A Product Market Fit Show | Startup Podcast for Founders
Every founder has 1 goal: find product-market fit. We interview the world's most successful startup founders on the 0 to 1 part of their journeys. We've had the founders of Reddit, Rappi, Cohere, Glean, Huntress, ID.me and many more.
We go deep with entrepreneurs & VCs to provide detailed examples you can steal. Our goal is to understand product-market fit better than anyone on the planet.
Rated one of the world's top Startup Podcasts.
A Product Market Fit Show | Startup Podcast for Founders
In 2019, he went all-in on AI, grew to $3M ARR in 2 years—then to $85M ARR in 5. | Shubham Mishra, Founder of Pixis
5 years ago, Shubham had just graduated college and had no network. He bootstrapped to $3M in ARR, then raised $200M & grew to $85M ARR.
But he started trying to sell AI for marketing to enterprises with no network.
So for 2 months, he'd go to the lobby of a bank and sit there for an hour. He'd wait for the CMO to walk by, just so he could say hi. He'd tell receptionists he was waiting for an interview.
Finally, one day the CMO had enough and asked him who he was. Shubham got his 15-minute moment. That turned into a free pilot.
That free pilot turned to $1.2M ARR.
2 years after he'd started the company, he bootstrapped to $3M ARR and was profitable— since then, the growth never stopped.
Why you should listen
- How to start a startup top-down from a market vs bottoms-up from customer problems.
- Why doing insane things in the early days can lead to insane results.
- How Founder-led sales and tight feedback loops are the key for product development and customer success.
Keywords
AI startup, product market fit, challenges, marketing, scaling, fundraising
Timestamps
(00:00:00) Intro
(00:01:52) Origin Story of Pixis
(00:07:05) Getting the First Customers
(00:09:14) What is Pixis
(00:11:26) Video Generation
(00:14:54) Finding the Problem Going Top-Down
(00:19:55) The Pitch
(00:24:30) The First Pilot
(00:27:08) Scaling the Team
(00:31:57) Raising the First Round
(00:33:59) The AI Market
(00:35:54) Finding True PMF
(00:37:13) One Piece of Advice
Pablo Srugo (00:00.00)
So I spoke with Shubham , the founder and CEO of Pixis, an AI for marketing startup that has raised over $200 million, is now doing $85 million in ARR. And Shubham grew from zero to $3 million in ARR in two years, fully bootstrapped. And when I asked him how he landed his first customer, he told me that what he did is that because he had literally had no network and you had no one in the space and he wanted to land like a financial institution as a customer. So for two months, he would go to the lobby of one of the banks and just sit there and wait for the CMO to walk by and literally just say, hey, what's up? Hey, how are you doing? If people asked him, he just said he was waiting for an interview and that's what he would do every single day at lunch hour for two months until finally CMO was like, who the hell are you? And gave him like 10 to 15 minutes to pitch a startup. And that is how he landed his first customer. Welcome to the Product Market Fit Show brought to you by Mistral, a SeatStage firm based in Canada. I'm Pablo. I'm a founder turned VC.
My goal is to help early stage founders like you find product market fit.
Pablo Srugo (01:06.766)
Shubham, welcome to the show.
Shubham Mishra (1:08)
Thanks, thanks Pablo. Nice meeting you.
Pablo Srugo (1:10)
Nice meeting you too, man. Well, look, I'm looking forward to hearing your story. I see you've raised quite a bit of money, including some from Softbank, which sure is an interesting story into itself. $100 million around in 22, another $85 million last year. But really today, I mean, we'll be talking like we always do about those early days because what you told me is that you hit $3 million ARR, which is 15 people, which is...pretty exceptional, at least in the world of startups.I mean, typically startups raise a lot of money and hire a lot of people before they get to those sort of numbers. So let's start there. mean, let's start at the beginning. Like how does Pixis even come about? Like what's kind of the origin story and maybe even kind of your background that kind of feeds into starting Pixis in the first place.
Shubham Mishra (1:53)
Sure. So taking a step back. I was primarily doing a lot of research when back in the engineering college when I was in India, primarily on the side of machine learning.in the side of gaming. And that's when I would say 19 years old, I started my first venture, which was on the side of using machine learning for gaming business. And that's where I would say I took a first stab at building a business back there in college. It was just like me and my couple of other founders, and we build the whole piece, scale it up to somewhere around 20 ,000 game developers. tons of learnings and tons of failures throughout the journey, which we realized, I would say the three key highlights were don't start something which you feel, which is cool. Like because we love gaming and we found it cool, so we just started it. And we combine two things which we like. It's like, we realized that it's not an art, neither a science, but it's not an art that you like something, you combine your two passions and something big comes out of it. That's not how entrepreneurship especially the early days really work. The second piece is there are early days you need to go buy the books because you need to understand the tab, you need to do a dipstick with the customer, you need to understand the scalability of the model even before doubling down into anything. So that was the second set of learning. And the third one was, which was a big learning, which pretty much led to the downfall of the first company itself, which was in terms of understanding the margin profile of the business and longevity. And I would say it also reflected upon on the timing of the business because if you look at it, when you're in a tech industry, you could be too early, you could be too late. And both are equally dangerous. Too late, you'll be too slow in the market in terms of growth, your growth will be too expensive. Too early, you'll never grow and like eventually you might just, and even if you grow, you might not be profitable enough because your gross margins will be bad. That's what we learned because we were too early to the market when it came to the first venture, we did. Even though it scaled to a huge number, there was, I would say, a great need in terms of the customers who could use it. But the overall ecosystem of AI and machine learning wasn't developed. So the gross margin profile wasn't good because computing costs were so high back then that it just didn't make sense. Taking, I would say, we took all the learnings into one bucket and went out there to the market to understand, I would say, a kind of a survey session, understanding the problem with different levels of CXOs out there, right from different CMOs, CROs, CMOs, CEOs, to understand what's the problem. What we understood, and I think the thing that we did right, was try to speculate and predict where...
what would be the problem which will hit the market really hard in next two to three years? So that because that's a time which any software as a service solution takes to stabilize. what we did, I would say we took a step back, went ahead, spoke to tons of tons of customers out there, potentially customers out there and got unanimous feedback that when it comes to the space of marketing, especially the digital marketing piece, everyone was spending money by throwing bodies at it. By that I mean that let's hire four more people, let's hire maybe 10 more people in the creative side, maybe four more people in the narrative side. To run campaigns, we can have a couple of more agencies. Now, that whole piece and structure seemed to be, I would say, designed for a failure in long term. And we were speculating even more market wins because even back then, Google always wanted to move out of cookies. that like privacy concerns was slightly being
Pablo Srugo (5:55)
What year is this?
Shubham Mishra (5:56)
Like five years back.
Pablo Srugo (5:57)
Five years back, like 2019 -ish.
Shubham Mishra (5:58)
That's when we tried to figure out, say, that if we can build something which can solve this problem. And then look at it from the engineering standpoint and seem to be a, like I would say, very complex optimization problem, build them three pillars, but it to targeting, building the right set of creative communication. took a decision that, okay, AI would grow. AI will reach to a standpoint where all these three problems statements could be combined together. And when we started the first piece itself, even if it wasn't, I would say it was just the first version of the solution. But the impact was so high that we just set it out across a couple of customers. And I'll come to the point of how we landed upon the customers, but the impact was so high that the conversion happened right So, I would say we were lucky on that side that we got the product market fit pretty fast. But when it comes to, would say, selling into the customers, getting into the lead, that was a tough part. And coming right out of college, it was difficult to get the lead. But we decided that we'll mark two or three customers who would potentially buy this kind of a solution and directly walk into the offices. So I used to just sit three days each of their office. I still remember those days. And I used to go to the sixth floor and say down stairs that, I have an interview. And those were like, this guy has a pretty prolonged interview. It's going on for like two months. So, and finally, every day I just used to greet the CMO that, hey, good afternoon, good evening. And then eventually I might want to use the word I greet him out. he's like, dude, what do you exactly need? I told him that he -
Pablo Srugo (7:42)
I just want to understand. It's
You go to the office every day for months and when you see someone, it's like, Hey, how are you? And if anybody asks you, you see you're waiting for an interview that that was the plan.
Shubham Mishra (7:53)
Yeah, that was a plan because we didn't have any connection, nothing else out there, but the desire to sell was, the drive to sell was really high. So we're trying to, like, I was trying to meet more people in their office to build a connection.
Pablo Srugo (8:08)
And when did you go, like you just went in the morning, like you kind of figured out when he'd gone to the office. You just go in the morning for like 20 minutes and then leave or how did you make it happen?
Shubham Mishra (8:14)
So it was generally during the lunchtime because that's when people walk around. that's a time. So otherwise they're busy in the meetings, not the right time to, and it's like banking sectors, you know, the lunchtime is like pretty, prominent over there. So you walk in there, you just sit there and like,
Pablo Srugo (8:34)
that's awesome.
Shubham Mishra (8:35)
And eventually they gave me a chance and opportunity to meet the VP of marketing over there. That's when we got the first client, eventually they started the pilot. Pilot was free of cost. We gave them a free of cost pilot. And when they saw the value, they bought the solution and deployed it across the country. That's when it became like our first million dollar client and eventually expanded to one and a half million dollars.
Pablo Srugo (8:59)
Walk me through, if you don't mind, like just even what is, what is Pixis? And like you talked earlier about finding out the problem, but I have a few questions around that. Like what even is the problem space?that you were investigating in the first place and why did you go that way?
Shubham Mishra (9:14)
For me, time is very important. And the second piece is I'm a big believer that AI will disrupt three industries completely, three services industries completely where services revenue would be eventually transformed into software revenue. The first one being BPO's, which we also, UiPath did it in a huge way. The second one being marketing and third being call centers. So that was my whole hypothesis marketing is a space where $450 -500 billion is being spent just using human bodies out there so a software which can integrate all the 3 pieces right from understanding whom to target, a creative generation and a coherent creative generation and finally running the campaigns across different channels can absolutely disrupt the overall piece.
Pablo Srugo (10:04)
and even in 2019? You thought that AI was gonna be able to do all that Gen AI was non -existent in its infancy like totally back then.I mean, it's like night and day what we can do today with Gen .ai than what we could do in 2019 if anything.
Shubham Mishra (10:17)
Yeah, absolutely. But because the previous companies experience we had the leverage in the gaming space and gaming already had, I would say gaming was the most advanced when it comes to the AI side of things. So we had the experience in terms of building out solutions which could be used over there.
even though the algorithms were not as strong as what they are today, but back then, like you still had very strong, I would say, multi -arm band -aids, just small set of neural networks which could be trained and reused. And for what we did was using our gaming experience, we built our JMAI solutions inside Unreal Engine. So that gave us an edge in terms of right from the starting itself. And even now, our systems are capable of generating really high quality video. We call it the first
cinematography engine, which can do the camera motion in each and every aspect. that's something which allowed us to get the edge. Walk me through that. Actually, I'm really curious. I know this is a bit of a tangent, but like just the video thing, like we've got Sora that's like not live out yet. There's the other one, I Stintigia or whatever it's called. How does this, how does what you're doing fit into that when it comes to video? Sure. If you look at the video space, we categorize it into three pieces. First is the general purpose video generation where you have Sora and other players diffusion models and other players are coming in. The second as the second field is human based generation where you have Heygen and Cintasia and few more folks which are which which used for human generation that's primarily I would say on tutorial videos and sales content. The third space is cinematography generation where you you train the system to understand the overall camera motion how to pan in pan out how
design the scene, how to do the animation, then each and every aspect. So there are like two players, us, and there is another player which is building it for the gaming engines. But we use it for marketing, they use it for gaming, so that's a difference. But imagine that you can control, you can type it in, that if there is an all -wood shoe or if there is a fury shoe, they could just put their shoe model inside the system and say that, I want an animation, breathing animation flipping the shoe three times in the air and bring it down by breathing. it's like run by LLM, controlled by LLM, but you can run all the parameters of an animation, scene, design, and everything. So that's what we've been working upon since a long time. And even we launched our, we call something called Adroom, which is a very interesting tool already in use with tons of our customers, where you can generate
like backgrounds and stuff like if you typically use any of the diffusion models out there there are two large problems for large enterprises the first one being it's trained on low and no switch data so even the companies don't know on which data so it's it's like for auction 2000 it doesn't make sense to use it right now because until you get the indemnity from them the second one being diffusion models still do not have the control
to not change the PNG of your product. So if you put in your product and generate the background, will also change the product. So we solved that problem by taking the whole piece and generating it in the 3D space. And all the generation that we did was trained on synthetic data, which was generated inside Unreal Engine. So that makes it really powerful and unique.
Pablo Srugo (13:47)
And this stuff is live now?
Shubham Mishra (13:49)
Yeah. So we have somewhere around 25 customers already using it. Video is being used by three, it's under beta with three of our Fortune 2000 customers. So it's scaling up, but we are building it only for large set of customers, primarily for Fortune 2000.
Pablo Srugo (14:06)
So that was a bit of a tangent, but I just had to go into that rabbit hole. So going back in this space, I think what's particularly interesting to me, and I just want to make sure I get it right, is there's multiple ways of starting a startup. Usually most founders do more of a bottoms up. Like they either see a problem themselves, they notice a problem with somebody else and they kind of build up from there and then hopefully they're in a large enough tam and all these sort of things. You seem to have done it the other way, like the Jeff Bezos way of like you see the internet and then you're like, what can I solve? Cause this is a huge market. So you kind of saw that AI would drive huge efficiencies for the marketing world. And then you went into that and started talking to, and this is where you were talking about talking to CEOs, CMOs, cetera, about what specific problems.they were having today that you might solve with AI? Was that how it played out?
Shubham Mishra (14:55)
Absolutely. Because if the time is small, if you go the bottoms up way, generally you end up with a small term and that restricts you and jams the growth. Whereas I feel in the AI world, technology won't be a barrier. The only barrier would be the mindset. So if you, how big you think, how small you think, that's the only barrier.
Pablo Srugo (15:17)
Got it. And so what were some of the initial problems- Well, let me ask a few things. First of all, how did you get the conversations? You talked about sitting in the offices, but that was to make the sale. Like how do you have the conversations in the first place to discover problems and what sort of problems were top of mind in marketing in 2019?
Shubham Mishra (15:36)
Sure. So I tapped into the alumni in a true of a five college. That was the first thing which you can do. And then from there, kept building and building and out, like I would going to the different events and stuff where you can meet people and when you do not come across selling something, people are more willing to talk to you. So that's an easy phase. I won't say that's a very complex phase in general because you'll always find people who want to share their opinion, who want to express their thought leadership about how the industry is evolving, especially if you're talking to them something relevant to their space.So that was one. And the common feedback which we're getting across that, hey, know, like we all saw this. So back then it was all about like, how do we personalize more? You know, there was 2018, 19, 20, everyone was looking at personalization more to an extent where how can you personalize the content more, make it more contextual. Well, but the problem was that, and it was a derived problem whereIf you make it personalize, then you make it personalize to such a huge volume that even the first question was what to personalize? And the answer to that was an AI solution can, which we call it as targeting AI, which could figure out whom to target. Then the second one was what to say, which is creative AI. And then the third problem, a solution was now you have, okay, now you have 20,000 different communication lines. How do you even run it? Because if you put that number of ads on Facebook, Google, LinkedIn, Twitter, you'll end up spending all your money in just one day, inefficiently, until you have something which can control it to that volume, because it then becomes impossible for a human to control it. So it was a chain of problems, but it was also underlying, the underlying thing was that this is not solved as of now because there was a human approach.
being deployed over there rather than a machine.
Pablo Srugo (17:45)
like, was it personalizing emails? Was it personalizing landing pages? Was it personalizing? I mean, ad sets on
Shubham Mishra (17:52)
ad sets.
Pablo Srugo (17:52)
Ad sets Okay. On like Facebook or whatever Instagram, whatever it was, it was deployed. Okay. And the specific problem was, I mean, personalizing ad sets. mean, that's a pretty high level problem. What did you kind of decide to focus on first? What did you tackle first to deliver value quickly?
Shubham Mishra (18:07)
So the first one was targeting because whom to target. So that was a big thing. I would say even before personalization, it's about creating the ad sets. So we build a solution which could find the right set of audiences and set the targeting. But in order to do that, you need to like, would say connect with different internet tools because Facebook has limited information. So you need to read. It's, would say it was back then it was a clustering problem where you need to cluster data from different sources like Semrash, similar where maybe Google search and different APIs, and then finally put forward what should be the targeting audience right out there. And then it becomes a process automation that now I have so many audiences, can I publish it on a single click, the audiences that I like on specific platforms like Facebook, Google, LinkedIn, Twitter. So that's like the step two. And we always call it like, you build robotic process automation with intelligence, those solutions are the most scalable. because if you just have one piece, it won't scale because if we just build into intelligence, it's a recommended engine. So people don't see an uplift or a value right away. And if you just build a robotic process automation, it sells at a very small ECV, but when you have both of it, it becomes like of a high value. And we see it in the LLM worlds today where LLMs are absolutely built on the same phenomenon.
Pablo Srugo (19:32)
So take me back then now you've got this idea, you've got this problem set.
you talked to this CMO that you've pestered for like two months in the office. What is your pitch to him exactly? what kind of, how do you set up that pilot so that you would succeed with a product that I guess wasn't really in market yet. You kind of had this idea, but you hadn't actually executed against it.
Shubham Mishra (19:55)
Since I knew that I'm going to pitch to this person, so I already had the presentation ready with me. So I was just waiting for the moment to get in. So that's the least that I could do. If I were not ready for that moment, then what am I doing for the two months? So I was preparing for the presentation, which spoke all about like a small presentation because you don't want to bore them with like tons of data. So it was probably just about your current approach.
Pablo Srugo (20:27)
I'm curious, like, yeah, what are the specifics there? Like, exactly were they doing? How are they trying to solve this problem?
Shubham Mishra (20:29)
So they had like four agencies and roughly 30 people in -house to do it. Everyone was building, trying to build tons of creatives and each and everything. And they were in a geography where there were tons of different languages, different cultures. So they needed a solution like this. So it even became more prominent on their part. And since we got being a part of banking, the compliance took such a long time that when they decided to personalize and take the campaigns live, to the time that it'll go live was a huge distance. It took months and months. So the level of frustration that was being built up was also right and right for me to get in and sell something like this.
Pablo Srugo (21:20)
That's the asset. How would they even personalize it? Because you talked about generating audiences. Like what kind of work were they doing? I mean, Facebook itself, like some of these platforms will do some audience generation of their own. Were they just limited to that or were they already? trying to like manually tie a bunch of things together.
Shubham Mishra (21:36)
Being a bank, they had their huge analytics team of 20 people, which is separate. So they used to run analytics on their CRM. Then there was another agency which will take that analysis, map it back onto the possible audiences on Facebook, Google, LinkedIn, Twitter, give that audience back to the next agency, which would publish those audiences. And even before that, there was another team.
Pablo Srugo (22:00.91)
which would create those creators, generate those creators and then try and generate those creators and then publish it. So, and since they wanted to build a control because they cannot spend tiles of money, everything was published as a separate campaign not to like base tiles of money. that was a very tiresome process. So what I told them that, this whole chain could be broken. Just imagine the whole chain could be done using just one single flick of a button Uh, and, and all you need to do is focus on approvals, which, which is the compliance side of things rather than managing the complexity of each and every aspect.
Pablo Srugo (22:40)
And how did you scope that kind of first pilot to make sure that the tech would work and it would succeed. Cause I mean, we're talking about like, guess what's shocking to me is like, we're talking about a pretty big product, right? Like it's, it's 2019. First of all, it's hard to your head around, but things have evolved so much in five years. You're talking about taking a bunch of data. tying it together, potentially creating ad sets themselves, like generation part, which is shocking to me that you could do that back then, or to what extent you could do that. I'm really curious on how did you scope things out so that your tech would get there? Two and a half, three months, mean, it's great you bought us some time, but it's still, that's not a lot of time to build a product, right?
Shubham Mishra (23:20)
No, for sure, because we were doing tons of research on that, and back even before starting the company, I was publishing tons of papers with the...different offices based out of Asia, Ukraine on different algorithms, which was not exactly on that side, but it could be applied in the space too. So that, the background of technology, understanding of technology definitely helped us out. And even those connections, because when it came to, I would say tech hiring, it was fast for us because we could just tap into our, I would say, Eastern European network of professors who were working on this space, our Asian network of professors who working on this space. And immediately like the team was ready, which was like to help us out in terms of doing the initial bunch of research. Like I would say we tapped into the front -end, back -end development in India itself because it's like very fast where you can move things. And also my CTO
She was one of the finest coders out there.
Pablo Srugo (24:26)
What parts of the puzzle did you solve for them in that first pilot?
Shubham Mishra (24:29)
We solved for the first piece of clustering, which was, I would say, the generation of the targeting audience and publishing of the audiences using APIs of Facebook, Google, LinkedIn, Twitter in an automated manner. And in between, we build a bridge for an approval. So that even when the audience has been down, there is a compliance piece being connected. And we solved the final. in terms of where we gave them the initial, I would say, basic level of multi -arm band -aid kind of algorithm, where they could run those campaigns if there are, like I would say, 45 campaigns published. Now, which campaign or which ad said how much to allocate, they could just give a high level of instruction, and the multi -arm band -aid will keep doing that. Which now has it all to a much deeper regardless but that was what we did.
Pablo Srugo (25:20)
And then in terms of creating the actual ad sets back then you just worked with the agencies to actually create the content and the imaging and all those sort of things.
Shubham Mishra (25:27)
Yeah so that we told them that that would be a part of the paid machine so once you once you this is the part of the trial so that like you know how you sell says that this is a locked feature for now and you can unlock it when you pay for it
Pablo Srugo (25:47) so that's part of that was that kind of part of it buying yourself time to develop that part of the product?
Shubham Mishra (25:47)
Absolutely.
Pablo Srugo (25:50)
And how did that pilot go? what was, I'm curious on what were the KPIs that they were looking for? Was it just the performance of the ads, the performance of that campaign versus campaigns they'd run before? Like, what was the number one thing that they were looking at to determine success?
Shubham Mishra (26:05)
So I would say three things. First was the time saving of the team. And second was the throughput. So the throughput was like if their team could, if their team put in the same amount of time versus the machine use, What's the output that the engine rage and dumps of volume? And the third one was, which was also important, was a baseline that after doing all of this, the performance still remains the same, better, or it falls down. So these are the three KPIs. The best thing was that they saw an uplift in all the three sites, which is pretty obvious to us because it had to happen, because now you're going deeper. And that's how That's how you do it. It should be done, but it's impossible for manual approach. But it's like a eureka moment for when you see automation for the first time. So that's what happened over there.
Pablo Srugo (26:57)
then what happens, how did, you know, because we were talking about earlier, you quickly end up getting to three million in revenue. This turns in from a pilot to paid. Like what contracts size are we talking about?
Shubham Mishra (27:10)
a huge bank out there. So it scaled up from just I would say a credit card line to debit card line to loans, personal loans. like I would say the upsell inside, we had to hire a person just to do that and manage it. So which took like I would say one year to scale it up to $1.2 million worth of recurring revenues. And We sold into one of a $10 billion food tech company out there, which had a similar problem. But it was more dependent on digital marketing than a bank. These two combined together and a couple of small companies combined together, we reached $3 million pretty quickly.
Pablo Srugo (27:54)
This is the interesting with enterprise, right? And especially when you have such a big opportunity, well, there's a few pieces. One of them is how quickly do move off of that enterprise and try to land a new customer. How did you think through that? Because the last thing you want is your first customer to not be successful. You also don't want to wait too long and then you're overly dependent and it actually is much harder to sell it to others than you thought. So like, how did you manage providing a successful product to this customer and expanding versus moving on to new customers, adding new customers and everything that that entails like from a go -to -market and product and customer success perspective.
Shubham Mishra (28:32)
I mean, that's when the hiring and scaling of the team comes into picture because it has to be like, it depends on how good the team you have. And it completely depends on that. It's like no other external factor dominates that because if you have a good team and if you can manage things well with the team and if they can take up the account and scale it up, which should be the case, then you should immediately move to the next one and build that because as a founder, During early days, need to build the playbooks and hand over the playbook to leaders who can run the playbook for you. And that's what we did. We kept selling into enterprise, built a playbook for Asia, then we built it for US, and then handed it over to experts who have been doing it for many years, because that's when the chances of failing even falls down. Because there is a limit to which you can reach just with hustle. because hustle and energy is great. It's cool. But when it comes to the wisdom and experience, it helps you scale the things to the next level, which is much needed for a startup.
Pablo Srugo (29:45)
And how did you make sure working through that bank that you were working on product that was scalable and would transfer to other customers versus, again, this is the other problem of kind of your tiny little startup working with a massive enterprise and they've got all these needs and you're not necessarily sure. What are bespoke needs that they have that they would love to see in a product, but other customers couldn't care less about it. And all of a sudden you're developing custom software that won't scale.
Shubham Mishra (30:09) Yeah, absolutely. So that's like the classic dilemma between, and that's like the pivoting moment which decides that the new base services company or just the services arm for a large organization or will you scale to a SaaS which will end up serving thousands of customers. So I would say it's doubling down on sales, parallel right after your first success. but sales in a manner where it should be a founder led sales because founder led sales is more receptive. Like you can go out and figure out these things on the spot itself rather than hiring someone during early days. So we didn't hire any of the sales guys until we were like beyond five, six million ARR until then it was just
Pablo Srugo (30:53) You did stuff yourself till five, six million ARR?
Shubham Mishra (30:56)
Yeah, like completely end to end and ensure that like we reach to a point where there is some level of playbook, some level of confidence built out. And that's when we raised a decent amount of capital invested in sales. Otherwise, it was all investment in tech and customer success rather than sales. Because the moment you invest in sales, beyond a particular point, the sales team won't bring the feedback to you from the customer, what would you expect, especially during the early days?
Pablo Srugo (31:39)
It's all about tight feedback loops and that's what's so important when you're trying to figure out what problems really resonate, what solutions really resonate. And the more layers you have between yourself and the customer, especially early on, but this is true always, but especially early on, the slower those feedback loops, the more, I mean, it's the classic like telephone game, right? Like the messages get twisted by the time they get to you. Another question for you actually is :When did you raise your first round? Like you started in 2019, you worked with that bank. When does that, what is your first round and when did you raise it?
Shubham Mishra (31:57)
We raised almost all the rounds. Like the first one after one and a half years of that, was a series A because we reached to that ARR level profitability and all the aspects. So we raised somewhere around $7 million of series A.
Pablo Srugo (32:14)
This was after you hit, so you got to 3 million bootstrapped?
Shubham Mishra (32:17)
Yeah, completely.
Pablo Srugo (32:19) ‘
And you were profitable?
Shubham Mishra (32:20)
Yeah, but at that point in time, we were super broke.
Pablo Srugo (32:22)
And what made you decide to fundraise in that case?
Shubham Mishra (32:26)
So, because we wanted to expand to the US market because we saw that the cross -border piece is extremely massive and we have proven ourselves in a tough geography to where it's like very hard to sell software. So now if this works here, then it can work very really well in the North and South American markets. So that's when we took more capital in future and a good partner who can help us out in terms of, I would say, but I knew like early days choosing the right partner is very important. If you get the wrong partner, you'll never scale and you'll be stuck with like wrong problems, which you should never focus on. So we are fortunate to end up with the right partner who knew how to allow us to peel from across whatever pattern from India to US shifted the complete focus to the US market and scale it up a little bit.
Pablo Srugo (33:23)
And this is well outside of product market fit, but I have to ask anyways, because obviously AI is top of mind for just about any founder today. And one of the questions is, know, startups versus incumbents in a sense, like giving your stage, you're, you're a bit of an incumbent in the sense that you already have a bunch of existing customers. What does gen AI do for you? Like, does it create a bunch of new startups that are competitive, especially in the generation space? I could imagine that there's a lot of people out there saying, we’ll generate different sort of assets, different pieces of the marketing puzzle. But is that, do you see that as a tailwind for you or is that a bit of a headwind because now everything gets kind of so crowded?
Shubham Mishra (33:59)
Because if you look at it, when we were selling two, three years back, everyone was like, okay, this guy is selling AI, we don't even know if AI exists or not. But now everyone believes in AI and everyone wants it. I would say the good thing. There are tons of general purpose AI solutions which are built out of OpenAI and on top of OpenAI, which solves the problem for SMEs, mid -market and a of things. But still there is a huge wide gap when it comes to enterprise and mid -market and upper mid -market, and that's where we focus on. And they are more concerned about the security on premises, their needs are very different. And we have been building this like since five years now, so The level of trust and the communication with which we go in right now allows us to penetrate the market even faster. So we are seeing super positive wins for us. And also I would say all the layered AI solution, was like, if you look at it, there were tons of startups which got funded. This is why building a layer on writing like a 10 line script on top of OpenAI. And that face like now going down because there is a limit to which you can scale a product which doesn't have a moat. GTM can be a moat, but product also has to have a moat, otherwise it can scale to a level then it flattens out. So I think it's a good time for us.
Pablo Srugo (35:24)
Perfect, that makes a lot of sense. So we'll stop it there. I'll end with the two questions we always end on. The first one is, when did you feel like you had true product market fit?
Shubham Mishra (35:34)
To be honest, after $10 million and when we got the first US customer, around that point, that's when we got, that's when I go, what are you, it works in the main job or feature because earlier, you you do the Series A, you do it, you do it because you feel like it will work, but eventually you need to see that the product really works in the US market. So that was a great relief and the point where we believe that, okay, now let's do it. So.
Pablo Srugo (36:03)
What is it about the US market that's so special? Like I know it's a big market, but I mean, India and East Europe, I there's a lot
money out there too, there's a lot of people, a lot of enterprises. What is it about the US market that makes it just so important to be the market you have to crack?
Shubham Mishra (36:17)
So I would say when it comes to a software heavy market, reception of software and the true usage of software of sales is extremely high when it comes to US markets. So especially North American markets, if you're able to crack and build your positioning over here, then you're at a different ballgame altogether. and which means that you don't have to scale your team to like 10,000 people to reach a number. Whereas if you operate in Asian geography, it's still like, I would say, services heavy, even after the product, even if you have a product, but still it's a low paying market and high demand. So whereas US is more value driven, if the solution delivers the value, then you get paid.
Pablo Srugo (37:02)
Got it. And then the last question is if you could go back in time five years ago when you were just starting Pixis with like some important piece of advice for you as a founder, what might that be?
Shubham Mishra (37:13)
If I were to change certain things, I'll directly start with the US market even earlier, maybe partly with the Asian market itself to reach the product market even faster. And even in terms of the positioning of the product, I would have made it like simpler because again, We complicated it too much during our early days when people didn't understand AI and stuff. we just kept pushing in our agenda of AI. Now it makes sense, but it's like lucky to be in this time, but we could have done it better. So we could have made it simpler to make people understand. So that's like the per up positioning. I feel like we were really bad at it.
Pablo Srugo (37:55)
And by the way, like, so you've raised $200 million How many customers do you have? Where are you at, more or less revenue -wise?
Shubham Mishra (37:59)
We're around more than 400 customers, and we'll be crossing close to $80 -85 million of ARR this year. And hopefully, we'll be soon in close to the $100 million mark, which is a big milestone for us.
Pablo Srugo (38:16)
That's a huge milestone. Well, the milestone is always the ones, right? $1 million, $10 million, and I'm sure you're going to cross $100 real soon, man. So it's been a pleasure speaking with you today. Thank you for taking the time to jump on the show.
Shubham Mishra (38:27)
Thanks a lot for having me here, man. Thank you. Bye.
Pablo Srugo (38:30)
If you've listened to this episode and the show and you like it, I have a huge favor to ask for you. Well, it's actually a really small favor, but it has huge impact. But whichever app you're listening to this episode on, take it out, go to a product market fit show and leave a review. Please. It's going to help. It's not just going to help me to be clear. It's going to help other founders discover the show because the algorithms, whether it's Spotify, whether it's Apple, whether it's any other podcast player. One of the big things they look at is frequency of reviews. It's quantity of reviews. And the reality is if all of you listening right now left reviews, we would have thousands of reviews. So please take literally a minute, even if you're just writing like great podcast or I love this podcast, whatever it is, just write a few words. Obviously the longer, the better, the more detailed, the better, but write anything, leave five stars and you will be helping me. But most importantly, many other founders just like you discovered the show. Thank you.