The following interview is a conversation we had with Anand Kulkarni, CEO of Crowdbotics, on our podcast Category Visionaries. You can view the full episode here: $58 Million Raised to Power the Future of Rapid Application Development
Anand Kulkarni
Hey, my pleasure, Brett. Great to be here.
Brett
Yeah, I’m super excited for our conversation. Let’s go ahead and kick off with just a quick summary of who you are and a bit more about your background.
Anand Kulkarni
Yeah, sure. So my name is Anand Kulkarni, CEO and Founder of Crowdbotics. I am a serial entrepreneur. This is my second time running a venture backed technology company. Crowdbotics is an AI powered software development platform, been around for seven years. We help enterprises and founders build software products using artificial intelligence and a universe of reusable components, which we call modules. And we’re part of a broader movement called code operations. I’m actually technical, despite being the CEO here. I got my start as a research scientist over at UC Berkeley ages ago. Was an NSF fellow, worked on a lot of computer science stuff, put out a decent amount of peer reviewed research in the field that is now called human computation, which is a subfield of artificial intelligence. That led to some exciting developments and some cool stuff.
Anand Kulkarni
So I’ve been excited to lead the team here at Crowdbotics as we’ve grown in the last two years.
Brett
Tell us about your first company, lead genius.
Anand Kulkarni
Yeah, lead genius was a spin out at UC Berkeley. We went through Y combinator, started out as a company called Mobileworks, which was competing with Amazon Mechanical Turk. It was a little early, but at the time, space that is now dominated by companies like scale AI, thought we’d do data annotation at scale. It turned out the niche that became most exciting for our customers and for us was sales and marketing. Data annotation, and lead genius is probably still the best product in the market around for that, still going strong.
Brett
Was it a hard decision to make to leave lead genius and go out and start Crowdbotics?
Anand Kulkarni
Oh, well, I think anytime you jump onto the next thing, it’s always a little bit of a tough decision, but I think that for anyone who’s been there before, you recognize that there’s always the urge to create and launch something new inside you. So I think that’s something that anyone as a Founder can appreciate, given.
Brett
That you’ve been in AI for so long. And I think it’s fair to say you were doing AI before. It’s cool. What do you make of the market today, and how do you summarize the state of AI?
Anand Kulkarni
Look, we’ve been an AI powered company almost from our inception. We were one of the first commercial users of chat, GPT, excuse me, of GPT-3, on their API, and we actually started the company Crowdbotics based on what I was seeing happening in the early research space around large language models. These were systems that were things like recurrent neural networks, predecessors of today’s llms, that just look extraordinarily promising, and they look like things that could make code or things that looked like they could make things that were almost code. And when I saw this in 2017, 2016, I said, look, this is probably the future, and we got to go out and get ahead of this. So, look, the current landscape of enthusiasm around AI is phenomenals that have a vindication of our vision.
Anand Kulkarni
And I think that for companies who are already in this space, of course, we’ve experienced massive tailwinds from the market, driving customers straight into our hands and demanding to use our product. And it’s funny, because were working on the same technology in the market two years ago, and people just didn’t believe us. We showed them technology showing that you could use artificial intelligence to plan software, use AI to write code. People said, yeah, well, okay, but can you really? And today, I think the current landscape is really awarded folks who are already in the space with good solutions. There’s credibility, there’s enthusiasm, and of course, market appetite to buy solutions like this has gone up to a ten, which is great to see.
Brett
So, obviously, November of last year was the watershed moment, I would say, for AI, at least just in the wider consumer world, where everyone started to talk about it and understand it. What about that time period before, so that five years before you’re evangelizing this idea, you’re trying to push this out into the market? What were those conversations like? And were there ever any very low points that you experienced where you just thought maybe the market wasn’t going to be ready, and maybe you were just too early to have a viable business here?
Anand Kulkarni
Okay, well, this is a great question. So I think that as a Founder and as a entrepreneur, you always have to balance the strength of your ultimate vision for the company with the reality that you’ve got to make sure you’ve got a product in the market that people want and that people will believe independently of anything else in terms of the popular conception of tech. So for us, were fortunate in that many of the first things that were able to produce and sell to customers were part of our bigger vision. Didn’t require customers to really believe or understand the bigger picture of artificial intelligence. It just needed them to think about and understand that they needed to have their own software built, and to believe that there was a better way possible to build software.
Anand Kulkarni
So if you look at how Crowdbotics works and the model of code operations in general, we’re a little bit different from how most people think about AI based software development. While we’re compatible with tools that do character by character code generation, we actually take an approach that is modular. So we are snapping together, recommending, selecting big building blocks of code into functional software. We help our customers do that at scale, using their own code, and we do this also with our commercial libraries of code. That was an easier narrative for a lot of customers to understand, because that’s the way that most people build software by hand today. Go out and find libraries that you think you’ll need, you put those together and you have running software.
Anand Kulkarni
We didn’t need to have people buy that bigger vision, as long as they were willing to understand at a different level how the software was able to work. Of course, now that we are seeing popular validation of this concept, of course things have come together in a nice way. Now, the other side of your question was around, were there tough moments? No question. I don’t think any entrepreneur or founding team can say that they have avoided those kinds of tough moments, especially when market sentiment may be going up and down all the time. And I think that we saw a number of moments like that.
Anand Kulkarni
Places where we saw tough customer types for flavors, difficult use cases, people pushing the envelope for the boundaries on things that we couldn’t do, and having to take difficult decisions on whether we wanted to take a bet on a space and persist in spite of challenges, or if we wanted to pivot away from it. And some of those investments really paid off the decision to persevere. I can give you an example. Early on in our process, we had the opportunity to work with the US government. This was probably three and a half years into the company, and our first efforts to get anywhere through the US government’s procurement process proved to be extraordinarily difficult. We just got rejected over paperwork issues, over not knowing how to sell or engage with buyers inside the US government. It was extraordinarily demoralizing.
Anand Kulkarni
We said, hey, we should probably just pull out of working with federal sales entirely. Of course, today the US government is one of our biggest customers. So it turned out to be a good decision to persevere. And of course, they’ve been now some of the biggest proponents of artificial intelligence. And that’s been a rewarding shift.
Brett
To see, what advice would you have for other tech founders who want to work with the US government and do business with the US government?
Anand Kulkarni
It’s a great question. So there used to be a conventional mindset in Silicon Valley that government was too hard to work with, too hard to sell into. I think that mindset has now started to shift. I think there’s been a surge of interest in and around government tech. I think there’s been a number of changemakers who have shown up inside positions in the federal government that say, we really need to adopt us grown technology and US grow innovation fast to stay competitive and efficient. And I think those change agents, those innovators, those folks who are looking to stay competitive, those are the ones to find and engage with. And I think that piece has been really powerful for us to watch as a cultural shift in how the US government approaches technology. So there’s been lots more attention paid now to innovation dollars.
Anand Kulkarni
There’s been lots more attention now paid to engaging in the private sector and with the private sector. I think that one thing that entrepreneurs should be aware of is that when working with the US government, there’s always compliance factors to think about. Security, safety. Those are obviously places where platforms like Crowdbotics do make a difference. But it’s also places where there is a level of rigor that you need to go through to be able to work effectively within the US government ecosystem. But once you’ve done that, things get a lot simpler. Of course, finding the people who really care to drive change and positive outcomes, those are the champions that entrepreneurs really want to empower and enable.
Brett
Trey, did you lose any employees or did you have any employee pushback when you were considering working with the government? And I guess that probably depends what branch of the government you’re working in. But I. I think it was Google or Microsoft. I remember there were a lot of news articles a couple of years ago about people protesting that they were. They were doing work, the government that they didn’t want to support the government in that way or have their technology be used in that way. Did you experience anything like that?
Anand Kulkarni
No, we’re a us company, so I don’t think that anyone was too concerned about that. I think the concerns that are typically faced inside startups are about whether you can actually survive the length of a buying cycle with a customer that can be as slow as the US government. And that’s really the bigger concern that I heard from our team and from our investors. Of course, the landscape has shifted. I think these days there’s been a lot of good successes from companies who work with the US government at scale. So that’s been positive. To see.
Brett
That must have been a big trust and credibility boost, I’m guessing as well, right? For an enterprise company that’s considering using you, if they find out that you are working with the US government, that’s a pretty high level of validation that you’re able to deliver at a large scale.
Anand Kulkarni
It certainly did not hurt. So Crowdbotics, as a platform for software creation, actually has seen its biggest uptake among regulated customers. So think government, finance, health, as well as non technical ones who are, by virtue of changes happening in their industries, being forced to become software companies, where the importance of their software, the mission criticality of what they’re building, requires them to act in a way that looks a little bit like a regulated industry. So we’ve been excited to see strong endorsements from lots of customer types. The US government is one, of course, that meeting their security standards means that we can work in a wide variety of contexts that are a little bit less meticulous.
Anand Kulkarni
But I would also say that if you’re a large healthcare chain, if you’re a large financial institution, your needs are more or less comparable in terms of the rigor of your process for software creation and the quality of the outputs that you need to see happen from day one.
Brett
Did you know that this was going to be a big focus, going after regulated industries, or did you uncover that later on in the journey?
Anand Kulkarni
Gosh, that’s a great question. I had no idea. I would never have predicted that. Let me tell you why we ended up going into regulated spaces. It was really because customers pulled us in. So Crowdbotics has an approach for software creation that, as I mentioned, it’s modular. What that means is that if you, as a customer showing up to build software, you are describing what you’re trying to create in natural language. We are identifying, based on historical similar products built by your organization or by your enterprise standards, the different components of code, the modules that we can make use of, or that you can make use of to put that application together. Each of those modules has been verified whitelisted by somebody inside your organization, and it’s a full component of code, meaning it’s living inside your code repository.
Anand Kulkarni
It’s something that your developers can go in and edit, and it’s written in an ordinary programming language, not something proprietary or esoteric. So think JavaScript, Python.net comma Java, something that you actually want to work with. So what that means is that you’re building these systems that have really structured elements of code that have been whitelisted individually, which is a very different way than systems like low code have built software, because first, you can actually see the entire set of code that you’re working with. Second, each of your components has been whitelisted by somebody in your organization at some point in time. And third, you actually have this governing set of standards that you have applied using artificial intelligence across your code base so that you have some ability to scrutinize what has been done.
Anand Kulkarni
You can also go and look at the code, audit it, modify it, change it, which if you’re a CIO or a security officer, matters quite a bit in terms of understanding what your software is actually doing and if it is actually going to be secure or insecure. So that is primarily why we have seen such fast adoption inside regulated spaces. It’s because the approach just works better if you are building software for high security purposes. But we’ve also seen lots of customer types, lots and lots of customer types who are not regulated, but who want those same benefits, owning your own ip, being able to edit and modify your code, or just the idea that you can get to market faster using pre built components.
Brett
What are your views when it comes to your market category? Is this a category creation play? Is it selling into an existing category in a different and better way? What are your views there?
Anand Kulkarni
Yeah, great question. So when I started the company, I actually had a specific set of reference points that I was looking at that weren’t just artificial intelligence, because back in 2016, AI was not viewed as something you were going to build a giant category around. Really what I was interested in is the related question of how we could do better than low code. Low code is this category that’s been around for 15 years now. This idea that you could write less code, empower people who are not developers to go and write the code and make the developer obsolete.
Anand Kulkarni
I think that a lot of people, software engineers, people in the Valley enterprise, CEO’s, CIO’s, we just found this notion a little bit off putting from the get go that you would actually eliminate developers using low code because it’s just really hard when you’re building systems using these drag and drop tools that hide the code or hide the complexity of code to say that we can build something that’s going to be scalable, important mission critical. As developers, we always know when you’re building on systems like that, you hit a limit. You have to go and switch into a software system, a proper programming framework, and go modify something, and then you’re back to building a real open source framework, the way that you normally like to build.
Anand Kulkarni
So really we thought about what we could do that was better than low code because look, nothing really important in terms of the software that we all use and drive every day, none of that’s been built on low code. It’s all been built in full code. Crowdbotics takes this approach that’s really opening up a new category, code ops, which is the idea that we can use artificial intelligence and historical requirements to change the way that we are building software. Codeops is a little bit like a superset of DevOps or dev sec ops, which were motions that thought about how to empower the developer to be more efficient. Code ops shifts this entire discussion left.
Anand Kulkarni
So we are asking, how do you design software using requirements in a way that takes advantage of what’s been built before and allows the product side, the business side of the house, to work directly with people who are developers and with components of code in a way that accelerates our time to market and lets us make good decisions together about how to build software. So this is really an emerging category. Not everyone calls it the name cod ops, but that’s the term that we see that is really emerging around this. GitHub calls this inner sourcing, this idea that you’re going to find modules and reuse them. And I think that cod ops as the category looks like a growing movement that we’re excited to try and get into the front of.
Brett
Did you coin that term or did you see that term being used and say, yep, that’s a great way to describe it? We’re going to run with that.
Anand Kulkarni
Oh gosh, we’ve seen it kicked around a bit. I think we are probably one of the first and most widely used advocates of the movement, but I don’t know that we have coined the term ourselves.
Brett
That’s one of the things that I see founders tend to get wrong when it comes to category creation, is they are really obsessed with this idea of if they didn’t create the term, then they can’t use it for a category creation play. But from studying category creation, what I’ve seen is a lot of times they do exactly what you’re describing, where you essentially watch for a trend that’s emerging, no one’s really owning it, and then you essentially hijack that term and make that your own and then become the leader of the movement and evangelize that louder than anyone else can. So it sounds like you’re following that same playbook.
Anand Kulkarni
Well, good movements exist independently of companies. Good companies exist to empower and enable that movement. The idea of code reuse is not a new one. It’s actually something that when we talk about code reuse to customers or code ops to customers, they all say something like, we have always known code reuse is possible. We want to be more efficient. We don’t know how to start in finding which modules of code we can reuse from our own ecosystem while maintaining governance. We don’t know how to use artificial intelligence in our software development process to help enable that. And that’s a great starting point for our discussions about how to get started with code reuse, how to launch a code ops practice, and how to increase your software development efficiency 30% to 70% is what our customers typically experience.
Anand Kulkarni
And so that movement is one that has accelerated now with the advent of tools like Crowdbottics and GitHub Copilot. But it’s a movement that is likely going to continue as we see more and more companies thinking about how to incorporate software development practices using artificial intelligence, or just looking inside their own assets to figure out what code is important.
Brett
This show is brought to you by Front Lines Media podcast production studio that helps B2B founders launch, manage, and grow their own podcast. Now, if you’re a Founder, you may be thinking, I don’t have time to host a podcast. I’ve got a company to build. Well, that’s exactly what we built our service to do. You show up and host, and we handle literally everything else. To set up a call to discuss launching your own podcast, visit frontlines.io podcast. Now back today’s episode. Some of the companies I’ve talked to who are doing a category creation play, they say they essentially have two different marketing plans. They have the movement marketing plan, which is, you know, pushing the movement forward, and then the product marketing plan. Do you have something similar?
Brett
Is it two separate plans where you’re independently trying to market the movement separate from the product, or is that all under one plan?
Anand Kulkarni
Well, I think we are looking for people who are like minded, people who call them co champions in the movement with us. Whether or not they talk about code operations under the name code ops, or if they call it something else, those are tools that we seek to partner with and to enable. So GitHub is one example. Crabotics has been built around GitHub, actually, since its inception. Now we support GitLab as well, because that’s one of the places that when we’re talking about building, storing and reusing modules of code, you’re going to be looking inside those repositories. From day one, were excited to see that GitHub has been thinking about inner sourcing for a long time, and we’re excited to see places that we can continue to align with folks like that who are looking at this element and these questions of reuse.
Anand Kulkarni
Similarly, we look closely at what folks are doing around efficiency measurements inside development. We look closely, of course, at the wide range of emerging tools around point problems in AI with software development. Things like can you auto generate test cases, or can you auto generate, let’s say, front end react components? Because those are likewise kindred spirits in the movement of code operations tools that we want to enable our customers to have access to or have the ability to team up with, which is a great place to go.
Brett
Are you sharing numbers in terms of growth?
Anand Kulkarni
Yeah. So Crowdbotics has had a great run. We actually, for the last three years, have doubled or tripled top line revenue every year. So 200% to 300% growth. A lot of that growth has been lately in the backs of large customers building mission critical software systems on top of the stack. Some of these customers are building really enormous large scale products, things around instrumented hardware, connecting fleets, measuring large scale data across their entire organization, standing up large mission critical e commerce, retail or health systems. Also some really innovative stuff around machine learning applications. People have built to do things directly in the end user environment.
Anand Kulkarni
We’ve even seen venture backed tech companies built on top of Crowdbotics, which is exciting to see as well folks who have in some cases, raised much more than we have on the products that they created first on Crowdbotics. So those are phenomenal stories, and I think we’re going to try to maintain this pace if we can. Depends on if we are able to continue serving the needs of this movement that’s emerging as effectively has been so far.
Brett
We talked a little bit earlier about, I’m sure, the positive benefits that came from chat GPT and everything that happened last November. I would guess that one of the downsides is there’s a lot more competition and there’s a lot of noise in this space, and there’s a lot of companies that are really battling and fighting for demand. What are you doing to rise above all that noise and make sure that you capture all of that demand that’s out there?
Anand Kulkarni
Yeah, this is a fun question, right? So look, solutions that are in the market today, or that have jumped into the market lately, often think about a very narrow approach to software creation using artificial intelligence. And what I mean by that is, they are often doing things that are, at their core, simple ways to try and accelerate the literal creation of code on a character by character basis, just by putting a thin wrapper around tools like GPT four or llama two. And while that is something you can stand up very quickly, you can build a lot of products that are just wrappers around GPT or your favorite LLM in a weekend.
Anand Kulkarni
Those are not places where you can create durable value, since there is essentially no barrier to entry, since users can just go directly into chat GPT if they want to, and do the same thing, or have them to get up copilot and do the same thing. To generate code, to generate durable value here, you’ve got to have an approach that actually creates systemic value for the customer by doing something that is more than just what those tools are doing. That’s been the approach we have taken for a very long time. Crowdbotics of course, we’re not focused on the raw individual character by character code generation. We are focused on creating an ecosystem of requirements and components that are unique for a given organization and allowing them to scale up by snapping those components together.
Anand Kulkarni
That synthesizes data from inside an organization’s own repos, their own backlogs, and Jira instances, and also takes organizations forward through governance of those applications at scale. Really, we don’t look at the noise of these small, fast, sort of flash in the pan point startups that people are building around very specific elements in the software creation journey. What we think about is how do we generate the most value for customers holistically, by allowing them to manage their use of tools like that at scale. And that’s really where the durable value is.
Anand Kulkarni
So, yeah, it’s interesting, because an important takeaway for entrepreneurs who have not been through cycles like this before is you got to build a company that’s going to be relevant over ten years, not over the six months or twelve months that there is a dynamic hype cycle around language models, and I think you’re going to see that out of the current wave of enthusiastic startups that are trying to form around problems and around the product development lifecycle. There’s going to be a small number that end up becoming durable, scalable companies and then there’s going to be lots of people who are maybe jumping into this thing too fast and haven’t really thought about the dynamic, defensible value that they can create for customers.
Anand Kulkarni
So look, this is also a trend that I think unlike some of the previous transformative revolutions in tech, kind of privileges people who are already playing in the space. There’s advantages for incumbents that, at least for companies like ours. Let us take a head start into this ecosystem thats been a good place for us to be.
Brett
As I mentioned there in the intro, you’ve raised $58 million to date. What have you learned about fundraising throughout this journey?
Anand Kulkarni
Gosh, so much. Look, i’ve raised now for two companies and I think that i’ve raised in up markets. I’ve raised in down markets. The fundamentals of how and when to fundraise remain constant even if the specific environments, themes and appetite from investors shift. For me, its always been around focusing on the company’s fundamentals. So I know some founders love to raise on hype. They love to raise on what is hot. They do really well when things are booming. If you’re a crypto Founder raising in 2021, lets say, or if youre im dating myself on demand delivery application in 2011, you can ride some waves. But my approach has always been about the fundamentals of the company.
Anand Kulkarni
Do you have a giant market that you are serving with disruptive, best in class, competitive product that’s going to create meaningful large scale value for customers in the biggest possible segment of the market with a great team of leaders in and around your management team. And when you have those fundamentals, you can raise in an up market or a down market independently of almost any conditions. That also delivers the best outcomes for your and investors and your customers who want to see a durable, stable and scalable technology organization being created. I don’t know that there are other approaches to creating durable companies at scale, at least not the way that I have built companies. I think you’ve got to have the basics right, super strong product vision, great set of customers and really defined outcomes as opposed to a lot of hope.
Anand Kulkarni
The other piece I’ll say is that fundraising is also about being systematic. Early founders who I advise sometimes underestimate exactly how much work is required to run a systematic fundraising process, you are talking to no less than 30, maybe more like 50 different groups to run a truly competitive process, especially early in your cycles. Of course, later on you can be a little bit more measured since your company grows on its own momentum and you get some choice. But it’s great to create the expectation that fundraising, it’s only presented in the news as a done deal. But the reality is you are really pushing systematically for three, sometimes six months to build and generate consistent outcomes and competitive end states. And it’s important to approach it with that kind of seriousness.
Brett
Based on everything that you’ve learned so far with this company and your previous company, what’s the number one go to market piece of advice that you’d have for a Founder?
Anand Kulkarni
Great question. Gosh, go to market is so important. And I see that first time founders, sometimes second time founders, just defer the decision to think about and plan for GTM until very late in the game. So the faster you figure out your GTM, the better your company will be and the easier the outcomes will look like. So come up with a plan and a hypothesis. Relentlessly. Test that hypothesis repeatedly and continuously. Identify new channels, expansion opportunities and opportunities to scale your GTM. GTM is actually one of the places where, if you get it right, almost everything else follows naturally because your company will be a business that is scaling and working well. So those are good places for us to continue to help educate founders. I think there’s only so many different strategies that work.
Anand Kulkarni
You know, direct sales, channel sales, low touch sales, advertising, referral, viral distribution. So understanding what those are scaling, those are really the purpose of the company in the early rounds. And then as you get later in the process, you have to make sure that you are continuing to expand and multiply on those, all of which are good places to go. So there’s a little bit of a context on how I think about GTM.
Brett
Super useful. Now, final question for you. Let’s zoom out three to five years into the future. What’s the big picture vision that you’re building?
Anand Kulkarni
Okay, Crowdbotics is a new way to create software. We are looking to challenge how people have historically built. Codeops is a movement that is taking the world by storms. People are realizing there are better ways to build software. People are shifting left in their thinking about how and when to think about the code part of their software. Products and initiatives and business units want to play faster and more competitively in the product creation discussion. We are looking to create the next billion applications. Mission critical, meaningful, large scale applications on Crowdbotics we’re looking to help enterprises understand this way of building software. Also, smaller creators who want to build the next category creating software want them to show up and build on.
Anand Kulkarni
And we’re excited to partner with teams and organizations with Cio’s and executives who are looking to bring software to market faster, to introduce AI into their software development lifecycle, and to be the product that people use to do that. So that’s a big vision. We have a large scale group of applications we want to see built onto the tools, and we’re excited to see those go forward.
Brett
Amazing. Well, I love the vision and I really love this conversation. It’s been a lot of fun, and I know it’s going to be a big hit with our audience. Before we wrap up here, if there’s any founders that are listening in, they just want to follow along with your journey as you build and execute. Where should they go?
Anand Kulkarni
Best place is crowdbotics.com. If you sign up right there, you can follow along with the story, get onto our email list, and even show up and plan an application using our tools. We’d be excited to help support you, and that’s the best place to go to follow along crowdbotics.com. I love it.
Brett
All right, thanks again. Really appreciate it.
Anand Kulkarni
Likewise. All right, thanks a bunch.
Brett
Yeah, keep in touch. This episode of Category Visionaries is brought to you by Front Lines Media, Silicon Valley’s leading podcast production studio. If you’re a B2B Founder looking for help launching and growing your own podcast, visit frontlines.io podcast. And for the latest episode, search for Category Visionaries on your podcast platform of choice. Thanks for listening, and we’ll catch you on the next episode.
Twitter
facebook
linkedin