 Hey, everyone. This is Carlos. I'm the founder and CEO at Product School. Today, I'm here with another product CEO. His name is Krish Ramineni. He's the founder and CEO of Fireflies.ai. Krish, welcome to the show. Hi, Carlos. Thanks for having me on. I'm looking forward to it. Your product is so hot these days. It's for once we don't know it yet. AI tool that records, transcribes, and analyzes meetings. So I just want to ask you, how long have you been working on AI before it became cool? So that's a great question. And I honestly will have to say that when we started working on it, it wasn't very cool. And to be even more blunt, we didn't think it was going to work. And most people didn't believe in the space because there was so much uncertainty around the technology, around the use case. And even just the customer experience. So for us, we've been working on this core technology dating back to 2018, specifically Boy starting in 2019. And the vision was really about how can I have a personal system that follows me around and transcribes and takes notes for my meetings and really recreate what usually executives that organizations have. They have their own business admin or secretary. And we wanted to create that sort of thing with AI. And it's just started with that simple vision. So I'm definitely going to dig deeper into that. But before I also want to take a sneak peek into your journey. I see that you are really a pure breed product person. You've been working in products since 2014 as an intern first. And then you work your way up. So I need to say more about how you got that first PM job. What happened was after my freshman year of college, me and a friend got together and said, hey, these internships are cool, but let's just go build something on our own and learn how to code, how to put things together. When you're a freshman, you're not going to have a lot of immediate skill sets. And through that process, we built the entire product from scratch. It was a simple product at the time. It was events for colleges. I think that's something that a lot of people that are in college do. So this was pre Facebook events going big. And we were able to just wear many hats, write the product, design it, hire marketers, ambassadors and try to grow it. And we went through the full spectrum, right? When people say you are the owner of your feature or the CEO of your feature, it's very much true as a product person. So I think that was our first break into building things, something on our own. And then afterwards, what happened was, hey, this is a lot of fun, but I don't know if I want to do like a traditional software engineering job. Like that sounds boring to me. I don't know if I want to go work at a big company and then someone introduced me to the PM space. So my first official PM gig was actually working with a startup out of Wharton, where it was a two person company and they got funded through Wharton while we were in college. They were MBA students. And I was really like their first person other than the founders working with them. And that was like PM experience again, like very hands-on. And through that, worked at Workday as part of their first inaugural PM program, I think it's one of the first or second programs there. And I went through that rotation program as an intern. And then fast forward, I was a PM at Microsoft. So I think it really started out with being an entrepreneur. First and foremost, working on projects, hackathons, and then translating those skills into a PM role. And I see a lot of overlap between building products or even building companies, especially at the early stage because when the company is small, pretty much the company equals product. You get to a point obviously where the company gets much bigger than the product, but I like what you said about building for fun. Maybe it wasn't even an official company or maybe you didn't have any official title but you were just having fun and learning. And then from there, obviously it gets to a point that you have to decide if you want to formalize that or not, but it's interesting to me to see that you've been building products. Even before product was a thing, even before you had the official title. And that's important for a lot of people who are thinking, well, how can I break into product? Well, sometimes you don't even need permission for that. You can try. Yeah, there's a lot of great books out there, resources out there now that when I was going through the journey and guidance, like even what you guys do at product school, but to be honest, when I first started, I was really focused around just building things and being a tinker with different things, putting things together, get it in front of customers, get their feedback. So I was very old school product and then I learned the data piece to being a good product manager. And when I was at Microsoft, that was actually the team that I was on. It was like the first AB experimentation team for office, which is a 30 year old product. And now you're running AB experiments and tests. You're not just sitting in a room and brainstorming and coming up with ideas, but you're also measuring and moving metrics. So I think a lot of the PM role today, having the data skills is a must. It's essential. And you can also go into like, how do you be a PM for an AI product? That's another thing, like what technical skills do you have there? But yeah, like I didn't have any formal training. It was just literally jumping in and we were learning about the industry. We're learning about the space, the customers. I didn't even know that PM was a title that existed when I started doing all this stuff. Me neither. And I also seen these role of all from being mostly an art into mostly a science with a sparkle of art. Because as you mentioned, there's now much more data and many more tools. This is a much better understanding, more maturity, more companies hiding PM. So they're better frameworks and best practices around how to actually maximize the chances of success. There's still a component that can only be acquired through any experience and in some cases even lack. But I think that this much data-driven approach within products is a much better approach than just expecting to be a visionary that has an amazing idea and somehow it works. Absolutely. And I know a lot of people refer to Steve Jobs and the things that he's done and I've listened to some of his talks. And I think that a lot of the visionary stuff that he gets attributed for also come from experiences and understanding and measuring and looking at like the psychology of users. So he had a really good sense of that. And then he drew insights and parallels and where it was able to build great product. So sometimes people defer to well, Steve Jobs didn't talk to people, he just envisioned something. But I think his life experiences really shaped that. And really what is creating data oriented products? It's essentially pattern matching. And so some of the most veteran PMs when it does seem like intuition, it's because they went through many, many cycles of trying things, failing at them. And you are able to create this heuristic and mapping because at the end of the day it's pattern matching. Sometimes it doesn't always work but that is also like a skill set. So nothing just happens spontaneously in your head. It takes many, many iterations. And just like what I tell my PMs today, there is no silver bullet that's gonna move a metric or wow a customer. It has to be like 10, 20 little shifts and adjustments that get you to the North Star. Once you have more of like the product market fit or like the area of the problem that you wanna solve. Firefly is the best example of that. The version of the product that we launched in the very beginning versus the product it is today, we look back over the last two, three years and the evolution of how much we had to adjust it. So it's not something that just happened in our head. We had to really listen to feedback and look at the data. Let's talk about those early days and your own transition from product into CEO. What was that pain? How do you come up with that pain point and that early MVP? When I was at Microsoft, I was very much novice. I was a PM with about a year, a little more than a year of experience in the industry. I didn't know much about SaaS. Again, same parallels to when I started my PM journey. I don't think I had the skill sets and background and experience to go jump into running a startup. That's something that I say very honestly and very humbly and as a result of that, it took us some time to walk through the desert and understand what it takes to be a founder, what it takes to build a SaaS product because I didn't have enterprise experience to go build an enterprise product. I didn't even know what Salesforce was at that time. So a lot of learnings. I think the first two years of the journey was figuring things out, making lots of mistakes. And maybe I could have avoided some of those if I had a few more years of experience under my belt at a larger organization. But what really helped us move towards being able to build Fireflies was that hunger and that almost naive sense of saying, we can do this, why not? We were almost oblivious to the fact of how hard it is to build an enterprise SaaS product. We also have some consumer elements in our product and I always believe that collaboration tools, some of the best ones out there have a lot of the characteristics of a consumer type tool and you're working in the B2B space. So that's the sweet spot for me. I love products like Slack, Dropbox and so we took a lot of inspiration from how those products were built. And the transition from being a PM at Microsoft, owning a few features to going and starting something like from scratch is you are doing everything. You're doing support, you're doing all of the operations, you are doing sales. So product ends up being like 20% of what you do and you're doing a whole bunch of other things. Now I've made it so that I spend more time on product because that's an area that I like to own and I'm very passionate about. But you also realize that a good PM actually can benefit from doing sales, talking to customers from that lens, doing support, hiring, all of those things actually shape me to better understand who our target audience was, who's our ideal customer profile. Yeah. And honestly, you can apply the product framework to any role. Yeah, but let's talk about the area that MVP. Like here, I see here you started in 2016. That's way before the pandemic, probably before Zoom was started or at least before Zoom was mainstream, right? And your homepage says that this is a tool to automate your meeting notes. So what was your minimum viable product? What type of notes were you automating in a world where it wasn't really that obvious that those type of notes would happen online versus in person? Yeah, so correction is that in 2016, we started working on a bunch of other tools before we got to the meeting notes assistant. So it was almost five, six iterations before we decided to double down in 2018 on the AI note taker platform. And so in 2016, we still started with the same thesis that there is knowledge buried inside conversations and how do we unlock them? So first, we built a email assistant that would go through all your emails and tell you what are the most important emails you need to focus on. Then we built a automatic project management system that would go through all your Slack messages and tell you like which Slack messages are actions that you have to complete. We built a Chrome extension that would track messages you're sending across like Messenger and LinkedIn and all these places and then create an automatic to-do list. So that was actually the formation of Fireflies, those early products where cool idea seems relatively valuable, but technology was way too early. Like we didn't have LLMs and open AI and chat GPT to make this stuff possible. You had to use good old school NLP and it's not always gonna be accurate. So those were our minimal viable products to test the technology. And then we later realized, well, we need a space that's a blue ocean that has a lot more value than emails and chat messages and stuff. And we said, well, I can send an email and I can remember an email I sent two years ago, but why can't I remember a meeting I had two hours ago? So it started with that premise. And then we said, let's apply all of this learning and technology to the meeting space and first build a platform that can capture meetings. And so yes, around that time, there was Microsoft Teams, Zoom. So we said, we'll start with video conferencing since a lot of people tend to have meetings on video calls. We had no idea, right? This is well before the pandemic, so that everyone was gonna shift to online. So we actually built for a world where some meetings were gonna be in-person and some meetings were gonna be online and it was really gonna be a tool for the conference room. The moment you enter the conference room, you have fireflies be able to do all this stuff. And through that process, we were lucky too on the timing because once the pandemic happened, everyone went online and this became more center stage rather than a fringe tool. So I would say it's hard to say what an MVP was at the time, but now when we were starting to work on this in 2018, 2019, we learned from all those trial and errors and stuff and we did like a pivot to the meeting space. And that's what I was trying to get with the MVP at the journey, right? You had to go through multiple pivots and that V1 in 2016, very different than where you are today. It's important that you were able to recognize those situations. You mentioned timing is important, right? I was trying to grab my head around, okay, this makes so much sense today, but I think in 2016, it's just early. So but the common thing for you, I guess was more the productivity side, right? Like how to save people's time through automating certain tasks that don't require a human. And you eventually found that sweet spot with the meetings. My other question to you is about at what point do you leverage AI, right? We're talking about now, chat GPT and other LMMs, but before what was that early version of how were you automating that without the type of access to AI that we have today? Yeah, so one thing that I think also was very helpful as we built the first iterations of Fireflies was realizing that what are the problems we're solving for ourselves? I realized when at the early beginning of the journey, we're talking to a lot of investors, a lot of customers, a lot of salespeople, and I couldn't remember everything now. And the first thing I would do is like, I'd be taking notes down on a pen with a pen and paper, and then re-uploading that online. So it wasn't like the most effective way. So I wanted to look at something where it's an action we do every day, it's a habit, and then go improve that process. So we went in earlier, which was a mistake, saying we have this cool technology which is like natural language processing, classification models, let's go find a problem that matches this technology. That's why we did all that Chrome stuff, Chrome extension stuff, email stuff, Slack stuff, because Slack bots are cool back in 2016. So that's a mistake I wouldn't do. Don't fall in love with the technology, fall in love with the problem, and then figure out what technology you need in order to help address that. So because of that, it took us a lot longer because we were trying to find a technology fit to a problem. Once we got into the meeting space, it was helpful because, okay, all the technology didn't go to waste, so we could reuse some of that core components from building the Slack bots and all of this stuff. So yeah, it is a challenge to be able to say at that time, is the NLP gonna be good enough? And we said, okay, it doesn't have to be perfect, but as long as we can search and capture that information, there was enough good search technology, there was enough really good technology around the process that you have for doing natural language and classification and all of that stuff. So that helped. So I'm looking at your LinkedIn, it says that you've been building Fireflies for almost eight years. And now when everybody's talking about AI, it seems obvious, right? And it's like, oh, cool, you are now, you are not just adding the AI word to your homepage to try to raise money. You've been building and trying and hustling for a very, very long time. And it's only now, maybe like seven years of hustle to give you that overnight success. They want to talk about that moment, like that hockey stick moment, maybe even before this year, like let's talk about the beginning of the pandemic when clearly doing meetings online is the only way to do meetings when taking meeting notes becomes an absolute need for a lot of people. So how were you able to handle that type of increase in demand? The demand happened honestly as a compounding effect. And it was really because of the word about the virality of people seeing it in meetings. So at the start of the pandemic, there was a lot of heavy adoption. We thought that was gonna be the peak of excitement for this sort of technology. And then right now we're growing much faster today than at the start of the pandemic. And I think this whole generative AI wave has opened up a whole new interest in using LLMs and this sort of tech. But for us, we haven't spent much money on doing your traditional marketing channels like paid marketing. It was really a function of a product that people talked about the word of mouth, seeing fireflies on the meeting and then exploring it, getting the notes afterwards and being like, wow, this is cool. Let me go ahead and try it. So those were all the key factors. And usually, it takes one, two, three years and you're like, okay, we're growing steadily. And then there's this explosion that happens from all of this compounding network effects. That took a long time. And honestly, I didn't know that sort of explosion or exponential growth could be possible. But you usually hit a tipping point where there is an attraction where you go from the fringes to becoming, I would say, almost mainstream in a way. So those really helped. But from an engineering point of view, there's a lot of stress, right? You have to keep fireflies up and running all the time 24 seven. I can be late to a meeting and I apologize for being late to this one. But fireflies has to show up on time every single time for millions of people. So that is an infrastructure challenge that people don't realize that some of the simplest products that you need to build at scale require a lot of resources and energy. And you have to get that right. So this is why the Facebooks, the, you know, Instagrams and the Twitters of the World, the scale that they're supporting and their consumers that we don't have like that volume that we have to deal with, it takes so much energy and effort. So some of the simplest products are the hardest to build at scale. And from the funding standpoint, I can imagine at the beginning when you were trying to explain what you're building way earlier, right? Taking notes and paper to now, how were you able to also capitalize on the opportunity and make sure that you are funded to continue investing in the growth of your business? Yeah, at that time, we always felt like you had to complicate ideas. I think the best ideas are simple and you have to explain to them why you're gonna be able to do it. You can't put in a lot of bells and whistles and make it seem more complicated. And people like to complicate things because that they feel helps them be unique or differentiated from other products in the market. So part of it is having a simple, clear vision. The other part is execution and finding folks that believe that you are the right team to execute it, right? When YouTube came out, there were several other video players, but YouTube executed I think really well on the distribution side of things. So it was more of like when we went to investors, we were spending so much time trying to explain to them why we were different. And instead, what helped us more was, you know what, let's just go heads down and build it and show them instead of tell them. And we were really bootstrapped. We were really, really scrappy. And once we started executing, had something to show for it, that helped. There are other folks that, you know, second time founders more established folks in the industry, they can go out of deck with just a conversation and go raise a couple million dollars, right? We didn't have that luxury. So we had to actually go build the product, proof of concept, have some customers and show semblance of progress like the good old school way. But that was really tough because the cost of building sort of this sort of technology back in 2018 was really expensive voice and transcription, all of that. We had to go down that route. And I think that old school mindset that you're referring to is now becoming a new school. And being bootstrapped is cool again. And being profitable is something, it's a good thing. But I understand, like I also bootstrapped product school for the longest time. We also raised money during the pandemic. So I can't relate to that story. And looking at Crunchbase, and it seems like you raised your series A in 2021. That's correct. So we did our seed in 2019 and then series A led by Coastal Aventures in 2021. All right, that's incredible. So you bootstrap all the way through these pivot. Even at the beginning of the pandemic, then you raised capital, I mean, series A in 2021. And I can imagine a lot of the people that probably said not yet are knocking at your door when AI is, it's all over the place. But what are your plans to then take this AI product that you build to the next phase? Yeah. We are very excited about the adoption of the AI note taker. The way I think about Fireflies today is we built a platform. The AI note taker is one of a hundred apps that can be built around this conversation layer. And we are very much focused on what is knowledge inside an enterprise and how can we unlock knowledge inside an enterprise? Going back to 2016, same roots, same DNA. And we think about, well, notes is one form of knowledge you can extract from conversations. Can you extract other types of knowledge from conversations? Maybe, you know, I'll give a product example. I wanna be able to go back through my last hundred user interviews that my team did with customers and pull out all the top feature requests for, you know, generative AI features that customers are requesting. We can build an app for that on top of Fireflies. Anything that can go through those conversations and unlock, so whether it's like scoring calls, whether it's writing notes, whether it's extracting tickets and insights, objections, feature requests, that is the future that we're hoping for is where we create and extract knowledge from the platform. I like that approach for very specific reason. I've seen a lot of these companies that start pretty much as a feature. Eventually those features become a commodity. Like now Zoom, Google Meet, Microsoft Teams already allow the option to record a meeting, to transcribe a meeting. And if you can become a platform that is not just tied to a specific video conferencing tool but you can really extract insights across so many other platforms, you become that knowledge base for the entire organization. And you become much more specific to it. And really think about actions and like workflows as well. That's something that I think really big about. So my brother who was just here, they were working on an idea sometime back around, what if we had someone that does the interviews automatically instead of like a human? And so for them, it's a product that they can build potentially on top of Fireflies where you have bots that go out and do the interviews and transcribe everything and then bring it back. So we also like to think about this platform play as let's deploy AI agents and have them do work for you. That's what the rage is right now with all the generative and LLMs. So like the example that like my brother was working on at that time is a really cool thing of, yeah, we have the video conferencing layer where you can build all these like intelligent bots that can go do work and really follow you around, right? So right now Fireflies is a passive listener on a call, it's taking notes, but what if it can be active, what it can be like Jarvis like in Iron Man, those are really cool things too. But we ourselves may not be able to think through all of those ideas. So we wanna create an ecosystem for more apps and agents to do that. That's what I was gonna say. I've seen that evolution from product or from feature to product to platform and then eventually from platform to ecosystem and then allowing third party companies or developers to build on top of your core product. Slack is a great example of Salesforce, Zoom, you name it. Now my question to you is obviously as a series A startup trying to take over the world, how do you go about picking your battles and making sure that people can really see the value that you provide versus trying to take the easy route and stay on Zoom or any other like platform that has already been around for a longer period of time? We're always looking at where are people having conversations and how can we help them? So I think video conferencing is a good slice of the pie but there's more, right? People have in-person meetings. So we're working on something to help capture and transcribe and take notes for in-person meetings. People use telephony systems. So we have a partnership and integration in place with AirCall and Ring Central. So I look at the holistic of where are people having conversations and how can we help them get perfect memory over all of those calls and meetings that they're having? So I think video conferencing is great but that's only part of the story. There's way more that we also have to dive into. One other thing you mentioned around not spending a lot of money in marketing and mostly leveraging your own product as a marketing channels, some companies will call product-led growth but regardless of the label, I think that is really smart move because I've been in, that's how I became a customer. I was on a video call and suddenly I see a bot there. What is that? I don't know if it's a tool to transcribe meetings and take notes and it's like, oh cool. And then I ended up signing up. So how are you thinking about that type of marketing like creating that bottoms up approach with also supporting that from the top and making sure that a decision maker, especially a large company can understand the value and eventually become a customer. Yeah. So that was by intention and we did product-led growth before it was cool. It's also because we didn't know any better. We were product and engineering oriented founders and we really didn't believe in being sales-led or marketing-led and we thought that's not our strength and we wouldn't do a good job even if we did. And we got a lot of advice in the early days telling us this is cool technology but there's so much money to be made in like a particular vertical, go very focused on a vertical, sell like this enterprise solution, go charge 20X for the product. And then our thing was, well, we're very much focused on building a product that everyone can use inside a workplace. We want to democratize AI. And I learned from like the best of the best like Atlassian had the same thing, Dropbox had the same thing. We're going to first start with the individual inside the org, make them really happy. Pigma, another great example of they made designers extremely happy and then organizations picked it up. So there are other benefits to it where low cost of acquisition, virality, where you don't have to spend money on traditional channels like marketing on like Google and Facebook ads and stuff. But PLG is very, very difficult because the amount of upfront investment you have to put into engineering and product and building a self-service function because that takes a lot longer. And why would someone waste their time trying to build a product where someone goes in and buys like a $10 product? That's not going to scale. It's much easier to have a sales person go and like quote $10,000, right? That's the logic and feedback advice people would give us and we would push back and say, no, I would rather build a $10 product that everyone can use and go for high volume. So I think this is again, being sort of inexperienced helped in this regard because we were just naive enough to believe that this is the vision we would want to build. And now it's cool. Everyone's talking about product like growth and virality and like how you need to be more like a Zoom or Figma, but we did that in an era where they're like, this takes too long to get to revenue, don't do this. So yeah, sometimes the general consensus isn't always right. Yeah, I think being a contrarian and being right, it's a beautiful thing, it's hard. And I think also what you mentioned, like having the flexibility or the awareness to recognize when you are not right and make important changes, it also gives you more chances to stay alive until you hit that big swing as clearly you did. And it's just one of many as I learn about your vision. Since most of our audience are product leaders, I have to ask you about the AI skill in product. We've seen the news, right? How this article saying, oh, Netflix is hiding AI product managers and paying them up to a million dollars and how a lot of the jobs out there now require some sort of understanding of AI. So how do you see the need to learn AI from a product management perspective? Well, it's cool to say that AI is a unique skill set and you need a special type of PM for that. For me, to be honest, is it's the same as how you look at any sort of function and how do you apply AI? So I don't like to look at AI for the sake of AI and how you build a product around AI because what you end up having is just wrappers and wrappers around a cool technology like GPT. So there's so many companies that are coming out of YC who say, well, AI is the hot thing. So let's go just use that technology and build some use case. Again, same problem we had back then where we said there's some cool technology, let's go build some product or force some use case around it. And then that ends up creating a shallow product. We made some of those mistakes in the past. And so I would rather not have PMs think from that lens and instead think about, this is a problem we're trying to solve for customers. How can AI help automate or ease some of those workflows? And how can I sprinkle AI on top of the core platform? So when you think about it from that lens, you look at it as a feature, a function, a layer that you can implement deeply into the product rather than a bell or a whistle that you just throw on or like a little hack that you turn into a product. So I think that the PMs that are working with AI should understand what are its upper bound limits? What are its lower bound limits? Where does it get it right? Cause it's not like black or white with AI, like it's not always gonna get the answers right with LLMs. And what are the core areas that you can reliably replicate where it's actually valuable? Where it's not just like a cool like filter on a photo app but it's actually a fundamental value that it creates time savings or workflow automation. So yeah, I think PMs that are already good PMs will adapt very well to the AI way. I don't think you need to specifically go and say, how do I become an AI PM? But that I think you still have to have the right fundamentals and foundational skills. Thank you, Chris, for your time. It's been a pleasure to learn more about your own journey and also your investments and your good work with AI technology. Thank you so much. Yeah, thanks Carlos. Thanks for your time.