Yash From Momentum (00:00)
Hello and welcome back to Building Momentum, the show where we peel back the curtain on the exciting and often chaotic world of building a successful SaaS business. I'm Yash, your host for this show, where every episode we bring you the stories and strategies of founders who've been in the trenches, conquering churns, scaling their teams, and building products that people and businesses love. In this episode, we'll be chatting with Michael from Cerebrium. Cerebrium helps build, test, and deploy AI applications in minutes.
They're excited to hear this story and the lessons they've learned along the way. We'll be dissecting the wins, the losses, and everything in between. So buckle up, grab your headphones, and get ready to dive in the world of SaaS. Hey, Michael. How are you doing today?
Michael Louis (00:43)
Hey Yash! Good. Yourself??
Yash From Momentum (00:45)
Awesome. Thank you for joining in and thank you for having this conversation with us. So the first thing that let's start off with this, right? What's the big bad problem that Cerebrium is trying to solve?
Michael Louis (00:59)
Yeah. So I mean, we saw the Cerebrium out of kind of facing the problem ourselves where at our previous company, we tried to implement machine learning. And to be honest, I mean, I've been a software engineer for a while and it was probably the worst experience I've ever had. And basically what that issue was, you know, talent was too expensive. talent was also scarce to find tooling was fragmented. And, you know, we spend as a scale up months trying to get just one machine learning kind of base, I guess application of our product out to production.
And it took months to do that. And, we just said to ourselves, you know, it's inaccessible for any scale up or startup to do machine learning because of the amount of risk you take and without the uncertainty that it's not going to work. But the things that did make it to production completely changed our business. And so, yeah, we got to eventually acquire that business. And we said, you know, what if we could actually make this easier for companies and startups to actually develop and sell? mean, I guess that's kind of our whole premise is how can we make.
building AI based applications easier for companies, not just in an enterprise stage, but also just, you know, single day startups, series A companies.
Yash From Momentum (02:04)
understood. And then so if you can share with us and the audience a little about the status quo, a little about the stage at which cerebrum is today, that will help put everything that you are about to say in context. So where are we today?
Michael Louis (02:20)
Yeah, so we've been going for about two years now, which is over two years. We've raised a seed round from like a bunch of investors such as like YC, Angels, Sequoia, Scarf Fund, Authentic Ventures and a few others. You know, we've got thousands of developers, hundreds of customers. And yeah, we're a team only of four. I think we'll get later into my experience, but I like small lean teams because
I guess there's a lot of reasons and learnings that I've had for that. And that's kind of the stage work today.
Yash From Momentum (02:48)
-huh.
Interesting and so you mentioned so two things, right? So one is that the previous company OneCut, which is where you were a CTO. You were instrumental in building our tech operations, you scale the team from 10 to 100 and then it got acquired. So I'm sure you were working very closely with the leaders of the startup and you saw the whole journey of scaling up of OneCut and then
eventual acquisition with Walmart as well. Why go to an incubation center? So why go to YC having been through the whole journey yourself? What is it that YC has to offer? I'm sure you would like 120 ,000 is I think what they invest. Why go to YC?
Michael Louis (03:42)
Yeah, I mean, invest half a million. But my whole reason for it is actually the advice I give to many companies, sorry, to founders who look at doing YC is I'm originally from South Africa. And, know, I previously started companies in South Africa. And when we had the like the idea of starting Cerebrium, we knew that the machine learning forefront was kind of in the US. And so, you know, my kind of whole goal was how can I get to the US?
or at least not even live there or work there, but how can I tap into the network of meeting companies just because, I mean, even the companies here are further ahead of machine learning as well as have access to the talent. And so with YC, it was a phenomenal experience, but I would say the advice they actually give you, it's quite funny because they give you the simplest advice, but obviously it works and that's right. But the biggest thing that they did for us is they kind of got us into the whole networking community. So.
Yash From Momentum (04:29)
Yeah.
Michael Louis (04:37)
We met a ton of our first customers, both being YC companies or even just people I met in San Francisco while I was there. And two, it gave us the networks for investors and it gave us a network for engineers. And so for me, that was the biggest value add for YC. I think if you've been in San Francisco for 10 years and you've been working at fan companies, you already have all those connections that the money wouldn't really be worth it. And I think the value add in terms of network wouldn't be worth it. But for us, it was our entry point. We didn't really need the money at that stage to be honest.
Yash From Momentum (04:57)
Yeah.
Michael Louis (05:06)
It was purely just based on a network decision.
Yash From Momentum (05:09)
Understood. what you're saying is that YC as an incubation center or whatever, a very, very early stage investor, the enabling part that they do in terms of putting you in touch with the right people or the access that you have is almost as important as the capital that they put in or in some cases even more so because capital is something that you might be able to find somewhere else. But this is something that just requires the amount of time.
for you to be there. You have to be in SF for like five years, seven years to be able to build that sort of thing. Interesting.
Michael Louis (05:43)
Correct. I I think one thing is I don't want to discount the advice. I think if you're a first time founder, hundred percent the like the learnings YC give you are definitely there. I think for me it was my fourth company and I did the YC startup school previously. And so I guess I just kind of, I mean, I definitely learned a lot still, but at least for me it was less about the learnings and more about the network.
Yash From Momentum (05:48)
Yeah.
Yeah,
-huh.
Interesting. So there's something for everyone. So let me try and understand this a little more because why raise capital so early on? So what is it that the capital enables you to do so early on that? I mean, if it's a four people team, why not bootstrap it out? You've been part of a company that's just gotten an exit. Why have
external investors take up an equity of your company
Michael Louis (06:34)
I mean, I think there's many reasons for that. think the one thing that I try to do with at least technical companies, which I guess is, we're a service platform is you want to build a technical that at least takes kind of, I would say the perfect range is about one to three years to actually execute on. So what that means is that it's a sufficiently hard problem, but it's not this like furry vision that's so far in the future, which means that at least it puts you a bit ahead of your competition in order to execute on. And so, you know,
The one thing with infrastructure is there's a lot of like complicated pieces, especially what we were trying to do. because you know, GPUs are expensive and there's many other reasons. And so for me and just my co -founder to do it, I would say it just was going to take a long time. And obviously we want it to move fast. So I think the one thing with funding that I've always tried to do is I only raise if I'm like, when I have a very specific way to prove an hypothesis. So, you know, us raising our seed round, the very specific hypothesis.
Yash From Momentum (07:12)
Yeah.
Yeah.
Michael Louis (07:31)
was and which is common for most of companies is, there actually like a problem worth solving in? You know, is it large enough and is the problem I guess paid for enough for book like businesses?
Yash From Momentum (07:41)
Got it. you advise that for, I mean, so one of the other things that we also see is that SaaS founders, as soon as they start building, as soon as they have an idea, the first thing some SaaS founders would do is they start looking for external capital without actually defining their customers really well or the problem really well or the solution that they are building really well as well. They start putting an investor deck together. They start putting some projections together and like a team together and stuff like that.
Would you also recommend most SaaS founders to look at a hypothesis that you want to prove with a capital round? So let's raise 500k, let's raise a million, and prove whether this can be done. Can we build a product that people will use? Can the product that build that people are using, will they also pay for it? And stuff like that. So is that how an early -stage SaaS founder should also look at fundraising?
Michael Louis (08:40)
100 % I think the whole mentality of I should raise like just for the sake of it I you know, I just don't believe it also Maybe it's just my mentality from Africa There's not a lot of money to go around So you really have to be very careful about what you spend on and also the amount of time it takes to raise even a few million dollars in East and Africa You're looking at eight to ten months, which is just I guess ridiculous. We're obviously in the US It's much shorter. And so our whole thing was, you know, we had this idea. We started building it
Yash From Momentum (09:00)
Yeah.
Michael Louis (09:09)
I think we had one or two customers and then we actually raised like an angel round. And then from there we kept building and building to prove out again, like, you know, I practice and that's when we got into YC and then, you know, we raised out a seed round. But I think even as companies scaled up to even like a series a, one of the things that I just kind of disagree with is people always have these, once you hit like a million ARR, then, you you can go around and raise a series a, like they call it series a metrics. And to honest, I just disagree with that because
Yash From Momentum (09:32)
Yeah. Yeah.
Michael Louis (09:36)
I've seen so many companies get to a million ARR like so quickly, but they have the completely wrong foundation set. And so what happens is they scale up rapidly to 1 million and then they're either flat line or actually it's temporary and then it just kind of decreases. And so I think you can even get, if you get to 50 ,000 MRR like revenue, but you've done it extremely repeatable, there's so much of your customer, you've defined it well, then go out and raise the series there. I promise you, if you show that to any investors like...
Yash From Momentum (09:49)
Yeah.
big lines here.
Michael Louis (10:06)
This is like our conversion pipeline of getting customers. This is how much they pay. This is how long the onboarding time is. If you can show that even at 50 ,000, I promise you always this as Series A
Yash From Momentum (10:16)
That's a very fresh and interesting perspective, right? Not to raise capital just because you've hit a revenue target or a revenue milestone or a growth percentage for the last quarter or whatever. And I've also seen some SaaS founders, me being one of them who tried to do this. And thank God I failed at it was just because we were growing at, I think we were growing at 5 % week on week.
for a bit when I was running a SaaS product called ClientJoy. And we did that consistently for I think about three odd months. And so I was going out into the market to raise capital to maintain the momentum. So which is not necessarily a good way to, I mean, I got a lot of investor interest for sure. But then COVID hit, people went like it was a knee -jerk reaction. Nothing happened for six months and...
It didn't work out, but fair point. The other piece that I want to understand is, is you're solving, so the technology that you're building is not commodity. So it's not something that's fairly easily buildable. It's going to take some time to build it. You have raised capital, you have a CTO and you have yourself been also in tech roles before and you've led larger teams. Why a small team for
for Cerebrium.
Michael Louis (11:41)
So the one thing I always find, I mean, you hear it often from like large founders is in our previous company, when we grew to like a large, like a number of people, I ended up becoming basically an HR manager. Like even though was a CTO, I was actually just dealing with people's issues every day. And so I was kind of not doing the thing that I loved. And to be honest, I just didn't see the impact of like hiring more and more people to reach a certain goal. And so what I did is my technical team, I mean, we were doing millions of dollars in revenue.
our technical team was like 10 people. And I just told my CEO, was just like, I don't need to hire, like, I'd rather just pay our team more and just keep delivering because we're mean and like, everyone understands the product. I mean, obviously there's different arguments for hiring more people and you know, when you have different product segments and things like that. But with Cerebrium is like, I just know that with our current team, we can get to a certain goal. And right now, our biggest problem is an actually product.
and we're moving pretty well on product. It's actually like go to market and just figuring that out. for me, I prefer leaner teams. Yeah, I feel like you get more done. You get close as a company. And so yeah.
Yash From Momentum (12:50)
Yeah, as you said that, I feel like that is my life today at Momentum. That I am a very, very heavily paid HR manager. Thank you for that realization. I'll figure out a way to do something about it. So the other thing that I also want to understand from you is you were CTO at a company that scaled and then got acquired.
Why do you have a CTO in this company? Like why not have a co -founder and chief executive of the company who has built GTM motions before, who has run companies before, or who has played leadership roles in those things before? why, and this is a debate, right? So why should an experienced person spend time understanding something that they don't know when their time is extremely valuable?
Like why do that?
Michael Louis (13:47)
So would say the one thing is you really have to think about least the customer that we're selling to. And our customers are CTOs, like AI engineers, ML engineers, data scientists, and things like that. And so you really have to think about what that typical customer's like. you know, me and my entire team are all engineers. So the one thing about selling to an engineer is they're extremely difficult. When you get on a call with them, which by the way is the first miracle, is getting on a call with an engineer. Yeah. The second miracle is they come there.
Yash From Momentum (14:08)
Huh.
is very difficult. Yeah.
Michael Louis (14:15)
They've already looked at you. They've looked at all your competitors. They already know what your flaws are. And they're just going to like shoot out these questions. They're not looking for some slide deck and some pitch. And to be honest, they asked like very technical questions and most of the things come down to like scaling, performance, use cases, things like that. And to be honest, we just find that, especially in the case of AI, where like, I guess the problem set is more complex than traditional software problems. You're just going to have extremely technical questions that if someone doesn't have a taken one to understanding, then they're not going to be able to offer it.
Yash From Momentum (14:19)
Yeah.
Michael Louis (14:45)
And the worst thing already is when you have a total engineer, I'm going to connect you to someone else that then can handle your queries because they want to try and implement and see like, guess the time to value of the product straight away. And so with us again, our whole thing was, okay, my CTO can hold the product and keep holding it. And to be honest, I still code today and contribute to the product, but then at least myself, I would say I'm not your typical CTO. I'm I guess relatively extroverted and like pretty good with people. anyway, so.
It allows me with that taking understanding to do the go -to -market motions, do partnerships, I guess kind of talk to all our customers that come through our pipeline and try to them solve their problem as quickly as possible.
Yash From Momentum (15:24)
Interesting, right? So more often than not, and this is something that I've seen as well that it's not always the best products that make it to the end customer. It's the people who have built out the best distribution unless you are building for engineers, right? So because you're building for engineers, they sort of have looked at like five of the things, they spend their time doing the analysis only after that.
typically they get on a call right so they're not on the call to like they'll never be sold on brand they'll never be sold on trust they'll never be sold on testimonials they will be like hey do you do do this right if you don't do that then yeah and I feel like when yeah shut
Michael Louis (16:02)
Well, I would say developers definitely do care about brand and trust of products. think, for example, as soon as we sell, change our pricing a couple months ago, the entire dev community creeps out, but people love us all. I think with testimonials, if a good developer on Twitter has to tweet about a product, I think a lot of people would respect that person. I think the one thing is developers do not like pun...
Yash From Momentum (16:13)
-huh.
yeah.
Michael Louis (16:30)
a product unless they're like absolutely in love with it. Like even if they like a product and they use it, let's say like, you know, we're using Riverside to record this, is unless a developer doesn't love it, they're not going to tweet about it.
Yash From Momentum (16:33)
Yeah.
Yeah, yeah, yeah, that's true. That's a fair way to look at it as well. So the other piece that I want to talk about is how do you think about short -term goals at Cerebrium? So how do you set them, measure them? If I were to start a company today, how would you advise me? I should think about my short -term goals.
Michael Louis (17:04)
I guess again, it goes back to hypothesis that you're trying to prove. So for us right now, a long, I guess, part of the first kind of year or two was just getting to certain product metrics for us and performance metrics. So we hit those in the last couple of months. And now for us, it's purely based on like go to market and like, guess, getting our top of funnel right, getting our conversions right, and then listening to user feedback. So what we do in terms of goal setting is we set obviously quarterly goals and then we break those down into, guess, what we need to reach at the end of every month.
But what we do is every month we have like a kind of what we call a town hall. So we look at like, okay, what were the goals that we kind of set? You know, this is our goal for the quarter. Are we on track? How are we doing across products? How are we doing across customers? What are the massive like big issues? And then what we also do is we do like a customer review every two weeks. So we go about like, you know, all the customers that we spoke to the last two weeks, like what were their issues? What are they like about the product? What do they think is missing?
should we change up our current roadmap for the next quarter? And so that's kind of how we think about long -term goals. think one thing, it's quite funny that it's often said everywhere in like, I guess the startup manuals. But again, I think startups just don't do it as properly as they want to. Metrics are like vitally important across your product stack to know how it performs and across like your entire sales pipeline to know how you're actually converting customers and how long it takes.
I know I've learned that like every time I think I've set up the basic metrics just because like I'm checking that box I really feel like you actually just need to go so deep on metrics and it actually gives you so many so many learnings.
Yash From Momentum (18:41)
Yeah, reduces, it actually like while it teaches you a lot of things, it reduces your cost of learning from the company's standpoint, right? So if you don't measure, will end up making, like continuing to make that mistake. And it's like just having the stove on without, it just ends up.
really poorly. Here's a good analogy that I came across. It's like being on a diet plan and not measuring the weight every day. So you never know whether you're going in the right direction or not.
Michael Louis (19:15)
Well, I think for me, like to put it bluntly is it just cuts the shit in that it basically just, it doesn't lie. So it tells you exactly what is happening. You can't kind of fake the metrics And so I think with a startup, there's so many risks that you're already like experiencing. So why not be very sure about something? So when a customer churned, you can go talk to that customer and know that they churned and try learn from it. But just thinking like, I think like,
Yash From Momentum (19:22)
Yeah.
Mhm. Yeah.
Yeah.
Yeah.
Michael Louis (19:44)
they churned for a different reason or maybe they're going to come back and they haven't actually churned yet. I mean you can just tell from the data what's actually happened.
Yash From Momentum (19:49)
Yeah.
Fair, So the other thing that I'm really curious about is how do you maintain a balance between the things that your customers ask you to build and the things that you yourself sort of set out to build? So when you started, you must have created like a bucket list of things that the Cerebrium would have. But as it turns out, as when you have more and more conversations, you start to realize that things that are in bucket number two,
are a little low priority, but you sort of feel like it's because they've not experienced it, they don't know how good it will be for them. like an approach, like the Steve Jobs approach, which is where I know I will give the customer what they need. And then once they experience it, they will fall in love with it versus just building everything for the customer. Where do you sit on that and how do you maintain balance?
Michael Louis (20:42)
Look, I think that's a very tough thing to actually get right. think it's something we're still dealing with. I think the way that we found like a common ground is we obviously look at all the user feedback and then we plan like our quarter roadmap. And then what we do is throughout the, I guess the quarter, as we talk to customers, we really try and understand like what, like when they tell us they need a new feature, it's like, why? And like, is that going to get them to their end goal? And also is that the only feature? Cause often they might be like, I needed to have like light and dark mode.
And then you build it and they're like, but I also need like this feature. And then you're like, okay. And then you build that one and then like, you can't just change your roadmap, chasing a customer that might not actually need it. So we really try to understand like one, what is the complexity of this change? If it's something simple, like, I mean, we can probably do that like an hour or two, that it's worth getting that through. And again, I think customers like when you react fast, like the request. So the second thing is have enough customers that you brought it up that is becoming an issue now.
I mean, I can think of one that we're having now that a lot of customers have brought up to us in the last couple of weeks. But then I guess escalates the priority. But what we try to do is we try not to deviate too much from our roadmap unless a lot of customers have brought it up or we can like land a large customer actually releasing it.
Yash From Momentum (21:55)
That's I think a fair balanced way we do to look at it as well. So this brings us towards the end of our conversation and one of the things that we do is I ask you a question that our previous guest has asked and has left for you. And so that question that I have for you is in the direction of
Fundraising versus bootstrapping. If I were to start a company now, could you share with me or could you share with us some notes around under what circumstances should one be preferred over another? I understand that there are pros and cons of both, but under what circumstances should you prefer what path to build out the product?
Michael Louis (22:42)
I think the biggest thing is people need to just frame the problem differently. It's not whether I should raise funding or not. It's again, coming back to what we said earlier, it's what's the hypothesis I'm trying to prove? Is it, you know, I know there's a problem here and I need to build a solution to try and figure it out. I would say if that's the stage that you're at, you shouldn't raise funding because that's something that you can do in your spare time. You can test with like a couple of friends or clients, whatever the case may be. As soon as you've got a couple of people, then it can be like, I need a hire and they're asking for these features and like,
I need to test out more, then again, you can come to like the pre -seed route. I think the one thing that YC says that's very good is when you're at kind of the pre -seed seed stage, it's not really about who's on your cap table, it's about how quickly you can get cash in the bank and actually, again, continue testing your hypothesis. I think also at that stage, you're one of hundreds of investments with some of these firms that you're not gonna get that individualized attention. And so yeah, I like to, I mean, I also just don't take money from people lightly.
I put a lot of like shares of myself. So I prefer actually not raising. And I think it's Africa, we were always just building startup set with a lack of capital. We were just always kind of programmed to get to break even as quickly as possible. And so I would say, make sure that your union economics work, make sure that you can test out your hypothesis. That's what it initially is. And then once you've done that, you'll feel a lot more confident going to invest as art and income money.
you will come across more confident going to investors knowing that, I've proved this, I'm very confident it works versus just going for an idea. And so that's kind of how I, I try to like raising capital because it dilutes me as an investor. gives me like, you know, boards and shareholders to report to. And so it adds a lot of like administrative and like stress onto you. So again, I would delay it until you're very sure that it's right move to do.
Yash From Momentum (24:25)
Yeah.
Interesting. And so what's your question? What's the big challenge that you are trying to currently solve for Cerebrium?
Michael Louis (24:40)
So I would say ours at the moment is kind of go to market. Like I mentioned, engineers are very tough crowd to get in touch with. And so there's kind of two things, especially with infrastructure is there's just kind of two ways to sell an infrastructure product to an engineer. One is you either selling it as you're starting their company, which means your brand has to be very well known. So I guess getting high brand awareness. And the second thing is either through migrations where they're literally say on
Yash From Momentum (25:04)
Yeah.
Michael Louis (25:09)
AWS and you're trying to move them to like your platform or GCP. But the thing with migrations is it's very nerve -racking for every developer because like, is it actually going to be more like cheaper and more performant? And like, it kind of feels like your product staying the same. And sometimes it takes long and there's so many moving parts. so migrations, you really have to have your timing very specific to catch the developer at the right time that they're thinking about migrating. Cause maybe they've had a bad experience or it's too expensive. And so we've got to walk it into engineers. like.
What have they seen that works and that's scalable? What we've seen work is like content and partnerships with other like, you know dev tools or brands they might be familiar with however, we haven't seen that really be scalable as well as like Yeah, so I guess my question is what's a scalable go -to -market approach for kind of engineers and developers when you're an infrastructure black? Well, just I guess a developer tool
Yash From Momentum (26:03)
That's a really good question and I'll tell you why because it makes me feel good. I love it when my guests suffer from trying to give a good answer to a great question. So that's as difficult of a question as it could get and that's going to be extremely meaningful for me. Thank you Michael for joining in for this conversation and for all the people who've been watching or listening.
So this episode, whether it is on Amazon Music or YouTube or Spotify, you will be able to find Cerebrium's link in the description. If you are a developer who's building in the space of AI, do check them out. They're building out a great product, have a great platform, actually. Do check that out. Thank you for joining in.
Michael Louis (26:52)
Awesome, thanks, Yash
.png)
.webp)
