Originally aired on March 5 @ 1:30 AM - 2:00 AM EDT
This week is Cloudflare's Founder Spotlight on Cloudflare TV, featuring dozens entrepreneurs from across the tech industry and beyond!
This session features Josh Curl, co-founder of Hightouch.
Josh Curl is cofounder/CTO of Hightouch. Before founding Hightouch, Josh was an engineer at Segment helping drive the development of the Personas product and transition of Segment into the Customer Data Platform (CDP) space. Prior to this, Josh was one of the first 10 engineers at Rancher, which was acquired by SUSE in 2019 as one of the largest exits in the Kubernetes space.
to find the rest of these featured conversations — and tune in all week for more! Founder Spotlight hub Alright, we are live for the Cloudflare Founder Spotlight of Joshua Curl, Josh Curl, Josh Wgit, you might know him on Twitter. He is a very good friend of mine. I'm Evan Johnson from the Cloudflare product security team and the Cloudflare security team. Josh is the third interview I think I've done this week. And he is the CTO and co founder of high touch data, the one of the fastest growing companies that I've seen anywhere, really, at the current time, but also in terms of valuation. And then especially in the data space, I can't think of anything compared in the compared to high touch in the data space. So I'm really excited to have you on today, Josh. Do you want to give a quick intro about yourself? Yeah, I'm super excited to be here. It's gonna be fun being on Cloudflare TV. I'm from Michigan originally, I worked with Evan at segment for a while. I was also an engineer at Rancher Labs previously. I've been doing startup stuff since the end of I guess, mid 2019 now. And then been working on high touch for about a year and a half with my co founders, Tejas and Cash, based out of San Francisco. Love it. So it's funny you mentioned, so the way Josh and I met, I don't really know how we originally met, but he had followed me on Twitter at some point. And, and he eventually came to work at segment. But the way he came to work at segment was I just got up one morning and I had a DM in my inbox. And it was from somebody, hey, can you refer me, refer me to segment, I want to come work at segment. And I clicked on his resume, and then clicked on his GitHub, and then read all of it. Well, not all of his code, but I clicked on some of the links, tons of go code, we're talking like, dozens of repositories and hundreds of commits over the last year. And I instantly referred him. And now I owe Josh a steak dinner, because and it's kind of a running joke now because I think dinner. Mythical steak dinner. Yeah, I'll take it. If you come visit me in Austin, I'll definitely come take you out. I got a small recruiting bonus from that. But I didn't know the guy it was. And that's how we met. Then we worked at segment together. And, and it was great working with him. So I want to drill in to, I guess, starting at leaving segment. So you and Tejas, I was, we're supposed to have Tejas on today. And we don't. But because he's, he's a little under the weather. But you, you both left around the same time and both started different things. Yeah, I like to think that I inspire Tejas to leave segment a little bit. I think I left, I gave notice before him, he was kind of already thinking about doing, doing other things as well. So at the time, I had an idea that I was working on that completely independent of high touch was actually a software company, more oriented around IoT and edge computing. And so that's something that I was excited about. And I ended up leaving segment to pursue that idea. And then I think that's was the thing that pushed Tejas over the edge to leave segment. And so he ended up joining my other co founder, Kashish, I'm working on a company called carry travel. At that point, they had already gotten into Y Combinator in the summer of 2019 batch and sort of have that lined up. So he jumped right into that. And then we were sort of on parallel tracks for a while, both working on our own things. Got it. So you started, you started device plane, and you raised money for that. Yep, we did. I raised a preseed round at the end of the summer in 2019. And just for a little bit of context of what device plane is basically, the idea was, there are a lot of companies that use Linux based devices, or servers as part of their provide their offering. So it could be like a server running inside of an autonomous vehicle, it could be a Raspberry Pi strapped through the drone or a robot, those sorts of things. And so the pain point I was trying to address was basically doing over their updates for these kinds of devices. So primary industries and verticals I was looking at was like edge computing, machine learning on the edge, and robotics and autonomous vehicles and like those sorts of things. So yeah, embedded devices, I'm sure too. Yeah, embedded devices, but all kind of focused on like Linux based devices. Nice. And so you raised money, started that, and then Tejas and I haven't met the third co founder, actually. What's his name again? Kashish. Yeah. Kashish. And they were doing Cary travel, which went through YC. And then at what time did you all start to get the idea for Hightouch and start working on that? That was actually a bit further into 2020. And so we both worked on our respective ideas for about a year. And I think specifically Cary had to go through a bit more of an abrupt pivot given COVID. So they were doing a lot of services around travel, which kind of abruptly stopped in March of 2020. And so they started to just think of new ideas like what to work on. And they actually went through a few pivots even between Cary travel and what we were currently working on as Hightouch. But in that span of time, that's actually when the name Hightouch was first born. So originally, Hightouch was actually a customer success tool to help you manage customers over Slack. So think of like Zendesk, but on top of Slack rather than email. That was the original idea. And the idea was it was a high touch customer experience for doing support. This is actually before I joined. And then what happened was after I joined, we were going through many other pivots. We decided we actually liked the name Hightouch as a general thing where most of what we were doing was around the theme of helping people serve their customers. And so the name actually still kind of stuck and we just saved the domain and the name through all the pivots. I remember that Hightouch because I tested it. The Slack automation where you could manage multiple Slack channels from one central place, basically. A lot of shared Slack channels where you could talk to them in one place. Yeah, exactly. And I think even to this day, we're still not sure if the original pivot was a good or bad idea because we're hitting the point now where we have a lot of customers in shared Slack channels and we're like, okay, I wish we had a tool to manage these. It doesn't really exist. Yeah, definitely not. And okay, so leaving the segment, you said that was 2018, late 2018? It was mid 2019. Mid 2019, you both worked on your respective ideas for about a year, started Hightouch a few months into the pandemic and around fall of last year? And now you're about a year in. Yeah, we sort of came on the current version of Hightouch, I would say, in like July of 2020. And so it's about a year and a half old now. Okay, about a year and a half old. And so, I mean, most recently I saw you raise the Series B 40 million at 450 million valuation. And that's an insane valuation growth. Tell me what, I guess, what it's like raising money in the current climate and just how you've been successful there. Yeah, I guess all three of us have never really had to raise money in previous times. But from what we hear from others, it was a lot harder in the past and valuations were a bit lower. So I don't think it's ever going to be easy to fundraise for sure. But I do think in the current market, it's a little bit easier to go out and raise. I also think we did have pretty strong growth metrics, which really helped a lot. But one thing I think is exciting just in general about startups today is that there is more capital available, especially for seed rounds. And so if you have an idea, it is easier to raise, which means there's just more ideas being tested and more innovation happening, which I think is just a cool positive trend. But yeah, it was definitely a process for us to figure out who exactly we wanted to work with for the Series B and what kind of experience we wanted to add to the board and so on. And in the end, we ended up with Iconic, which is a great firm that's had a lot of other great data investments. Cool. So yeah, definitely helps to have a real business that's growing and everything when you're raising money. That's definitely something that I hope investors are asking about. How did you go about finding product market fit so quickly? And what did that look like your customer's journey? Yeah. So I guess this was the spring and summer of last year. It was a time where we just tried out a lot of different ideas. I think the way that we went about it was we had a lot of context on the customer data space from Segment. And so because of that, we had a pretty strong intuition of tossing out ideas that were bad. I think we were able to generate more ideas that were higher quality. But at the end of the day, it can be difficult to figure out what's actually going to stick and find interest. And so I think all in all, we probably went through about, I'd say, eight different iterations of ideas. And some ideas were just a slight alteration of existing idea. And it would just be a difference in the way that you pitched it, but kind of even the same code or product. And so we went through a couple of different iterations of the product. And at times, we were actually testing multiple things at the same time. And so it would just depend on what call, we would just see what was getting more interest. And so it was a phase of a lot of mass outbound on LinkedIn and just a lot of conversations. I think luckily, we were able to lean on our network from Segment a lot to just short circuit a lot of the conversations and just get thoughts from people that we really knew and trusted. And so I definitely think with customer discovery, having a domain that you're experienced in means that you'll accelerate your customer discovery by a lot, but there's still no silver bullet. You do at the end of the day, have to talk to a lot of people and do a lot of sales, know a lot of things. Because even if an idea is resonating, sometimes the implementation doesn't work out nearly as well. And so once you start to see repeatable sales with a specific product, that's the same across customers, that's when things start feeling really good. That's something we didn't really start seeing until August or September of last year. So that's fascinating. So you went through eight different MVPs almost. How are they slightly different? Can you permute them for me? Yeah, I can go through a couple of them, what the journey was. The Slack idea slowly morphed this idea. So I don't know if you remember at Segment, there was this internal tool that we have. It may have actually been built after you left, but it was the customer dashboard where it was an internal tool where you could search and find any customer and find out a lot of context about them. So it was like data from Salesforce, receipt tickets from Zendesk, what their sources and their destinations were at your segment concepts, as well as billing and so on. So it merged all of these things and gave you a UI, a dashboard into that. And so one of the first we worked on after the customer success tool was basically helping people build those kinds of dashboards. So you can imagine like Retool, which helps you build custom internal UIs, but really very focused on B2B use cases. And something that was a bit more data-driven where you can ingest from multiple sources and build a dashboard. This is sort of since emerged into a category of its own called PLG CRM. We actually explored this before that name or category really took off last year. And I think we found that people just wanted to see different things out of it depending on what role they were. And so in customer success, like what we found was there was kind of a big existing industry for selling to customer success. Like customer success tooling was actually pretty well established with like GainSight and Catalyst and other players. And so I think we were a bit worried that we were just building a different version of one of those things. But like what we found was we learned a lot about data warehousing and the data space through this. So what we started hearing more and more was that people didn't want to have us ingest data, but they wanted to ingest it from the warehouse. Like that's where they already had a bunch of their data. And so we kind of got the idea of like, what if we took that tool, but built it on a warehouse natively? And so we kind of followed that thread for a little while. And then I think we realized that we built this workflows feature on top of the customer dashboard where you could visually query data from your underlying data store and then do things like send it to Slack, send it to like Marketo and other marketing tools and so on. And so it kind of got us thinking like, what if we took like an audience segmentation tool and we built that on top of a warehouse and just tried selling that? And so that was like the next pivot. And then we started working on that for a while. I think we realized that a lot of companies didn't need the visual layer and they just wanted to sync data from their warehouse into different systems. They realized we could get into more companies quicker by having like a simpler version. And we were actually just wanted to like table and like bring the audience segmentation layer back. And so that's kind of like when the SQL sync core of Hitech was born. And our pitch was more like, you have data in your warehouse and you actually just need to move it into Salesforce, your marketing tools and so on. And that was kind of like the final pivot where we ended up like dropping anchor there. But what was kind of cool though, is one of our previous pivots, the idea of taking a marketer and allowing them to build a cohort of users on the warehouse and then syncing that without having to know SQL. That's actually since come back in an add-on product that we have called Hitech Audiences. And that's actually been doing really well. Now that we have like a pretty solid base, I think that we have more trust to do sales into larger companies. I think the value prop is a bit more for like mid-market and enterprise companies. And so we brought it back almost exactly one year later. It's been doing pretty well since then. Nice. That's quite a story. If you're like me, I went and Googled what PLGCRM was, product-led growth CRM, which is customer relationship management. That's so many different avenues that you went down. And I'm curious when people sign up for Hitech and when you talk to a customer, what is like the usual suspect thing that they're using Hitech for? So they want to get data out of their warehouse and like, what's the main use case? It's mostly customer data use cases more broadly. And then within that umbrella, the main two are sales and marketing. So if we're selling to a B2B company, it tends to be the sales team primarily first, oftentimes marketing. It could also be customer success use cases like loading data into Gainsight or so on. It could also be customer support, getting data into Zendesk. And then when it comes to B2C companies, it's usually marketing. That's usually the big one. But I think in general, the two biggest categories are sales and marketing. Gotcha. Makes sense. That's awesome. And so rewinding a bit, I'm curious, it's pretty obvious that travel and carry what Tejas was working on was not going to work in the pandemic. I'm curious, like if you had the retro device plane, I always thought device plane was a really good idea. I remember whiteboarding it a little bit with you and like how the GoBinary worked and how it like, updated. And I'm curious why, if you had to retro that, why didn't it work? Yeah, I think this is like one of the most frustrating parts about going through pivots is like when you look back on it, you're always operating off of limited data and then sort of extrapolating like what you think might be the case. And so I think like looking back on carry travel and looking back on our Slack idea and looking back on device plane, there was definitely hints of things that were very right about those businesses. But at the end of the day, there are certain things that add up to the point where you lose conviction about it overall. But there are times that you look back and you think, okay, maybe that wasn't like a bad idea. And there's like a different angle. And so you always, it always kind of like bothers you in the back of your mind that maybe there was something there. And now that high that you're doing pretty well, I don't have that nearly as much, but there was, there's a couple of different things, but I think one of the challenges was to generalize the infrastructure across verticals was difficult. And so I think a lot of times the requirements for what people wanted to do in autonomous vehicles was pretty different from robotics and so on. And edge computing was kind of different from that. And so I think the idea was to build something like Kubernetes, but for remote devices. And I think that the use cases for Kubernetes ended up being a bit more standardized across different domains or different verticals. Whereas what we were trying to do is people had pretty different requirements across those verticals. And so it was hard to build a product that appealed to all of them. And so the original value prop was very much around deployments. Turns out that's very hard to get into devices. It's, it was a very, very hard upfront sale. We had to convince people to take a bet on device point. And so the sales cycles were very, very long and it took a lot of convincing. And at the end of the day, it was hard for people to prioritize. And like people had their own built-in systems. It was a lot of like homeworld scripts and they would hire like a DevOps person to solve this problem. And once they did that, they didn't really want to touch it or think about it at all. So they got to a point where it was good enough and people would really, it was a lot of friction to change it. And I think what I found over time was there was more demand for sort of like an access control layer for these devices. And so basically just allowing people to SSH very easily to devices is actually something that was pretty valuable. And when we ran Y Combinator at the beginning of 2020, we ended up focusing a bit more on that and like changing our landing page to be like the access control layer for Edge and IoT. And then, you know, we were, we were seeing traction with the deployment side. We did have customers, but it just wasn't growing very quickly. And we were trying to figure out how to sustainably get more customers, but it was just kind of growing slowly. Once we pivoted to that idea of like more of like an access control layer, we did start to see like more organic adoption. But at the same time, I didn't think that there was enough value in that. We couldn't charge enough per customer, like the whole market wasn't big enough, but there was a time and it's actually really interesting, but I stopped working on device point in May of 2020. And I think we had about a thousand devices on the platform at the time. And then we had stats and data dog that I was just keeping track of just to keep an eye on things. And then we also didn't want to shut down the platform because people were actually using it to like monitor devices. And I think for three months after that, the number of devices on device point doubled month over month, just from expanding usage and like new people just find out about device point. And so in a phase where we were all pivoting on completely new ideas, I was watching my old idea basically double month over month. And I did have a little bit of hesitation where I was like, maybe I was wrong here. But the story since that about device has actually been kind of cool. So there's another company in the edge computing space called Edgeworks that eventually took over the SAS for device plane. And so they sort of like own the IP for it now. They've maintained the cloud version, which means everybody who is using device plan can still continue to use it. And they've actually incorporated a lot of technology around access control into their own practice too, which is pretty exciting. And then just because it was open source, they're also introducing it into the Eclipse foundation as to be like a first class open source project in Eclipse. So it is actually pretty cool that it still lives on, people still use it. And it's still an exciting idea, even if I'm not actually working on it. Yeah, that is cool. I'm glad. It sounds like kind of a happy ending it. So what is spinning down the company look like? Did that company buy your device plane, technically? No, it was almost a year later after the company shut down. And so the IP at that point was just public domain. And so, yeah, there wasn't any like exchange of money or anything involved. This is purely like, I didn't want to manage the service anymore, but I wanted people that were using it to still be happy. And so I kind of see it as them taking over a project that I didn't want to maintain anymore. And it allowed all the users to continue using it. So there wasn't like any purchase or anything like that. Yeah. And so you have all these users on it. Do you send them an email and say, Hey, I'm going to work on something else? I didn't really say much about it, because the plan was to just sort of leave it running. And it was kind of in limbo, they didn't really exactly know what was going to happen. When Edgeworks took over, we did announce to everybody that there was gonna make a new owner and that they're like taking over the service formally. And like props to their team, too, because we were running device plane in an AWS region, they migrated it to Azure, like fully with zero downtime, which is actually super impressive. I don't think our users even noticed that they took over on the back end. But yeah, no, I think it's a good continuity for the users of device plane. Yeah, that's good. Especially given I can see how the lot of points you made there, like the sales cycles are long, because it's such an important piece of technology that people have in their stacks. Like if it goes down, then their devices are basically unreachable and unmanageable. So I can see how both it's hard to convince people to adopt it. And then like, kind of not a very nice thing to do if you just shut it down while they are using it. It would have been a very bad thing. Yeah. And I remembered one thing that was just very anxiety inducing about working on device plans. Every update that you push, you were always like, okay, what if we actually break devices? Because like, basically, device plane was the bridge between our users and their devices. And if we went down, they'd have no way to access them. And so we had this key metric, which was basically like a heartbeat from all the devices. And every deploy, you would look at the heartbeat. And then if it ever dipped, for any reason, you would know there was a problem. And luckily, we never had any brick devices. But that was something that was always scary with real hardware in the real world that's remote. It's something you have to worry about. Totally. Yeah. And so now at Hightouch, you went from like being a pure technologist to now like, and working on this thing and trying to make it real. And now you're kind of like, leading a company as the CTO. And what does that look like? How have you adjusted to that change? And what are the ways you spend your time? Yeah, the roles definitely changed very quickly in the last year. And so during the summer of last year, it was a lot of coding, doing sales, talking to customers, doing customer discovery. I think once we realized that we had something that was working, the fall of last year, I actually did just a lot of coding myself. It's like all the initial versions of Hightouch, like me and our other founding engineer, Ernest, we built most of that together in the fall, just because we really needed to get up and going and getting satisfied with all these users that wanted to use the product. So there was a lot of just a heads down coding last fall. And then coming into the new year, we ended up hiring a couple additional engineers. And at that point, I think my role sort of turned into a bit like an engineering manager while still coding a lot myself, but reviewing a lot of PRs, working with new engineers, doing a lot of hiring, started to do a lot more hiring. And then at a certain point, I ended up not being able to code anymore. And I was really more focused on just doing reviews of PRs and being sort of like a tech lead, plus again. And then we ended up hiring even more. And at this point, our engineering team's up to 14 people. So it's even bigger now. But at the start of the summer is when my role started transitioning a bit more into just being a product manager and an engineering manager. And I didn't really do as much tech at all. And then once things started picking up even stronger in the summer, my role really turned into being a recruiter full time. And so a lot of what I do now is just purely engineering management, product stuff, and recruiting. So that's what kind of looks like today as I'm spending a lot of time building the team and looking for further engineering management on the team and so on. And it is a lot of recruiting for sure. But I think we're in a phase now where a lot of the founders are just still doing a lot of the day-to-day management. And so we're starting to introduce new executives that are between us and other people at the company. But until they're able to take over a lot of the recruiting functions and take over things, it's pretty hectic. We're running a lot of the day-to-day while spending substantial amounts of time doing recruiting as well. Yeah. So did you write code this week or today? No. When's the last time I wrote code? I think September was the last time that I wrote code. There was one specific change that I wanted to do myself. But before that, the last time was probably in May or June. So you can actually look at my GitHub contributions, my GitHub profile. There's a lot of activity in the fall, and then there's a little bit of activity in the spring, and then it just plummets in the summer. Nice. And do you like that? Yeah, I do. I think I'm pretty flexible. I like the variety. And it's actually been cool to have to play different roles within the last year. Even recruiting, I'm actually really getting into just because it's such an important function. And yeah, I think I like the variety of work. It's nice to have a bit of a break from doing coding, which I've been doing for five, six years now. And so I actually do enjoy the variety of roles. Yeah. I kind of like the... In my role, I end up having to do a lot of different things, and the variety keeps you, I don't know, keeps you engaged. It can be... If I was just in meetings all day, I'd get burned out. But also if I was just in... If I was just programming all day, it's the same thing. So variety is good. When's the last time that you coded, Evan? Oh, boy. It's been a while. I do a lot of grepping still, but I haven't written real code in a while. I think the last time I wrote code was a Cloudflare TV episode where I was getting some home security cameras set up, and I was doing a bunch of RTSP stuff and pulling down images from my surveillance cameras. And so I used to use Cloudflare TV 30-minute blocks to do that. That's cool, though. That's cool that you've adjusted. It sounds like pretty seamlessly, because sometimes you hear about the founders who don't adjust to their company being larger, and they end up staying as a programmer, and it can cause tension and everything. That sounds like you've transitioned really seamlessly into that. Yeah. I know the temptation is there a lot of times just to get into the weeds and solve things yourself and really get back into coding. But most of the time, it doesn't really make sense to do that. I do miss it a little bit. But overall, I like the variety, and so I think it's been fine. Yeah, for sure. Well, I'm curious. You've tried things and failed. You've tried things and been successful. You've been around a lot of other great founders. I know you got to work closely with Ilya and Calvin and all of those folks at Segment. What's your message to people who want to be founders or people thinking about starting things? Yeah, that's a good question. I think definitely patience is a big part of it. I think a lot of people go into starting a company expecting things to go from zero to one and always be positive, but it really is up and down, for sure. I think patience to wait it out, and I think going into it and knowing that it can take a long time. There's a lot of companies out there that have pivoted for years before finding what they actually end up working on. I think even though we had a lot of pivots and we had a bunker journey in the beginning, I think I've seen cases where it's even harder, where companies make it pretty far, and then they're at a series C and they have to pivot. That's, to me, a lot more stressful, where you have a much bigger company and a lot more responsibility and momentum, and having to pivot at that stage. What Slack went through, that must have been very stressful. Having to pivot in the early days where it's a small core team, it's one thing, but having to do that late stage is even harder. There's definitely a lot of ups and downs, for sure. I think when things are down, having the acceptance of it is actually pretty important, because just knowing that things will eventually get better. I guess sometimes they don't for a lot of people, but at least having the patience to try it out. Not everything works out. I do think as much as there is a lot of skill and entrepreneurship in starting companies, at the end of it, you do have to be in the right place at the right time, too, and to find those opportunities. Let's say somebody messages you after this and says, hey, I just started this company. I guess you gave one piece of advice already, and be patient and all of that, but in terms of the most important thing that they could be spending their time on and getting early traction, what kind of message would you give them? Yeah. I guess in terms of more concrete stuff, I think there's two things I would say. One is move fast. Don't wait until next week to do the most critical thing for your business, and get good at identifying what you actually need to move the needle forward, because once you move into a startup, everything's unstructured. You could spend your time doing a lot of building things that don't actually matter for your customers. I think identifying what really moves the needle and making sure you stay very, very focused and having strong week-over-week goals and even day-over-day goals. Accelerate your timelines. Do things very quickly and make sure you're actually moving at a fast pace towards what your goals are. Always be very analytical about how you're spending your own time. I think that's a big one. People come from big companies where they have quarterly OKRs, and quarterly is not the way you think about startups. It should be at least weekly, so I think moving fast is a big one. I think just in general, make sure you think about and be analytical about your own time. It's very easy to get distracted and work on things that don't actually matter. Looking back, if you sum up all the days where you just didn't, even in the case of where things are going well, there's still days where you end up working on things that just don't matter in the long run. I think being just really critical about what you're doing day-to-day is actually very important, and having that self-awareness of not chasing down the wrong things. How often do you ask yourself? Do you ask yourself every morning, what's the most important thing I should be doing? Very, very often. That doesn't change at any point. We try to keep the culture of having week-over-week goals and day-over -day goals versus thinking about things very long-term. Having quick momentum is always very important. Love it. Josh, we have just a few seconds left. I want to thank you so much. That was a very great segment, and I really appreciate all your talking on High Touch. Yeah, this has been great. Thanks for inviting me, Evan. This has been really fun.