How to scale your co-founder relationship alongside your startup — Manu Sharma & Brian Rieger of Labelbox
Episode 69

How to scale your co-founder relationship alongside your startup — Manu Sharma & Brian Rieger of Labelbox

Our guests are Manu Sharma and Brian Rieger, co-founders of Labelbox.In this interview, we take a microscope to their co-founder DNA, exploring the ins and outs of how they’ve made the relationship work over the years.

Play Episode

Our guests are Manu Sharma and Brian Rieger, co-founders of Labelbox.


In this interview, we take a microscope to their co-founder DNA, exploring the ins and outs of how they’ve made the relationship work over the years. We discuss:

Manu and Brian both have extremely valuable advice to other founders, either those in the early stages of looking for a co-founder, or folks who want to add a little magic to an existing co-founding relationship.


You can follow Manu at @manuaero and Brian at @RiegerB on Twitter.


You can email us questions directly at [email protected] or follow us on Twitter @ twitter.com/firstround and twitter.com/brettberson

Manu:

Piece on a particular company or profiling team, or is it sort of a collage of stories from different folks mixed in one single piece.

Jessi:

I think it will be a collage and we'll maybe break it down into different areas and topics. And maybe, we'll have a section on customer interviews and then we'll have advice from a couple of founders in each section. Just because I think one thing we run into with this is that people have such different business models, and people that they're targeting that a lot of times the advice doesn't feel super relevant to you if you're in a different space, right? So just trying to get a big variety in terms of different models, and people were talking to, but focusing on the folks that the partners like Bill thought did a really good job of this in the early days.

Manu:

Got it.

Brian:

Sounds good.

Jessi:

Awesome. Okay, so I obviously know a little bit about y'all already in terms of what you focus on and things like that. But I'd love to just sort of start with the broader timeline of how Labelbox got started. And if you were exploring other sort of ideas and wanted to do a start up generally, or had a really specific idea for this one, and went to go tackle it specifically. Because, we kind of see a mixed bag of some people just wanting to do a start up, and exploring many ideas versus people having a sort of a stroke of inspiration and going after something really particular.

Manu:

Yeah, so Brian and I have been working together since college, we first met in college, about 10 years ago. And this is actually our third foray into business world. The first two was sort of an attempt to start companies in renewable energy and space sector. And so we've been wanting to build companies products, ourselves. We've been the independent thinkers, since during the college time since then. And there has been this constant pursuit of trying to find a way for us to work and build something that is lasting and durable. Yes, that was that. We were for... So one of the interesting things about our story has been that the companies or the projects that we have started or work on, they've emerged from our just general intellectual curiosity about the world. We were very excited about climate, and the renewable energies in 2012. That was a time where that was, in everybody's mind, they were [opening 00:02:57] great companies. And we thought we could make an impact there. So, that was our first foray into that. Then because we are both aerospace engineers, we were also very excited about potentially building aero planes and hardware that would go into space. And so we found a opportunity to do a space business and we built hardware that went to space and all that.

Manu:

And we were also very excited about AI since college times actually, we in fact worked on projects, and published research papers on leveraging neural networks, in our domain of aerospace engineering. But, before Labelbox we were, again exploring all different ideas, starting companies, and they were not necessarily in AI. In many ways we would share or propose ideas that were inspired from our life or our recent experience and events, or opportunities that existed, right. So we would talk every week and just say that, is there a [image 00:04:15] on this one or that one? And So anyway, that was that. But in that format, I proposed an idea from my experience of actually seeing AI systems being built in Silicon Valley, from the companies I've worked at DroneDeploy and [PlanetLabs 00:04:39]. And there was a lot of insights there. Just purely seeing how people were building visual computer vision systems. And I shared it with Brian and Dan, and they actually very quickly validated that that was also a thing that was happening in their companies or teams. And that's how it began.

Jessi:

Cool, that's awesome. So how did the idea for Labelbox feel different than the two other ideas you mentioned? Why did you decide to go with this one? Was there just more excitement around it? Or you felt there'd be more pull around it? Or I guess how far did you get down the path of exploring those other ideas?

Manu:

Well, the first two things that there were actually companies, we built products, and we were working on them full time. And in both cases, we had access to capital, not in a form of venture capital, but through grants and customers and so forth. But for Labelbox we... Why we chose particularly this one, number one was purely first principles thinking we had experienced with neural networks, a decade ago. And the things that made it to not be useful in the real world, the constraints were no longer true. In 2018, there was ubiquitous compute, and that there was far superior neural networks that could learn to emit human behavior, human judgment, pattern matching. And the thing that was remaining was a very organized workflow. And, and we realize, that's a really interesting problem to solve, and potentially a very important company to build in the world of AI systems.

Jessi:

Mm-hmm (affirmative).

Brian:

So, I was working on a different product at the time, when we started Labelbox, as well as a full time job. And, that was solving a different problem and payment space. And we had a very serious conversation as founders. In the very beginning of Labelbox, because I had this app which was... Manu was helping me to some extent, but was something that I was building, essentially. And it was going fine, we had customers, and we had processed half a million dollars in payments. So, it was working. And I think we had a conversation about what do we do here? Right. How do we decide what idea has the most [inaudible 00:07:26] and things like that? What we learned in our prior two ventures was, you really want to look for a few things.

Brian:

And we learned this the hard way. Which is number one, what's the market like? How big is it? What's the opportunity? We built an excellent product in the space business, but the customer base was pretty small, right. So we basically saturate our customer base. So we learned really, that lesson of, even if you build something great, you can be proud of that. We are proud of that. And we have that framed, and we're very proud of that work and still holding on to the artifacts of that, and the impact we made. But, to build an enduring company, do you need to have a certain market size? And so we were interested in that. The thing I was working on... when you really backed out the calculations, this was going to be more of a lifestyle business, right. And that was really what turned the corner for us was, if we're gonna get together for [deals 00:08:17] in this living room and, and do something together, we should do something that can have a big impact, and build an enduring business.

Brian:

And while we had this product that I was building that was fine. The building an enduring business, from a first principle's perspective was much more compelling with Labelbox. And similarly with other ideas that we've explored, right. And the second thing you look for, really, from our perspective, as tool builders is, when you build a wrench, it needs to be applicable in its form to a problem that cuts across a large user base. So we love building tools and products. And so what we really look for is, can we build a tool that can generally solve a problem for a lot of people?

Jessi:

Mm-hmm (affirmative).

Brian:

We really focus on building that tool in its form. And we're not as excited about having a company where we're doing one offs for every customer.

Jessi:

Yeah.

Brian:

We love the practice of building the wrench, and the wrench works for the mechanic and the bridge engineer and watchmaker. We want to make that really nice and compelling. And so that's our bias, and so with Labelbox, it had those two elements to it.

Jessi:

Okay, great. That's a really helpful overview. So it sounds like you had validated through your own experiences that this was a tool that teams needed. But how did you take that next step of finding customers to talk to, and hone in on exactly what the problem is and where to focus your early product building?

Manu:

Yeah, first of all, We had validated the applicable general problem. And I had personal experiences of seeing that happen. Or the problem, particularly in my companies, and how teams were building or investing a lot of capital to build these tools. But what we did first was Dan, who was a CTO, he and I had a lot of friends in companies that were doing computer vision. Purely being in the industry, and we just started to have conversations with them about... Hey, if you were to go [pay 00:10:48] with a the tool like this, would it be valuable? And would you buy something like that? Or is this even a problem for you? And so through that, we got to some maybe handful of companies and within our network, through friends, we basically got to about 10 to 15 different companies that were building AI systems.

Manu:

And what we did was we were kind of doing this quick iterations of our ideas and potential problems, as you were going through those 15 or so companies. The first few were very... In ideas, and then the second batch of people, we had a little bit more concrete, and we were assigned to form a value proposition. And then [inaudible 00:11:37] this is exactly what the product is going to go do, and so forth. And by the third or fourth iteration, we had a very good sense of what we could go build. And we were trying to just subscribe potential customers who would initially pay us to even go build that. This was pretty unique in this experience, because we were very focused on generating revenue from day one.

Jessi:

Mm-hmm (affirmative), yeah.

Brian:

I think[inaudible 00:12:07] on the questions, which we can get into more as a pillar of our product development process today. It's one thing to ask somebody, if I build this, are you interested? Most people will say yes, because everybody wants cool new stuff. But what we ask and ask them was, what are you doing today? Right. Really the hallmark of a pain is that they're doing that because it's essential. That's what you're really looking for. This is an essential thing. And, tell me about how you're doing it today. Tell me about that. Oh, well, we have this consulting firm, we have this labeling service provider. Okay, tell me more. Yeah, and we have a full time persons running the operation and managing the service. And we send zip files and Google Drive links back and forth.

Brian:

And it’s a biggest pain in the ass Brian, and we don't want to build a tool. But we have to build a tool. And I have two put two of my engineers on it. So you really want to get to that. And we did that. I thought Dan did a great job of asking us questions, Okay, I get it that you're you're interested. But what are you doing today? Is this something that you would like to do? Is it nice to have? Are you really solving this problem? And tell me about how you're doing that? We call that the workaround today? But, I think that's a very essential part of this process.

Jessi:

Yeah, What are you firing and the job to be done to kind of framework.

Brian:

Mm-hmm (affirmative).

Jessi:

Okay, cool. Was there anything that you uncovered during this process that really changed your understanding of the problem, or set you on a totally different course from maybe what you had been envisioning as founders originally? (silence)

Brian:

We found two things and then we can maybe talk about others that we built into the product right away, that were not, I don't think obvious to us in the beginning. But the first was that the labeling process was operational in nature to some extent. Meaning that the problem that people had was partly operational, how do I give all these tasks out to people? Labeling tool is one thing, but I also need to distribute the work and I need to manage that work process like a factory and so that we built a queuing system into the product, which is still a mass differentiator for us.

Brian:

That came out with original Labelbox and was incredible. And then the second thing was that we... this is going to get a little technical, but we structured the labeling process through this ontology system. Where we said that, because all the tools on the market, allow the labeling team to basically write anything and put any label on an image. And we found that that wasn't how people wanted it. That wasn't how decision makers and buyers and data scientists needed it. They needed structured data, they needed everything to be consistent. And so the other thing we built into the product from day one is that when you set up labeling work and [train 00:14:58] your development, the task is very rigid. Such that the data on the other side is as clean as possible, right. You have constrained what can come out of this thing. And that creates better training data. And that was one of the other things he picked up on that drove a lot of adoption.

Jessi:

Okay, great. Okay. So I did see that from what Bill was talking to me about that you guys had early traction pretty early on, compared to a lot of companies who come in to raise at the seed stage. So in your mind where you bifurcating between validating the idea and seeing that it has legs, versus building for what those initial customers wanted? How did you juggle the two, because some other folks I've talked to first are spent a long time digging in and making sure there's and error there, and there's a problem there. But it sounds like you guys, as you said, took a sort of different approach of working with people right away, and building a product that way.

Manu:

Yeah, I think that's purely the spirit of the founding team. In many ways, it's a core culture of Labelbox, which is we were very action oriented, and just solving problems. And we wanted to iterate with the tool, rather than with ideas. Once we had gained conviction, this is all very fascinating, fun ideas, we should just go do it. And, and we were also very... We just rolled up our sleeves. And we would just do these things days and nights. And it was a lot of fun being in that environment, and so forth. So I guess it was just the nature of how we like to solve problems-

Brian:

I think you're underselling a little bit, so. Manu really had this very special chemistry, incredible understanding problems and driving through intuition, what a solution could be. And Dan was incredibly fast at building. And so they got into this rhythm of learning and building and learning and building and we separated out our roles. And quickly my role became kind of frontline talk to users. And really, my job in its best form early on was to synthesize the information coming in from our users. And in the support realm, and operation part of the business and try to explain that in problem terms, systemic problem terms to Manu and Dan. Here is the... If we boil everything down, and we get rid of all of the [cruft 00:17:54] of what people are saying, what is the thing people are trying to get done? And what are the systemic pillars of it? Driving that onto Manu and Dan's iteration loop. And that's how we worked for a long time, probably four or five months ago.

Jessi:

Okay.

Manu:

Yeah. The launch was in pursuit of, obviously, the validation. So what we did was, after all of those 15 or so customer situations, I had built the entire design and model of the application. Sometimes I would show that we in fact, even had an instance where an AI company wanted to recruit Dan and me, as a contractor to go build their internal tool that was exactly the product we wanted to go build. And so we kind of have these examples where we had enough confidence that we should just go build this out.

Manu:

And on January of 2018, we launched a very basic tool on Reddit. And, that's really where the snowball effect began, because we started to have people having conversations and they've signed up on Labelbox. They said, "This is great, I want x and y and z." And what we were just doing was, it's kind of like software, it's kind of like building a castle and we start with a room and as people come into the room, they want more doors or other the rooms. And so we started to kind of feel that a market pull. And we [basically 00:19:41], days and nights, we just continued to make more rooms, more doors and all those things organically based on users and their needs. And within a span of three months, we had that traction, that Bill mentioned. The reason to raise capital was purely To keep the growth and to keep ensuring that we can provide a great experience, that honestly we thought we should be doing. And there was no other way for us to do that.

Jessi:

That makes sense.

Brian:

And I think validating ideas is tricky. Because, if you have a user, a customer, whatever, and they're saying, I want a button here that does x. What you want to do is figure out, what's the underlying purpose of that? But what is the synthesized version, or the boiled down version of that stated in a problem, stated in an outcome. And then, in your next conversation, boil that down in a way that you can ask about that problem to someone else. Do that over and over and over again, if customer A says "I want a button to sort the list." You can't go to customer B and say "Do you want a button to sort the list?" You have to boil that down to? Are you trying to figure out which training data to care about? Is that a problem for you? "Oh, yeah, that is a problem." Okay, great. So I've actually validated two people having the same issue by translating from a feature request into a problem statement. And we did a lot of that.

Jessi:

Yeah, that's super helpful. How in terms of something also about your sort of model is that you're in different verticals, right. There's a lot of different kinds of customers you could have, in terms of their industries. So did you have focus in the early days on a certain industry or you got building pretty broad from the start?

Manu:

We we're building broad from the start. And that was purely from our desire or conviction that we wanted to build tools. And that tools should be broadly applicable for building all kinds of AI systems. And a lot of these beliefs were driven by our first principle thinking, and that, in order to this idea to work at a grand scale, it cannot be just in a single vertical. And the other thing was that, my experience was, I was one of the first engineers at DroneDeploy, and DroneDeploy software was a broadly applicable technology for people who would use drones for commercial activity. And the power was that when you make broad tools, you get all these amazing use cases, and you have a sample into nearly every industry as possible. And in a newer emerging technologies, betting on a single industry, can be highly risky. And so those were some of our experiences and belief systems, going into building it the way we wanted to go with, for sure-

Brian:

We saw it at the same time, as functional, we saw that problem as, like GitHub would see it. You're writing software for Mars rover or a radiology device, you need to follow a [probe 00:23:10]... You have a functional need around that. And there's a solution for it, that can be agnostic to the industry application.

Jessi:

Mm-hmm (affirmative). So in terms of the stakeholders involved, how did you think about the buyer, versus the end user, versus other players you might have to consider, versus just sort of the traditional user that you might think about?

Manu:

I think we continue to forget... I think we are more mature thinking about it now. But in the early days, it was AI teams or machine learning engineers. And that the companies or the teams that had the biggest problem, in those places, machine learning engineers had enough say, in that early phase of the market, to go buy a tool for a few hundred dollars, or a few thousand dollars, and so forth. So those were our customers initially, and now it has become our tool has become a platform. And then they are rooms for different cohorts of people within an organization, product managers, engineers, operators, then there are service providers. So over time, we've come to realize different folds. But I think in earliest days, it was very acutely focused to machine learning engineers or teams person.

Jessi:

Do you think, do you think looking back, you wish you would have considered other kinds of stakeholders in [that 00:24:49] early process, or you think it was fine to kind of have that laser focus in the beginning?

Manu:

I think it was totally fine to have that laser focus. I think if you were to do more of that exploration. My suspicion is, it might cause an analysis paralysis kind of thing. Yeah, I think there's something to be said by having a strong conviction and confidence and using that to inform actions. Quick actions.

Brian:

Part of the problem with the market, especially at that stage was that it wasn't even really clear, in the organizations were selling to, who was really responsible for solving the problem, they [inaudible 00:25:34] solved. It's more clear today. But this was a two way situation, right? We were having conversations with different titles solving the same problem. There's PM's that were doing data labeling operations, and there's machine learning engineers doing that. And so, there was a lot of noise in that conversation. There were a variety of different roles. But, we stayed true to the problem. We knew that this problem needed to get solved in an organization. And so we stay true to that we let the persona fall out of the [work 00:26:09]. It's not a mature market, still isn't, So.

Jessi:

Yeah. Were there any questions or tactics that were helpful for uncovering who the right person... Because, I imagine you probably had some conversations were you're like, "Oh, am I talking to the right person?" Or, things like that? Was there anything that helped you narrow in on that when you were seeing all these different titles and things like that? Or was it just such a case by case basis based on how that company was set up?

Manu:

First, it was case by case basis. We were looking for two things, basically, somebody who actually had a real problem in their company about training, labeling and clean[inaudible 00:26:47] data, who could obviously use that tool and provide a lot of insights. And secondly, we were looking for somebody who's going to write a cheque for our tool. And in many cases, there used to be one, and in some cases, there used to be different people. But those are the two... this one has been even in case by case basis, those are the two things two [buckets 00:27:11].

Jessi:

Yeah. Okay, cool. So in terms of the market more broadly, I don't think you guys face a lot of regulatory issues or anything like that? I don't imagine that you do, but wanted to check on that.

Manu:

We do in terms of security and privacy, because it's a company's most proprietary [chaining 00:27:32] data. Many companies are reluctant initially to get into the cloud, or to upload, share their PIA data into Labelbox [inaudible 00:27:44] label that. Yeah, that was a challenge for us. The way we address that was, we basically had a solution where the data then had to transfer to us, or to Labelbox servers. Basically, we found all these creative ways to go ensure that the data won't come to us, so we would never see it. And but (silence) we knew that also, this is going to be a constant for our business. So we proactively went out to seek software compliance certification. So which we are now certified, in two and a half years of history of a company. We have an on premise deployment solution, which was a [product 00:28:28] we did early on as well. But those were taken after our first experiences. Okay, people really care about security and privacy and that we need to mitigate that to[inaudible 00:28:41] become enterprise company.

Brian:

Yeah, I think ultimately, we met the customer where they were. We even have HIPAA compliant customers today that they run Labelbox, and we're not HIPAA compliant, in our system, but they run HIPAA compliant workflows with Labelbox. Because, we were willing to meet the customer where they were we. We thought that it was important to win the customer, was important to meet them where they were, and it was important to keep momentum. So sure, on-prem is a pain in the ass, it still is a pain in the ass or VP [manager 00:29:12] wants it gone as soon as possible. But, some of our customers want on-prem. So you really need to meet them where they are. There are other companies in our space that took a very opinionated or sort of rigid approach and said, "We're cloud only, we refuse, it's the world of cloud," Or We must store your data. We met our customers and said, "Oh, you want to store it on Amazon, and that's really important to you." We will come up with a solution, so you can do that.

Brian:

And it's given us a lot of tailwind, even today. It's been extraordinary for us. And it's in those times we just thought outside the box. Dan was incredible and coming up with solutions and helping us meet the customer right where they were so they could solve the problem, and make and we can make it as easy as possible to buy because we're a small company. And they don't trust us just because, we're 14 people they met half the team on the sales call, they just don't trust us. So we have to meet them where they were.

Jessi:

Yeah, that makes sense. In terms of, you mentioned competitors. How did you sort of survey the competitive landscape? What in those early days when you were starting, did you just, do a bunch of research on them? Did you try out the other tools? What was the competitive landscape like at the time? Because, you mentioned it was new and emerging, but it sounds like there were some solutions available?

Manu:

Yeah, they were only few, maybe two or three companies that we're tackling this space, [figurative 00:30:40] scale. But most of the solutions were in-house or open source tools, or that kind of thing. And we tried all those things, and it wasn't the experience, or it wasn't really solving the problems that we had been exposed to or seen in the conversations, or experience in the conversations. And that's it. Yeah, that was basically it, because it was [an 00:31:16] emerging, there was not any real mature solutions for companies. So there was this ample amount of opportunities ahead. So we weren't really concerned about it.

Jessi:

Mm-hmm (affirmative), yeah. Were there any other market dynamics that you really had your eye on at the time in terms of tailwinds, or headwinds, or things you were really concerned about working against you? In terms of other aspects of the market?

Brian:

Yeah, I mean, one of the things that made it harder to raise a seed around, luckily, we met Bill and others that agreed with us on this, was most people thought, and the venture capitalists thought that data labeling was a more of a discrete turn on turn off sort of thing. Meaning that people were worried that, oh they'd by Labelbox for three months, use it, label a bunch of data and then go back into their cave and, and work on it. We had a different opinion, based on the Manu's experience, and my experience, and Dan's experience. That was probably the thing we were the most concerned about, because there's this overwhelming number of venture capitalists, which we really respected, saying, "This is really more of a discrete, you're not going to be able to build, customers that renew and buy your software over years of time that that doesn't make any sense, Brian." So we worried a lot about that.

Brian:

But it turns out, it's totally the opposite. People are doing it like they do software development, just because Facebook works today doesn't mean they don't have 5000 software developers in there. There's always something to do, there's always innovation to be done. And machine learning is a technology paradigm for solving problems. And I think that now is very true and obvious. At the time, it was something we were concerned about, thinking about, worried about.

Jessi:

Mm-hmm (affirmative). Was there anything you felt at the time, you could have done to get more data or get more information or validate that further? Or was just something that had to be sort of born out of experience of a couple years and seeing it play out?

Brian:

I think the thing that really helped was having people like Bill, talk to our customers and helping them understand how these companies are actually buying, and why they were buying Labelbox, and how they were using it. I think that really turned the corner for us. We had champions in some of our in early customers that were willing to get on the phone, people like Bill and tell them, "No Bill, this is not a service we've installed this thing in our workflow, we're not going anywhere. Labelbox is part of our production workflow." And that really opened the eyes to people, that opened our eyes to it made us feel a lot better about it.

Jessi:

Yeah. Okay, great. And then sort of the last thing on the business model, go-to-market side. Are you guys still sort of self serve freemium? I know that was sort of how it initially started, but is that still the case now?

Manu:

Yeah. What I would say above 50% of our business is organic inbound, and 50% nowadays, direct.

Brian:

But I mean, we know people don't buy with a credit card today. So we do have a free tier and it's super powerful for us. But our first paid tier is expensive and so the buying motion kind of the go-to-market sales motion is very much looking like an enterprise, the first year starts in the mid 20,000 range. Then it goes up from there into the six figures. So if you look at the contract side of the business, it looks very much like a classical SAS Enterprise business. But the free tier is very powerful for us. And we do provide free academic licenses to students, professors, things like that. So we do have this very liberal free tier, and we do support the community, Labelbox is used a bunch of curriculums now in science machine learning at Berkeley and MIT support. So we do have that part of our business, but it's still expensive to buy. I think in the future, we may have this more self serve motion, but that-

Jessi:

Okay.

Brian:

That isn't what we do today, mechanically.

Jessi:

Got it.

Manu:

We never had self serve, actually, we never had credit card payments [inaudible 00:35:49]

Jessi:

Got it, Okay.

Brian:

Yeah.

Manu:

But, yes we have a free tier. Anybody can go try our tool.

Brian:

Which makes us unique. I think what it does is it focuses you to build a great product. And that's so important, right? Because, we're almost the only one with a free tier, there's been more and more lately, but that is so powerful, when someone can log in and use a tool and be, "Okay, yeah, I can use this, I can recommend this, I've experienced with this, I can look at their docs, and we can do something here."

Jessi:

Well, and also helpful for you, on the product side of what to build and what the problems are, and things like that, just learning a lot more through discovery that way.

Brian:

Mm-hmm (affirmative).

Jessi:

Awesome. So the last section, I wanted to just pick your brain on what advice you might share with someone who's looking to start a company, and they have an idea, but they want to really go deep on it and really prove it out as much as they can, as robustly as they can probably before fundraising. Is there any advice you would share or anything that you think you guys did really well or didn't do as well, but you think would be helpful to share with them?

Manu:

I think first principle thinking journey helps people to get to the deepest levels. Why is that this idea or this particular solution is going to work or will be fundamentally needed. Apart from people's perception of... It kind of teases out the world and the reaction of the world to a solution to a functional problem.

Jessi:

Mm-hmm (affirmative).

Manu:

If you look back in hindsight, that's what it feels like, when we were obviously building. We were not very cognizant of, that is the tool we use to solve problems. But in hindsight that's what it is. I think that's number one. Brian, anything else?

Brian:

I know this isn't always applicable, but for fellow tool builders out there, [producty 00:37:52] people. In all the things we've done, and built the common thread of things that have been used by people and bout that we've made. Especially early on, we've always been able to find something that a problem unsolved, right. So even if you're going into I mean... I can't even imagine what it would be like to start a company competing with Lever or HubSpot or Salesforce. What on earth could you possibly do in those spaces? But I'm sure there are things. And Labelbox had a few things I mentioned before, that were problems that were unsolved, that people are willing to pay for it, people are willing to pay a lot, I mean relatively a lot, let's say $12,000 a year. For something you make that solves a really painful problem in its isolation.

Brian:

And we started there we said okay, labeling sure people can do that. But what are these things that just really pain people in their work? And let's get those things right, let's solve those problems. And people will pay us for that, just because if it's painful enough, people will buy a point solution to do it. And I think that's really important. You don't want to be a [me too 00:39:07] to kind of copy surface level, necessarily. That's not how we approach it. We approach it by saying, sure, there's these things that definitely need to get done. But what are the things really under there that are really painful?

Jessi:

Mm-hmm (affirmative).

Brian:

People will pay good money for that. So you've always got to find your kind of your product market fit, and you got to find something or a few things that that people really want solved. And they're willing to bend over backwards to get it solved, right. They're willing to do business with a 12 person company, they're willing to pay five figures for it, they're willing to deal with a vendor that's super unreputable. The money is usually not the biggest problem when you're six people. It's usually, "Who the heck are you guys"

Jessi:

Yeah.

Brian:

"How do I know you're not going to be gone in six months." You have to overcome all of that. And the way to do that is to solve problem that people are really excited to solve, they think is going to have a big impact on their business. So that would be my thoughts and getting the product out early, and listening to your users, being very open to what they're saying, and what they need.

Manu:

I think the other way to... When we got deeper, [was 00:40:31] certainly we had a lot of depth of knowledge, before we got started. But in order for us to get to a position of pursuing even more deeper understanding of the problem. We just had to go build a tool that was enough value proposition where we could be in a room with a more customers, or users who saw initial value, but then they wanted to share more about their problem. So yeah, I think maybe that there's something there, that you've got to create enough initial value to get in the position where you can go more deeper. And I did it[inaudible 00:41:18] users and for customers.

Jessi:

Mm-hmm (affirmative). Okay, great. We have a couple minutes left. So anything else that we haven't chatted about today that you think would be helpful to flag in this piece or something else? You think you guys did a great job of that we haven't gotten to yet?

Manu:

I think they're some of the questions you had in there. We had a podcast interview that goes over, I think it's 30 minutes or something. And basically goes over a lot of that early formation of Labelbox.

Jessi:

Okay.

Manu:

So if you want to just... I will just reply with the link, and you can hear the more details. And there are many answers to the questions there as well.

Jessi:

Okay, Awesome.

Brian:

Yeah, things we could do differently, I was just looking at this. One of the things we didn't do well was we didn't hire sales engineers early enough. And we didn't really understand sales engineering that well as I did. And even though I was a sales engineer at Boeing [inaudible 00:42:19]. But, myself and Manu and this other guy, Cyrus, were doing sales engineering. We didn't really realize it, because we're engineers, and we just got the phone, and we're, chatting up API calls. And, we not even realize that we're, solving these intricate technical problems, and performing a sales engineering function. I would hire that role a year earlier, at least, I think it's a very important role. I think when you hire your first salesperson, you probably want to put a sales engineer in there pretty quickly. So that's one thing that I would do differently for sure.

Jessi:

Okay, Awesome. All right, great. Well, thank you so much. I'll take a listen to that podcast and pull out any other extra [inaudible 00:43:03] for that. And then if I have any other questions, I'll just ping you guys. But I'm doing a couple more interviews this week. And next and then my plan is just kind of start writing and hopefully have something to show y'all around November ish, the beginning.

Brian:

Thank you.

Manu:

Thanks Jessi.

Jessi:

Awesome. Thanks so much for the time today, really appreciate it.

Manu:

Yep, bye.

Brian:

See you.

Jessi:

All right, see you.