Ry Walker, Co-Founder and CEO at Astronomer, helps organizations achieve their data-driven initiatives by providing a platform that allows “citizen developers” to easily create and monitor data pipelines. Use cases include centralizing data from silos into a data warehouse, real-time data processing and algorithms, machine learning, and data sharing with partners and customers. He has twenty years of experience and previously co-founded a number of Cincinnati companies, including Sharkbytes, US Digital Partners, and Differential.
ML: As the company is fairly new, why did you feel that now was the best time to start an organization in this space.
RY: The interesting thing is we started with an analytics company called UserCycle and quickly realized that the space was really crowded. We were doing something called user analytics, similar to what Mixpanel or Amplitude or Kissmetrics does. After we started that project, we realized how incredibly crowded that domain was. Not only were the big guys there, but a lot of new companies were forming around the idea.
The other thing we realized was even when we made a sale and had a customer buy our product, it was really, really hard for them to send their data to us. It took 4-6 weeks of nagging just to get a trickle of data. We realized the reason for that is companies really don’t have the ability to wield their data right now; they don’t have the actual tools to. If they decided today to send a certain portion of their data to you and stream it in real time, it’s not happening. It’s not like they have a tool to just flip a switch and send it to us.
We were thinking, with all these analytics companies and a lot of new machine learning data science thinkers out there, people are going to need to be able to send their data. So we decided to pivot to that problem of trying to make it easier for companies to send their data to their partners. The converse of that is to allow analytics companies to collect customer data.
Really, we built an analytics company. We were struggling to get it going and decided to shift down to a problem that we thought was pretty open. There’s a big opportunity in the space, especially around customer experience. I think a lot of the reason companies aren’t using these tools is because they’re just too hard to use.
ML: I’m sure you had to do this before, but what would be the elevator pitch here?
RY: We’re building a platform to make data transportation easier. There are two parts to this, really.
First, we’re building an open source JavaScript framework, which has connectors and recipes, so you can find open source code. If you need to pull data from SalesForce and send it to Redshift, for example, the entire code base to make that happen is open source. We’re building all that as part of Astronomer. We think it’s really important to start building a community of sharable code that we can use and stop trying to reinvent the wheel every time we want to do something that we’ve done before.
The other part is a place to run those data pipelines. Let’s pretend you needed to grab data from SalesForce and ship it off to S3. You could write the code for that but then where are you going to execute that code in a reliable, secure way, with good monitoring? We think there’s a big opportunity there to provide a hosting platform for these data pipelines.
Basically, we’re building open source software that defines the task and then a really powerful cloud solution to execute those data pipelines securely, reliably and with a really nice monitoring solution.
ML: You mentioned at the beginning that this is not where you started, that you started in a slightly different space. Can you share with us the journey? Has it been very much like a startup, where you came up with something and showed it to people, got their feedback and then reworked it? How did you go through the process from where you started to where you are now?
RY: I think it’s the power of lean startups. I’m one of the organizers for the Lean Startup Meetup here in town. I’ve been into it for a long time and what you learn is that by trying to execute, to get customers to pay for your product, you’re forced to go through the entire lifecycle of the customer journey. It’s going to be horrible in certain places because you haven’t built out everything. That’s where we learn.
So we absolutely started with the idea of selling the product before it was ready, to figure out where it could fall down. With the analytics company, we were able to sell it and market it…none of that was a problem. The problem was on-boarding customers. It was so painful that it was severe enough to cause us to not pursue the business.
I’m a big fan of this thing called “running lean.” So we pivoted; we kept the same customer type but we pivoted the solution from, “Hey, here’s an analytics tool” to “Here’s a tool that lets you get your data to your analytics teams.” It was a solution pivot, which most people typically wouldn’t do, but we saw that there were deeper problems we could work on.
ML: I can imagine different types of organizations or industries are your customers or are going to be your customers very soon but, right now, who’s your ideal type of prospect or ideal type of organization that you want to work with?
RY: There are companies that call themselves analytics companies. They have a real need to collect a lot of data from their customers in order to execute on their business, so they really need to have a strong data pipeline. They may have a lot of different sources they need to pull data from and it’s pretty daunting to roll your own data pipes. We’ve met a lot of companies that have gotten halfway there and then realize this is a never-ending problem and they’re keeping it all to themselves. So we’re a big help for those companies that hit that point of reflection where they say, “Let’s not try to reinvent the wheel and write all this code.” We’re perfect for that.
The other primary target for us are data scientists within large companies. If you just joined a big company and you’re the first data scientist in there, you’re going to find data spread out all over the place. With the proliferation of SaaS and Microservices, data is in a lot of different silos now. So we really just jump in and help them pull it together into a data warehouse.
ML: How does one get started? So, let’s say an Analytics company decides to sign up, what do the first 30 days look like?
RY: The nice thing about us right now is you could contact us and say, “I need to be able to collect data from these 10 sources. I need to be able to send it to SAP HANA.” We’ll handcraft that entire solution for you and, in the process, we’re going to build open source connectors for everything. To get started, a company can just contact us at astronomer.io or request a demo, and we’ll create a plan based on their unique needs.
ML: How did you arrive at the name?
RY: I mentioned how we pivoted from analytics to data collection, but what’s interesting there is a pivot is usually a series of other pivots. One of the things we were doing initially was building something primarily for Meteor apps. Meteor is a JavaScript app development framework that we were using as Differential to build all of our customers’ web and mobile apps. Astronomer, originally, was the companion for these Meteor apps.
So Astronomer was created because it matched with Meteor. It was like, “Oh, they kind of have a ‘spacey’ thing” and we picked a name that was “spacey” too. We also looked at it like this: the Astronomer studies this array of light dots and tries to make sense of it, which is what we think analytics, or a data scientist, does too. Data scientists are kind of like modern age astronomers.
ML: Let’s say our company decides to adopt the solution. What should the company be measuring within 30, 60, 90 days to really see if your product has been beneficial to them?
RY: It’s tricky to say, “What’s the value of being able to analyze data that you didn’t have before?” but the thing I would say is that a lot of companies that are doing data science or customer science are generally doing it on revenue-producing activities. It’s usually not for cost control type stuff.
So there are a couple of hops here: We’re enabling data science, and data science is usually enabling revenue.
ML: Why do you think now, and maybe over the past 2 or 3 years, we’ve been really reaching this kind of tipping point where everyone started to get on the bus of Analytics?
RY: I have a couple of theories about this.
Mobile devices: everyone has a super computer in their pocket now. People are using Fitbit, for example, which is visualizations and graphs. I think that the rise of mobile has really increased the technical competence of a lot of people who normally wouldn’t be very technically competent and I think they’re getting more ambitious for what data can do for them. That’s one factor: the rise of mobile, wearables, and computers.
I think that there’s also a rise of SaaS, and SaaS has been happening since 1998. It’s been close to 20 years since the first SaaS company came out. Do you know anybody who’s building on-premise, packaged software at all anymore, for businesses? Not many. There are games. People are building things for computers, but they’re generally personal applications.
The final factor, I would say, is that the promise of cognitive computing is very interesting. I think that once we have self-driving cars around us, we’re going to question, “Can computers do a lot of the jobs that we humans are doing right now?” The world’s changing and companies can either be left behind on their own little data island or choose to get into this stuff and hopefully keep business going.
ML: How do you think your technology will contribute to the success or evolution of the Internet of Things?
RY: We have several Internet of Things customers. Let’s say there’s an IOT company. To analyze what’s going on with a manufacturing line, you need not only data from those devices but you also need other data from the company. You may have data in an Oracle system that you need to merge in with the IOT devices if you want to get a real picture of what’s going on.
I actually think IOT is creating more data silos, and each one of those devices is essentially broadcasting data somewhere. It’s generally not going straight into a company’s Oracle database. It may be going out to some third party partner. It’s contributing to more data communication and, 5-10 years from now, the amount of computer-to-computer communication around data is going to quickly surpass human-to-computer communication. The computers are going to be taking the stuff over soon.
ML: The question I feel organizations are still often struggling with is somewhat of a difference between buying an Analytics solution and using it. Similar to buying Organic products and living a healthy lifestyle. So how do you think we can bridge that gap and spend money not just on tools, but on actually using them?
RY: I’m a big fan of telling companies that they need to own their own data vs. completely relying on thirdparty products. I just see too many companies using five analytics products because they were sold five analytics companies. They’re not using it because they’ve made any conscious decision.
We just started using HubSpot at Astronomer, and you either have to buy in to the whole HubSpot theme or you can stay independent of that. But once you buy into the tool, you have to use it. I also say, “Don’t be 100% committed to that tool. Have a copy of your own data. Control yourself a little bit more than just being completely controlled by vendors.” Think about it and have the ability to try tools and shift around. If you send all of your data only to Google, and they’re the only ones that have your data, they have you locked in pretty good. I say keep a copy of it yourself and send them a copy. If you have your own data, you can potentially have some paths away from full reliance.
ML: There have been many articles about Cincinnati, how amazing it’s becoming and I’m curious on your perspective. Do you think by launching the company in Cincinnati you have an advantage?
RY: I think there are some definite advantages. For one thing,there is no major tech company in town, sucking up all the tech talent, so I believe that the best talent in Cincinnati can be recruited if you have a great company. We have so many developers we could hire. That may not be true for every company but the quality of the company is a big factor in that equation. If you’re a solo founder with a questionable business idea and questionable technical understandings, I don’t want to go work for you as a programmer. Competent companies have a greater advantage because there’s not a lot of other competent companies in town, so you can get to the point where you are that company.
Another advantage is that dozens of mid-sized American cities are within an 8-hour driving radius. Many of these are largely ignored by startups in Silicon Valley but comprise a significant percentage of the national GDP. Cincinnati is not only accessible but also relatable for these cities.
The disadvantage, of course, is the access to investors is lower and it’s hard to really be connected to what’s going on in San Francisco without being there. I’m a big believer that despite all that, you do what you can to build a bridge to San Francisco. It’s kind of the tech mecca right now and a lot of great thinking and investors are there. There’s also a great ecosystem in LA and a great ecosystem in Chicago…so there’s precedence that you can build a great company outside of San Francisco, in Cincinnati.
ML: Now on the subject of investors, it’s often said that investors invest in people and invest in teams. Can you tell us a little bit more about your team and what makes it unique?
RY: I think that’s absolutely true. They’re looking for as few gaps as possible, but they’re also looking for something extraordinarily interesting. If you’re trying to build a robotics company and you came out of the MIT Robotics lab, that’s extraordinarily good. If you’re just some guy, though, that was selling insurance but has this idea for a robot, that’s an extraordinarily bad match. So I think the first thing to ask is, “Do the founders match? Do they have any reason to be doing this?” More than half of the time the answer’s probably “No.” Both my co-founder and I have experience running running tech startups and have a huge, shared vision for the problem we’re tackling. Running Astronomer makes sense.
The other big factor is “How well rounded is the team? Do you have a technical person? Do you have a salesperson? Do you have marketing? Do you have design? Do you have all the pieces?” From early on, our first five people were two technical people, a data scientist, a marketing person and a design person. We felt like that was really practical. If you want to build a whole company, if you want to be a lean startup and deliver value as soon as you can, you have to have the whole team in order to do that. It forces you to have a team that can deliver the entire equation right from the start.
ML: Can you give us a preview of things to come? Over the next 6-9 months, what can we expect from Astronomer?
RY: I believe there’s going to be a big growth in the need for companies to transport data from point A to point B. You can find these products inside of other products, but I actually believe that just having a dedicated product to help move data around is critical.
Right now, what we’re building is what I call “dumb pipes.” They don’t really do much more than what you ask to move data from point A to point B. We believe there’s an opportunity to build some intelligence into them so you can do data quality analysis as the data flows; give alerts to say, “Hey, this field changed. It used to be 1s and 0s; now it’s Trues and Falses.” There are some very basic algorithms we could run across these things to be more intelligent than dumb.
That’s the big thing for us. We want to build the infrastructure to move data but then we want to build intelligence on top of that, which would be a real differentiator. If suddenly your data went from 100 records a day to 10 records a day or from 100 to 1000, it’s probably important to be notified that there’s been a real volume metric change. We think things like that are low-hanging fruit that we can build to deliver additional value.
Even now, we’re offering that intelligence piece through our machines + humans hybrid. As we build “dumb pipes,” we have a human dedicated to helping companies understand which ones to implement and how to implement them well. In the next 6-9 months, you can expect to see us connecting with many more companies. We’ll work with them to solve individual problems, and we’ll continue to build a platform that’s increasingly flexible and impactful across a wide range of industries.
Resources
You can find Ry on Twitter @rywalker and check out Astronomer.io’s blog, which covers everything from culture to data science.
