audience and why?

[00:46:30] Peter: I, I really think Four Steps to the Epiphany was a really fantastic book. It, it's hard to say Why. It just had a lot of good bits and pieces. I really, I, I was just, I loved it. I thought it was a fantastic book. But another one that people should definitely read is Innovator's Dilemma.

[00:46:43] That's a fantastic book. People should definitely read that book.

[00:46:46] Omer: Great. What's one attribute or characteristic in your mind of a successful founder?

[00:46:51] Peter: Well, we already kinda touched on this, but I think it's the ability to simultaneously manifest existential optimism while also being completely paranoid about everything going wrong.

[00:47:03] And, and this idea that courage isn't a lack of fear. Right. You, you have to see all of it. You have, it does fear, like you feel it, but you do it anyway. And so I think that is an, an attribute that's really important is being able to manifest both those things.

[00:47:18] Omer: What's your favorite personal productivity tool or habit?

[00:47:21] Peter: I just take a lot of notes. So for a long time was an Evernote user. I switched over to using Joplin. I'm trying to put other things in obsidian now, but it's basically just I'm scrapbooking ideas. I'm putting meeting minutes down. I have like. Readings and other things, I, I really wish I had a better tab browser organizer, so I don't have like a perfect system for this at all.

[00:47:40] But I definitely feel like having a single place where I take all these notes, I. Is is just been super critical. I couldn't, I couldn't function without that.

[00:47:48] Omer: Cool. Yeah. I'm trying to use obsidian more and more as well these days.

[00:47:52] Peter: It's fantastic. Yeah.

[00:47:53] Omer: Yeah, yeah. What's the new or crazy business idea you'd love to pursue if you had the time?

[00:47:58] Peter: Well, there's some things that are AI related and adjacent but I think the crazy of it and this is something I'm definitely going to do is create a network a sort of research network. A research group, maybe institute is too big of a name that really is in support of looking for.

[00:48:18] I. Weird and interesting physical phenomena that we don't have explanations for. So you can call it sort of fringe physics, but not like just crazy random theories that people have no evidence for, but actually fringe, sort of evidentiary physics. Like what are the weird things that people observe, but we, we actually don't understand or can't, can't explain.

[00:48:36] And creating really a single place where people can catalog all these things, talk about them, and we can really start to start understand some of the stuff. So that kind of, radical physics or heretical physics Institute is, is sort of my crazy idea.

[00:48:50] Omer: What's an interesting or fun fact about you that most people don't know?

[00:48:54] Peter: I'm, I'm really into astronomy, so I've been an amateur astronomer since I was like 13 or 14 years old. And I have a really big tele, I have, I have unfortunately too many telescopes, but one of them is a really big telescope. It's a 28 inch diameter pretty large telescope. And I, and I do, the thing that I do that most even astronomers don't do is I use military night vision equipment.

[00:49:15] To look through the telescope and that is an amazingly cool thing. And you can see Nebula, you can see all these things that would otherwise be photographic objects, like you have to photograph for hours and do long exposure. But with military night vision gear and with the right kinds of filters.

[00:49:29] You can see amazing things in real time even in smaller scopes, but in a big scope. It's really spectacular. So I, that's my hobby.

[00:49:37] Omer: And finally, what's one of your most important passions outside of your work?

[00:49:39] Peter: I, I'm tied into a group of thinkers that are really looking into, like, I. What is the future?

[00:49:47] What should the future operating system for civilization really look like? How do we do infinite game, not zero sum games kinds of dynamics and really bring that to life? And this does tie into my, my open source work because I think open source software, you know has been, is one of those proof points that human collaboration at scale can yield incredibly powerful artifacts that is.

[00:50:09] Way more effective than, than what capitalism and even market economics has produced. And as we get to a world where we have more and more technology to suffice human needs, we need to start thinking about how it's not about limits of growth or degrowth or anything like this. It's more about how do we as a collective action principle organized humanity so that we, we don't keep running, you know, burning up the earth and, and burning up our own internal psyches and, and whatnot, chasing.

[00:50:37] This, this infinite growth, but actually having a mentality, a growth mentality that, that actually can, can rest in the, the biosphere that we have, but still produce a lot of really wonderful things. So that term my friends I've taken to calling game B instead of game A, which is the current system of the world, hypergrowth, hyper financialization, all these things.

[00:50:57] And and so if people are interested in that, they can look for some of my podcasts I've done in, in interviews and, and writings around Game B.

[00:51:04] Omer: Super interesting. So, th thank you so much Peter, for joining me. It's been a great conversation. I. It was great to just unpack the Anaconda story and, and a very, very unique business and very different to, you know, 99% of the types of, you know, SaaS businesses and guests that I have on the show.

[00:51:24] But it definitely, I think it opened my eyes and hopefully it'll, it'll get, you know, people listening to this, you know, maybe thinking a little bit differently about things that. They could or, or should be doing with their business. So I appreciate you making the time to do this. If people wanna check out Anaconda, they can go to anaconda.com and if folks wanna get in touch with you, what's the best way for them to do that?

[00:51:45] Peter: I am on

[00:51:45] Blue Sky @wangsocial. I was actually Angel investor, founder of, well not founder, but angel investor, blue Sky. I am also on Twitter @pwang, but I've been sort of moving to Blue Sky as my primary social thing. You can also find me at LinkedIn. My dms are open there on LinkedIn. And one thing I would say is people can also use Anaconda directly through Excel by typing equals py in their formula bar, and they can now type Python code and that runs on Anaconda backed container and Azure.

[00:52:12] So so you don't even have to install anything. You don't, if you go to a webpage, you just use, you already have Anaconda if you have Excel.

[00:52:17] Omer: So that's a great integration.

[00:52:19] Peter: It's, it's fantastic. So yeah. Well thank you so much for having me. This is a really fun conversation. Thanks for asking me that really beautiful question.

[00:52:26] The start that. No one's ever asked me before. Yeah. Thank you so much for having me. This has been a pleasure.

[00:52:30] Omer: It was my pleasure. Thank you. And I wish you and the team the best of success.

[00:52:34] Peter: Thank you so much.

[00:52:35] Omer: Cheers.

Book Recommendation

The Show Notes