Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
How Krish Ramineni’s AI startup has revolutionised 200k workplaces globally (Fireflies.ai)
148 Plays9 months ago
Fireflies.ai is an AI powered app that is transforming meetings across the globe and is on a mission to derive actionable intelligence out of conversations.
It was 2016, Krish had got admission for a masters program at Cambridge University after completing a stint as a product manager at the tech giant Microsoft. In the few free months he had before he was to join the course, he created some products as fun experiments with his cofounder, and those few months ended up becoming the start of the journey to build Fireflies.
He realised that there was a lot of value in conversations, and ended up spending the next 8 years in building a product to truly help companies capture that value.
In this episode Krish talks about
- How to find Product Market Fit
- How to make Product Led Growth work for your business
- How to using pricing as a strategic lever for growth
Get notified about the latest releases and bonus content by subscribing to our newsletter at www.founderthesis.com
Recommended
Transcript
Introduction to Fireflies.ai and Venture Capital Trends
00:00:00
Speaker
Hi, everyone. I'm Chris Ramanini, co-founder and CEO at Fireflies.ai.
00:00:17
Speaker
Venture capital investing has a flavor of the season kind of an approach. There was a time when cryptocurrency was all the rage and a lot of VCs were investing in crypto projects. During the pandemic, it was a tick. And right now, clear of the season pretty obviously is the relative AI based house.
00:00:38
Speaker
In this episode of the Foundry Thesis podcast, I'm talking to Krish Dhammedeni, who runs a generative AI-based SaaS startup, which he started eight years back, much before an LLM was even like something which existed inside the lab. And over the last eight years, Krish has wished up my advice as an extremely accurate meeting assistant.
00:01:02
Speaker
Firefly adds an agent to your meeting who does stuff like taking notes, making actionable points, creating insights out of meetings, creating summaries, and distributing it to everybody who attended the meeting, thereby allowing businesses to really derive intelligence out of meetings and there is so much more which can happen once you have this base of being able to capture the meeting information and data and transcripts.
Finding Product-Market Fit: User Love and Development
00:01:30
Speaker
This conversation is a masterclass in finding product that fit and using ELG or Prickland North to scale your startup. I'm Akshay Dutt and you're listening to the Foundry thesis products. All right.
00:01:53
Speaker
So let's start with your original story. You know, what was it like growing up in the U.S.? What led you to becoming an entrepreneur? For me, I grew up in San Francisco, California. I moved from India when I was about five years old. Having been in Silicon Valley, all I got to see was other tech startups and other family members that were in tech. So that was the only world that I knew. What did your experience do?
00:02:21
Speaker
My dad was in tech. Yes, he works at Oracle. And I've had cousins, aunts, uncles, or other relatives who were also in tech. So it was a very interesting upbringing in that regard.
00:02:36
Speaker
Got it. One of the things that I felt was that I didn't want to go into tech or anything to do with tech. I thought I wanted to be a doctor. So when I went to school, that was my initial goals. But very quickly, I realized I wanted to do entrepreneurship, work in business. So it was some weird way of getting pulled back into the tech world. I also wanted to be back on the West Coast after going to school.
00:03:05
Speaker
near Philadelphia and spending time there. And like, did you work before starting off on your own or, you know, what led to becoming an entrepreneur?
Lessons from Microsoft: Data-Driven Decisions and Culture
00:03:18
Speaker
Yeah. So prior to starting Fireflies, I was actually at Microsoft. I was a product manager there. I started when I was about 20, 21 years old. So relatively early. Yeah. Amazing. And what product were you handling there?
00:03:35
Speaker
So there we were working a lot with Microsoft Office and my specific team was responsible for the growth engine for Office. So all the A-B experimentation, data, metrics, and so any feature that was getting shipped from
00:03:51
Speaker
Microsoft for office would go through our team and we would measure the metrics. And this was something that happened for the first time in 30 years at Microsoft because you were shifting away from shipping CDs with software to digital software and everyone wanted to be more data informed and understand metrics. So this was a huge new cultural shift and I was really excited to be a part of that while at Microsoft. That must have been amazing learning for a 21 year old.
00:04:20
Speaker
Yeah, as someone that was just starting off their career and wanting to work in a growth sector, your initial opinion is, oh, it's office. It's been there forever. What innovation can you bring there? And I was very surprised with the rate at which you can shift, the cultural changes that happened as well. So all in all, I think we were in this precipice of a very exciting time at the company.
00:04:47
Speaker
and I personally learned a lot. I got to work with a lot of other folks, but more so than anything else, I got to learn about the culture of how it's like to work in a big tech firm and what are some of the inefficiencies that you can solve. Okay, so how did the idea of Fireflies come to you?
Origin Story: From Meeting Challenges to Fireflies Innovation
00:05:05
Speaker
For us, fireflies was a culmination of a lot of different experiments, but in reality, the true inspiration for it was, look, I spent so much time in meetings. Why is it that I can remember an email I sent two years ago, but I can't remember a conversation I had two hours ago? And so that started the quest for wanting to work on fireflies.
00:05:29
Speaker
Um, but we started working on a various set of problems, build Chrome extensions, Slack bots, other tools. And the general theme was there is knowledge buried inside conversations. It just so happened we were focusing on emails and text messaging, uh, type products and bots. Uh, but we realized that, Hey, meetings is the blue ocean. This is something that I would actually use for myself. Uh, and then ended up building for them.
00:05:56
Speaker
Well, did you quit and start these experiments or these were happening on the site? So I actually left Microsoft because I got a full-time master's program at Cambridge in the UK. So I've always wanted to study in the UK and London. So that was a great opportunity. I had about a month of free time. I went to MIT where one of my friends was graduating.
00:06:24
Speaker
We got together and we said, hey, let's start working on some projects. We have a month. Let's just try a bunch of different things. And the intention was not necessarily to go full time, but we had so much fun working on all these projects. By the end of the summer, I decided to say no to my master's program and go move to San Francisco and start this full time with my friend.
00:06:50
Speaker
Well, so you must have had data inform you that which idea or which experiment is working, which is not. Tell me about the journey of discovering that this is what is getting tractioned and we should zoom in on this particular idea of meetings.
Gaining Traction: Engaged Users and Feedback Loops
00:07:10
Speaker
The reality is the first couple years of fireflies was like walking through a desert without a map.
00:07:17
Speaker
because you don't know what's going to work or what's not going to work. Ultimately, the most important signal for any entrepreneur or founder is, are you building something that people love? But what is love? And how do you know that you're onto something?
00:07:33
Speaker
So we started with several different projects, but what we found was the adoption and uptick in usage was not always that strong. We spent all this time launching something only for it to spike up and then like go back down and things would normalize to essentially nothing. You don't need fancy metrics or data.
00:07:54
Speaker
Is it that you have 10, maybe 20 or 100 people that absolutely love your product and can't live without it? I think that's the best barometer to measure your success around. And when you find that critical mass of those 100 people, you can then find the next 1000 and then the next 10,000 and so forth.
00:08:15
Speaker
A lot of times what people do is they try to force acquisition of customers before they found something that is truly magical or something that people would truly love to use.
00:08:26
Speaker
wow that's an amazing insight Seth Godin has this thing he calls 1000 true fans like if you can find 1000 true fans you've built a business amazing so yeah I think that's very much true it was the same thing that companies like Airbnb had to follow
00:08:46
Speaker
The best products in the world started with a few people that really used it and then started talking about it and sharing it with others. Things that go viral don't go viral because of some social media posts or some sort of ad, but the real products that go viral are when other people talk about it and like to use it and want to share it with their friends. Amazing. Tell me the kind of products you tried which didn't work out. What was the product journey like?
00:09:15
Speaker
Yeah, one of the projects we worked on was a Chrome extension that would track all the messages I'm sending across different platforms. And if I ever promised to do something like, Hey, I'll send you the notes next week, or I'll forward you the report by the end of the day, it would create an automatic to-do list for me and say, Hey, these are all the things that you need to do. These are tasks that you've said.
00:09:37
Speaker
It was a cool idea at the time. This was pre-open in AI, pre-LLMs, pre all of this generative AI magic. So we had to build a lot of this stuff ourselves. So while the technology wasn't always perfect, the concept and idea was really good. And maybe we were a little too early for our times at that point in time. So something that we've incorporated some lessons back into Fireflies and maybe we'll bring it back.
00:10:03
Speaker
But again, as you can tell, the theme was around conversations. The theme was around action items and tracking and project management. It was a really cool idea, but it was a little too unclear if it was a consumer product, a prosumer product, a B2B product. We got various pieces of advice. Someone said, this is cool, but you can't build an entire company around it. So you go build another project management system like Asana or Trello. And that's not something we wanted to do at that time.
00:10:31
Speaker
So that's just one example. We also built like a Slack bot that would send you reminders, help you do daily standups and check-ins. So a lot of different things that we did.
00:10:44
Speaker
which I think if we maybe spent a little bit more time refining and getting to the core truth, we would have built ourselves something that people would have loved to use. But I just feel like the pull towards meetings was so massive at the time, because we were having a lot of meetings. We were trying to take a lot of notes. And we had to ask ourselves, where do I spend so much of my time as I'm going through this process? And it was just a natural calling for us, and we decided to go down that road.
00:11:13
Speaker
When you think like this, this was like the pre LLM era in which you were trying to build. So you would have had to hard code that if the message says, I will the verb, then make this into a task. You would have done something like that.
00:11:27
Speaker
We were actually using machine learning and natural language processing. So we were still using algorithms. They were just a little bit more root of entry. They're a little bit more fixed. But again, just using keyword matching would not work. Simple rules will not work. So there were still a lot of research that's been done around NLP and BERT and all of this sort of technology.
00:11:49
Speaker
So I think that it was a hard problem, but we had to build classifiers, we had to build a lot of things that we learned from doing. And I think it helped us build a really good discipline and muscle for when we decided to go work on firefly.
00:12:05
Speaker
So what, what do you mean by classifiers? And what was the kind of tech stuff that you were building? Help me understand. Yeah, we were using BIRB. You give it a bunch of examples and say, Hey, if it's like this, then make sure that it's a task. If it's like this, it's a false positive or false negative. So you're still building models at the end of the day. I just think that with open AI and LLMs, a lot of that work.
00:12:27
Speaker
people built entire companies around just got thrown out the window like you don't need to do sentiment analysis in that rudimentary way anymore or classification these llms with the push of an api call gives you that sort of power and ability and i think that's what's so powerful is the way that companies were built pre-llms and openai had to change
00:12:51
Speaker
There were some of these developer tools that were built to help you train your data, manage your data, build these classifiers, use these like models. And all of a sudden, OpenAI comes out, you don't need that much data to train some of this stuff or do some of these functions. And many of these functions come out of the box. Now you're equipping an ordinary developer with the skills of what historically took a PhD to two.
00:13:15
Speaker
So I think OpenAI cannibalized an entire market of not just end user tools, but also like developer tools and developer companies that are no longer needed anymore. Wow. Interesting.
Technical Evolution: From Telephony to Bots
00:13:28
Speaker
Okay. So when you started the meeting product, how did you like transcribe the, because essentially it's an audio that you're capturing, right? What they're speaking, that you capture that. So how did you build that product?
00:13:43
Speaker
When we started building that, back then we were using tools like Twilio, which are very expensive. You're dialing into those calls as a participant and you're paying per minute for every minute of telephony that you're doing. Very expensive process. So that was the first time we started building that. We'd then pull out the audio, we'd clean it up and we would transcribe it. We would use our own internal engine or we would use some off the shelf stuff like Google. So it was,
00:14:11
Speaker
all about just hacking something together to get it out to market and testing people. And then we realized, wow, this has a fall short on so many levels. It's also very expensive. Like we can't continue to afford to pay per minute. And so we learned about like better ways to do it. And we also essentially pioneered this whole new technology around
00:14:35
Speaker
having a bot join your meetings with the name Fireflies. So it actually is a physical like VM that gets spun up in the cloud. And then that bot will dial into your Zoom meeting or Google Meet meeting or WebEx meeting. And we started scaling that out and no one thought like you could have millions of Fireflies note taker bots that can go and join all of these conversations and do all of this capture.
00:15:02
Speaker
But we were able to architect it, we're able to scale it. It's not an easy problem to solve. And I think a lot of other companies started looking at that and saying, hey, that's probably the way we want to go about it. And then they started emulating some of the things that Fireflies does. I think it's okay to get started when you're having a little bit of volume, but try doing that for millions of conversations.
00:15:26
Speaker
It requires insane R&D and a really strong infrastructure focus. So I always like to tell people that Firefly, as much as it's an AI company, it's also an RPA company, robotic process automation, because we have all of this RPA technology that's able to navigate and go into meetings and do all these functions, pull the call, detect the speakers. It's a really complicated process.
00:15:51
Speaker
Wow, fascinating. So, version 1 Twilio. Twilio is like a cloud telephony service. So, people would call into a number which would essentially connect all the participants and because it's cloud telephony, so the audio would get recorded and that you would transcribe and make it available to the subscriber. Did this find you your 100 true fans, this version 1, the Twilio one?
00:16:14
Speaker
At the time, we had given the product away for free. And so we would have intermittent usage here and there. But what I like to say is that you won't really find your true fans till you start charging for a service. Anyone will try it and give you feedback.
00:16:29
Speaker
But it's only that person that's putting their money down, right? Your mouth goes to where your money talks. So I really believe that like when you start charging for it, that's when you will know the real stamina of what you're building, right? So if
00:16:45
Speaker
A movie's out in theaters and it's free. I'm sure a lot of people will come to the theaters and watch it. Like what you're giving up there is time. But then when you have movie tickets and prices, how many are willing to pay to go there versus saying, nah, I'm going to just wait till it comes out on OTT, right? So in the same manner, like building products is so much similar to releasing, you know, these sort of things out into the market and getting a litmus test of what people like and don't like.
00:17:13
Speaker
The disadvantage with a movie, I use that as an analogy, is that once it's out, you can't really tinker with it or tweak it. It is what it is. But with a product, you have opportunities to save it, improve it, make it better, and look at what parts of that product experience has frictions and how you can eliminate that type of friction and help customers get to that magic moment with your product.
00:17:38
Speaker
As much as it is a science, it's also an art to understand what your customers want. There's a lot of qualitative understanding that needs to go on. And then you get into this like science experiment, data testing iteration mode. I remember that Google had like a team that tested 50 different shades of blue for a button, right? So that they can see how much ads people will actually click on. And you can do that at scale when you have millions of users where that small amount of change can have a huge impact.
00:18:08
Speaker
But when you're a startup, you have to take as much of the data as much as your gut intuition and what you're seeing qualitatively from what customers are saying. So there is no shortcut to that, but you need to have both that science as well as the arts. What do you mean by the magic moment? So you said you have to remove friction to help your customer discover the magic moment.
00:18:31
Speaker
In Fireflies' case, the magic moment was right after a meeting when I got the notes and I see it and I'm like, wow, this is an accurate transcript I can search through and these notes are pretty good. And that was a vision that took two to three years of building towards. Not only like, did we promise that at our seed, we were building towards that at Series A. It was only really, I would say in the last one year where the technology met the expectations of the market.
00:18:57
Speaker
where people can say, Oh, wow, fireflies actually takes better notes for me than me. And that is a huge thing. Like to say, like, it can take notes as good as me or better than me is a huge step up versus, you know, you just putting out a few bullet points and saying, Hey, here were some keywords that were talked about in the call.
00:19:14
Speaker
Now, fireflies actually paraphrases the notes, you see it, you get them within five minutes after the meeting, your participants get them too, and they are like, wow, this is actually what we talked about. This is a really good reflection of what we spoke about during the meeting, and these are the important parts, and these are the action items. So that aha moment was how do we get people to get there? Now, there's going to be many pieces of friction, like how do you get them to set up their calendar? How do you get them to invite fireflies to their meetings?
00:19:41
Speaker
How do you get them to view their recap? So you have to eliminate friction along the way. And that's where the science experiment is. But ultimately, you can do all of that friction removal. But if your magic moment doesn't pack a punch, it doesn't matter. Amazing. Just take me through the steps. Like what you said last year is when you finally reached there, when you were able to deliver the magic moment, what were the steps you took to reach there? What had to be fixed along the way?
00:20:11
Speaker
So for one thing, we set our ego aside and said, look, we need to adopt LLMs. This is the way of the future. And we were fortunate to be working with OpenAI well in advance, well before the launch of Chad GPT. So our investors at Coastal Ventures were also investors in OpenAI. So got a chance to get connected to Sam Altman, got a chance to be put in touch with their resources and their team.
00:20:36
Speaker
So we were able to start experimenting with it and eventually we had to make the shift and it's a risky ship but you have to make that shift and so we make the ship. We made our AI summaries and notes and action items all powered by these LLM capabilities. We threw a lot of like other work out the window and
00:20:57
Speaker
Over time, we realized the quality increasingly improved. And we were able to tweak and fine tune it, change the prompts, do iterations, all of those sort of things. And then we launched this thing called Super Summaries earlier this year, which is a comprehensive set of meeting summaries that has both an overview, shorthand bullet point notes, a meeting outline with time stamps, as well as action items. And this was from learning from those millions of conversations and pieces of qualitative feedback
00:21:26
Speaker
What are notes that people actually like and want to see. And I've come to the realization that no set of notes are going to be perfect and everyone has a personal opinion on what notes look like. So we said we were going to try to give them the best we possibly can.
00:21:43
Speaker
with the extent of the technology today, whether it's GPT-3, whether it's GPT-4, whatever it is, we're going to go all out. We're not going to compromise on quality, even if it's a little bit more expensive. And people see that, and they start talking about it. They start sharing the notes. And they are like, this is one of the most powerful use cases of using LLMs and OpenAI. So I think that was a testament to us just being very maniacally focused on releasing that.
00:22:12
Speaker
And over time, we went a step further and said, how about we get Fireflies to write notes the way you want to write notes. And again, not all meetings are the same too. Notes you want for a sales meeting are very different from immediate notes you'd want for a recruiting call versus notes you'd want for a podcast like this.
00:22:31
Speaker
And we started giving users the ability, the tools, to build their own custom notes, templates, and custom apps. And that's something that we secretly, quietly rolled out over the last couple months. And we've seen incredible adoption of that. And so it's just like I said, right? From the day one to now, it's about iterating on that
00:22:54
Speaker
aha moment, that kernel of magic. And it's a never ending process. So we're maybe light years ahead today in terms of the quality of the notes and summaries than we were maybe two, three years ago. In fact, two, three years ago, we didn't even think this was feasible with the technology that was out there.
Competitive Edge: Execution and Platform Evolution
00:23:11
Speaker
But now we're at a place where like, how do you keep pressing, pushing forward, keep pushing the envelope forward so that you get to a place where this is magic. Like we want it to be better than what a human can do and be as detailed and as crisp as possible. So yeah, just the rate of change has just been dramatic over the last 15 months. Okay, interesting. So
00:23:34
Speaker
See the fact that TAT GPT OpenAI allowed you to create these very, very precise transcriptions of conversations and very relevant notes and summaries. So that means that this is no longer a mode, right? This is available on tap for any other startup. So what's your mode now? The fact that you want to transcribe and summarize is no longer a mode.
00:24:01
Speaker
Am I right? We built for a world where summarization and transcription is a commodity. And I think that a company that starts today versus a company that started three years ago, that can't be the reason you start the company. Also, if you are starting that and you want to compete with the likes of Fireflies, that's like a very difficult challenge, I would say. What I would also
00:24:27
Speaker
Say it's just because that it is possible means like it's one thing to be able to execute really well on it. I think execution is a huge piece of the puzzle. The other thing is that if you want to build for this sort of idealistic future,
00:24:42
Speaker
I do believe the companies that started a few years ago and built out the proper infrastructure, proper SaaS platform, the billing, the payment systems, the user experience, all of that, it's not going to go to waste because you need to have done that right versus just launching with an interesting hook. I don't think necessarily turns it into like the biggest platform play you can.
00:25:07
Speaker
So for us, Fireflies is evolving from just an AI note taker, just an AI meeting assistant to an overall work assistant. And we're turning this into a platform play. We're gonna have hundreds of apps on top of Fireflies that help you during meetings. So the AI note taker that you see is just one of hundreds of different applications. So I can have an app that will help do call coaching for me when I'm doing an interview.
00:25:32
Speaker
or an app that will pull out information and score that candidate or help me understand the objections of a customer and then send that immediately into Slack for my product management team. So we're building a workflow engine platform. It's similar to how you can say like Slack building a really beautiful chat experience is no longer a mode if everyone can see how that's done. But then they turn that into a platform play
00:26:00
Speaker
where you have all of these integrations, all of these bots, all of these workflows, and it's where work gets done. And Fireflies has evolved into that and we will continue to do more work on that so that it is a platform where all of your organization's knowledge is searchable and accessible in one central place.
00:26:19
Speaker
So, you're saying one more is the fact that you have built it as a SaaS platform and billing and user experience and you already have a certain amount of user base and this is inherently I'm assuming it's like there's inherent virality because
00:26:35
Speaker
Every time I attend a meeting, I will see that there is a participant in that meeting called Fireflies. So I will be curious. And I'm guessing that even if I'm a guest in the meeting, I'm not the host, I will still get the summary. So all those viral hacks are there, which is like your moat. And the other thing that you're saying is your moat is the fact that you are now creating this into a platform.
00:27:02
Speaker
When
Transition to AI App Store: Expanding Meeting Productivity
00:27:03
Speaker
you say creating a platform, are you talking of it in the sense that like an app store where other developers can put their apps or are you creating these apps in-house? So help me understand that vision. I'm not very clear.
00:27:16
Speaker
The Fireflies AI App Store is where you will be able to access hundreds of different apps to be able to extract different things from meetings. You may want an app that's going to pull out feature requests or objections or complaints or sentiment, create automatic to-do lists inside Asana, all of those sort of things, automatically fill out fields inside Salesforce.
00:27:37
Speaker
So we're going to launch a few flagship apps ourselves, and then we're going to give people the ability to build their own apps. And if they really want to also be able to publish them to the app store, share it with their teammates. So yeah, like the app store that you see with Apple or Slack or some of these other companies like the Google Play Store, we'll be having that. And I believe that the apps are going to get more and more
00:28:03
Speaker
powerful over time. And what's not to say that there will be an app on the Fireflies platform, whether we build it or someone else builds it, where Fireflies will actually go to that demo and do the presentation on your behalf, or go and interview the candidate on your behalf, ask all the right questions, gather all that information, bring it back. And so now you're getting into this world of agents, which a lot of people are talking about in the AI sphere,
00:28:29
Speaker
Where, how can fireflies be my teammate that can actually do work on my behalf. Right now it's already doing back office work like taking notes and filling out your CRM and other project management systems, taking those notes putting them into Google Docs, or sending updates to your team on Slack, which is what you're usually a secretary, AI secretary would do.
00:28:49
Speaker
But I think those agent-like capabilities are going to become more powerful as we see better technology, right? GPT-4, GPT-5, Gemini, and so forth. We will be in a really good space. Amazing.
00:29:07
Speaker
As of now, a user's interaction with Fireflies, the UX of Fireflies is not really there, right? I mean, they would probably do a one-time sign-up and set up the calendar and then after that,
00:29:20
Speaker
It is automatically reading your calendar. It is automatically attending meetings, sending you emails with all the summaries and the transcripts of the conversations, et cetera. So how do you get people to actually go onto the Fireflies app or website and discover other features and stuff like that? How do you make that happen? I'm guessing it's like a set it, forget it kind of a product as of now, right?
00:29:49
Speaker
The set it and forget it makes the ability to capture meetings very easy. But if you had 14 meetings today and you're like thinking about, oh, what did I just say to Krish earlier? Well, you're going to go to the Fireflies dashboard and you're going to search across all your meetings. And with one simple query, you're going to be able to find your answers.
00:30:08
Speaker
Or if you are looking for a specific meeting and you wanted to generate a report or some sort of a follow-up email, you can use our Ask Fred experience, which is like our chat GPT assistant, and talk to it and say, hey, can you turn this meeting into a thank you email? Can you turn this meeting into a product spec? Because we brainstormed all these great ideas, but I want it into a product spec. So it can do those sort of things for you.
00:30:32
Speaker
If you want to turn your meeting into shareable sound bites, like highlight reels, right? If you watch a cricket match, you'll get the highlight, the five minute, 10 minute highlight reel. You can ask Fireflies to create a highlight reel of clips that you can then go and share with your team so they don't have to review the entire meeting, but actually just go through those clips. So there's a lot of host meeting work people want to do, and there's so much knowledge there.
00:30:54
Speaker
My co-founder Sam would argue that my two years worth of meetings all centralized in one repository or dashboard is more valuable to him than any other source of knowledge inside the organization. Because if you have documents and wikis,
00:31:12
Speaker
they will get stale over time unless someone keeps updating. We all have like confluence or some sort of SharePoint type system in our org, Google Drive, but a lot of times new employees come in and say, hey, Chris, this hasn't been updated in like three months. Is this still relevant?
00:31:28
Speaker
But with Fireflies, it's a self-updating knowledge base. It's a self-updating knowledge graph because with every conversation, the knowledge is evolving, it's changing, it's updating. The decision I made six months ago on architecture may be different with the decisions I'm making right now.
00:31:43
Speaker
And it gives you that audit trail of how decisions were made, where disagreements happened, and what is the latest and greatest that the company is doing. So when we have employees that join Fireflies today, I don't have to do onboarding calls with them and repeat the same thing 100 times over. I say, hey, here's the five onboarding calls we've done in the past. Here's some best practices. Here's some sales calls. Here's some decisions we made around architecture. Review the Fireflies recaps.
00:32:13
Speaker
Within a couple hours, you'll be up to speed and have the context that we have. At the end of the day, for any organization, context is king. The reason why managers make the decisions they do is because they have somewhat more context than maybe other people on their team or their direct reports. But we're democratizing that so that you don't have to handhold an employee. When they're making those decisions, they have the same context that you do.
00:32:36
Speaker
So there's so much more to Fireflies than the note taker. So a lot of times people say the Fireflies note taker is the hook, it saves them immediate time, but the knowledge base is why they have to use it for an entire organization. Fascinating. How do you make this knowledge base available?
00:32:58
Speaker
in an automated way? Or does each person need to decide that, OK, I want this portion of this conversation to be shared? How does that happen? Does it automatically create a publicly available knowledge base? Because there would be conversations which would have something which I'll private and I don't necessarily want it to be shared with the organization. So how do you thread that line?
00:33:22
Speaker
As we have grown and the desire for enterprises to want to use Fireflies, we've spent a lot of time around meeting access controls, security, privacy settings, and so forth. So again, when you ask like, hey, aren't summaries commoditized? Sure. But then go try building an enterprise access control system. And you realize it's really, really insanely difficult.
00:33:46
Speaker
and you have to respect the privacy controls of every person on the team, and maybe the admin wants to override those privacy controls. We have a concept of channels, private channels, public channels, very similar to Slack, and then per meeting individual privacy settings as well. So all of those things will be adhered to when you search through the knowledge base. So even if certain pieces of information may have had the answers that you're looking for, but if that was made private by the admin,
00:34:15
Speaker
it shouldn't be surfaced to you, right? So you have to think several layers deeper in terms of that. But I think search is a very powerful function. Knowledge retrieval is a very powerful function. So before, I would just have to search by a phrase or some sort of keyword and I would get things like, hey, at six minutes into a call three weeks ago, this phrase came up or this competitor came up. But now we have this idea of with Ask Fred, you can also ask more meaningful questions across your meetings.
00:34:44
Speaker
And so if I wanted to ask a question about why did we lose this deal to last month, and were there other deals where we lost because of the same reasons, that is a very expensive, time-consuming query if you had hired a human to go find that. This is why people hire McKinsey's, BCGs, and the Baines of the world to go say, hey, go talk to my employees, talk to my customers, look at my processes, and give me answers.
00:35:13
Speaker
But Fireflies is now able to do that. It's able to retrieve that knowledge and take your natural language queries and go find answers to that. There were companies in the past that started where a lot of them usually in the enterprise space that were focused around meetings like for salespeople.
00:35:33
Speaker
And they used to come up with all of this sort of fancy analytics and they said, hey, if you talk about pricing seven minutes into a call, you're 20 times more likely to close the deal than if you talk about pricing six minutes into a call. In theory, no one actually finds that useful. No one actually adheres to that. Every company is different. That sort of analysis is static.
00:35:53
Speaker
So what we said is let's flip the script and no one wants that sort of archaic stale data and instead give them the truth. So the query I told you is like, why did we lose this deal? You're now backing it up with actual statements that the customer said versus why did we win this deal? Uh, you'll actually say, Hey, I really like fireflies because it's more affordable. It's 10 times cheaper than.
00:36:15
Speaker
say a Gong or another platform, and it doesn't matter when they said it or what part of the call they said it, but it could be really important. Same thing with what questions did my salesperson ask and which of those questions, I guess, created a negative response from the customer. It's a very qualitative question, hard to quantify with numbers, but if I can just take you exactly to those moments in the calls,
00:36:43
Speaker
That's what this sort of technology can get you to. So again, going back to that same theme about OpenAI, LLMs, knowledge retrieval, this is some of the magic that can be enabled today that wasn't even possible two years ago.
00:36:58
Speaker
So what you're saying is if, for example, a 20 headcount organization has signed up for Firefly, then there would be a certain amount of meeting transcripts accessible to all 20 based on the settings, based on how it's set up by the admin. And so all the employees can ask certain queries and if that query is available in the
00:37:23
Speaker
Uh, transcript which are publicly available, then they'll get an answer straight away and some conversations could be made private either by admin or by the user and those are not available when any other employees searching.
00:37:38
Speaker
Yeah. And then the more people inside your organization that use Fireflies, the better it gets. It's great as a single player product, but usually if you get your entire organization using it, then the value it delivers to both the individual employees, the management, the leadership.
00:37:55
Speaker
the different teams. It also helps break down cross department silos because sales has no idea what product is planning, product has no idea what customers are complaining about, and then marketing is pitching something completely random versus what their customers actually care about. So it aligns the voice of your customers, the voice of your candidates who you're recruiting, as well as like the voice of your colleagues into one central place. So those three C's form the voice of the company.
00:38:22
Speaker
Wow, amazing, amazing. So which means that this sounds a lot like how Slack did its growth early on. Slack's growth hack was that people would use it without really the corporate buy-in and eventually the company would realize that Slack is where all the data lies about what people are talking about and things like that and then the company would sign up. I'm guessing you had a similar strategy for getting employers to sign up.
00:38:52
Speaker
Yeah, what's interesting is if you think about the Slack acronym, it's something like searchable log of all conversational knowledge. So that's what Slack stands for. And they're doing great with messaging, right? But we felt that.
00:39:10
Speaker
the amount of knowledge that's generated in meetings, especially at an organization level, is more valuable, more expensive, and it's tedious. So I think Slack is phenomenal, I use it every day, but there's this entire other product category around meetings, and we're starting to build the conversational data around there. As well as telephony systems, we connect to like RingCentral, AirCall, all of these sort of systems,
00:39:36
Speaker
you have audio files, you can put them in. So yeah, there's there's amazing amounts of knowledge there. I think what one thing that Slack did really well was making it easy for everyone inside an organization to use it. It's a horizontal product. And in fact, the chief product officer at Slack is also an investor in Fireflies. So that's another thing that
00:39:57
Speaker
We know when I met with her and she decided to invest, I said, look, we want to build a horizontal platform. We want to democratize this AI note taker meeting assistant for everyone in an org. But I'm getting a lot of advice from folks that are saying, that's not feasible. You should only build this for salespeople and charge a lot of money. And she told me,
00:40:18
Speaker
No matter what, don't give up on your vision of bringing this to everyone inside an organization. That's really important. That will influence the way you price, the way you build the go to market function, the way you do self service, and ultimately,
00:40:34
Speaker
it's going to be more valuable. It's going to be slower to get off the ground than hiring a bunch of salespeople and going and selling a product that you're selling for $10 for $150, which is what the gongs of the world do. Instead, focus on creating value for everyone inside an organization. And we did take those lessons from Slack, for sure. And it was something that felt intuitive to me at that time, because I'm no salesperson. I don't know how top-down enterprise sales works. I didn't want to get into that.
00:41:04
Speaker
A world of trying to convince an executive and then forcing some software down someone else's throat. I wanted it to be organic. I wanted one person inside an org to use it, talk to their friends about it and then more people use it and then their managers like seems like you like it. Let's buy it. So that whole product led growth.
00:41:22
Speaker
thing also aligned with our horizontal strategy. So again,
Growth Strategy: Organic, Self-Service, and Minimal Sales
00:41:26
Speaker
as much as the technology needs to be better and execute well, you also need to think about how you go to market. The nice thing was that aligned also with my philosophy of sales and product and how you deliver value to people.
00:41:42
Speaker
I'm not someone that was comfortable trying to make a thousand cold calls every day and try to send a hundred cold emails and try to convince someone. I said, let's build something valuable, easy enough for someone to try. Both will have a free tier as well as some paid options. But if they find value, they can choose to upgrade and it reduced a lot of the friction. I think there are a lot of other founders out there
00:42:08
Speaker
who don't want to do sales. And it's not that I didn't want to do sales, it's just that the way that enterprise sales worked just didn't align with how I was as a person. So I know that enterprise sales is about not even showing the product, doing weeks of discovery, and then coming up with some crazy quote, some insane price for implementation,
00:42:31
Speaker
You as the end person don't even know how this is going to work. You spend six to 12 months on it, only to later realize that, hey, this was a total waste of time and a lot of unnecessary meetings. So I said that, hey, if we're going to build this product, the iteration cycle, the learning cycle has to be fast. Someone should be able to get value, not in months, not in weeks, not in hours, but it has to be in minutes.
00:42:53
Speaker
and let them make that decision and let them be the evangelists for us. So yeah, very relatable to how Slack disrupted the PLG market.
00:43:03
Speaker
Amazing, amazing. Help me understand that journey of acquiring the first 10,000 customers. You must have had to constantly tweak the onboarding journey to make PLG work. I mean, PLG is not something which happens on its own. You need to build your product in a certain way to make it work. So for founders who want to learn how to do PLG, can you share your own lessons?
00:43:32
Speaker
PLG takes way longer. PLG is way harder. PLG is also harder to monetize because you're charging a lot less. PLG requires more R&D than sales. PLG requires you to understand pricing and packaging really well. PLG needs to be more data oriented.
00:43:52
Speaker
You can skip all this if you're sales led where you have a pretty decent product. You don't have any onboarding or anything like that, but a salesperson can smooth over the edges and they can get people to that aha moment by hand holding them. It's just that it's not going to scale. You're going to need to hire hundreds of salespeople to do that.
00:44:11
Speaker
And in my mind, I do not want to have someone handholding a customer when a customer can help themselves. Today, we have salespeople for enterprises and stuff where even with all of the amazing self-service stuff we do, they still are a bit old school. They need someone to talk to. We'll have someone there for you.
00:44:30
Speaker
It's never say never right like that's the thing with that like the world of I'm not vilifying the sales process I think it's necessary could be a necessary evil but what I instead think about is sales should be in the act of service to people and helping them understand get educated. So when you're building a PLG company one figure out if your product is
00:44:52
Speaker
meant for that type of market? Is there a broad enough customer base where you can get lots of people? Are those customers people that can quickly try out your product? If I'm selling an enterprise security national defense project, yeah, I can't do PLG for that. If I'm trying to build some
00:45:11
Speaker
project product that is going to like help you catch like really serious threats and bugs and those sort of things harder to sell as a self-service product. You can try but ultimately a decision maker needs to make it before it even gets rolled out.
00:45:25
Speaker
So if you feel like your product has broad utility, it can service a lot of people, and you can go and get it to an individual inside an organization versus trying to have to deploy to the entire team, go ahead and try it, build for it. Make sure you're very crystal clear on your onboarding, crystal clear on how you're pricing, because you don't want to scare people away with pricing either.
00:45:50
Speaker
and get ready to do a lot of customer support. A lot of support tickets, a lot of opportunity to learn through those sort of things. And try not to force yourself to throw humans at that problem, even if it feels like that's the easiest thing to do. Build processes, build systems.
00:46:09
Speaker
So even after I've said all of this and you still want to do PLG, you have to be borderline crazy to want to believe that. And you have to have a strong, deep conviction in that. I feel like introverts are naturally going to be better at PLG, as funny as it sounds. Extroverts are naturally going to want to do sales. Yeah.
00:46:30
Speaker
What I would also caveat all this is with is we were still talking to customers. We were still learning from them. It's just that we chose not to sell to customers when we were having those conversations. We were learning, we were helping them, but we said, we're not going to sell to you if you want to buy the product. The pricing is on our website. It's transparent. I'm not here to negotiate or discount or any of that stuff. It's transparent. And we know that we have some of the best prices in the market.
00:46:55
Speaker
Buy it on its merit. Don't buy it because I'm a really smooth talker and I can convince you to buy it. Get rid of all of that subjectivity. Make it very objective. And that's the magic. Amazing. What does the journey look like? So let's say I'm working for an organization. I sign up for a free Fireflies version. It's a freemium product. They would be like a free version and a paid tier and so on. So what happens next?
00:47:23
Speaker
So when you starting up for the free product, you'll get a certain number of summaries, certain number of transcripts, and you'll hit some natural storage paywalls. So you can keep using it just for recording and capturing your meetings, but eventually you'll need to delete some of your old meetings and so forth.
00:47:39
Speaker
At that point in time, you've already realized the value and you said, okay, I'm going to pay for it. You're going to start inviting your teammates and then they're going to start using it. And there you go. Now you've created like the magical workspace that you need. We also have a free trial so that if you want to just get like unlimited access to the business tier for a week, you can do that and test out every single feature from analytics to integrations. And you can sign up that way too. So our whole philosophy is allow people to buy the way they want.
00:48:08
Speaker
And if you're someone that says, log, I need to buy a thousand seats and it's cool, I want to try it, but I need to go through a security review and I need someone to help me through that process, then you can request a demo and talk to sales. So the whole point is that you choose your journey and we'll help you and support you through it. We would prefer that you do self-service, but I also understand that for some organizations and certain deal sizes, that's not always going to be the case.
00:48:34
Speaker
How did your pricing evolve? What is the first year like and what does it go up to and how has that evolved over the years?
00:48:44
Speaker
Surprisingly, we've never changed our pricing from day one. I mean, we've added more tiers, but like the core product, we wanted to make it ridiculously affordable. We've gotten advice from people, especially during recessions and down markets, where the growing cost of supporting all of these free users, where they said, look, you should just raise your prices by a couple dollars.
00:49:07
Speaker
you should do that. You've created so much value in the product, and when you first started, it seems like a no-brainer to raise your price. We've refrained as much as possible from raising prices from the day one of the product, so you can get the Yearly Pro plan for $10 per month billed annually.
00:49:23
Speaker
And in the past, if you wanted to get two hours of meeting transcription at high quality, it would cost you that much. Now you get a one-year subscription for that rate. And then there's a monthly rate as well. And then we have the business plan where it comes with unlimited storage, analytics, API access. That one is $19 per month if you're buying it for the year versus $29 if you're buying it monthly. And that plan,
00:49:51
Speaker
A lot of people compare to, wow, there's other sales products where I have to pay $150 to $200 per month, and I have to lock into an annual plan. It makes no sense. So we're very competitive on pricing. And I'm sure that there will be people that say, you should raise prices. You should think about this. But we will try to go as long as possible, not raise prices, keep it stable. And in the event that we do have to raise prices, we will try to make it as minimal as possible.
00:50:21
Speaker
so that the value you're getting, you're getting 10x the value that you feel like it's a no-brainer. That's the key. All right. I'm guessing another source of pricing is going to be or another source of revenue is going to be the app store because all the apps there would be priced over and above the basic subscriptions.
00:50:43
Speaker
Right now, the app store is early, but the way we price for the utilization of some of the most intelligent features, the features that would cost a lot of compute or LLM capacity, we have this concept called AI credits. And depending on how complex the app is or the query that you want to do, if you want to just ask a simple question about how many seats did this customer want to buy for one meeting, that's one AI credit, versus if you are trying to
00:51:11
Speaker
hold together a list of all the feature requests from the last three months for your sales calls, that's going to require maybe 20 or 30 credits. So it's like, credits is pay as you go, and it's an add-on that you can buy so that you can unlock more intelligence. And if you want to use the most advanced model like GPT-4 or something with a big context window, then it's going to be more expensive.
00:51:36
Speaker
So the reason we've done it this way is that instead of trying to play this musical chairs around pricing and quality, we're saying, hey, you want the best quality and the most complex queries. This is what the utility of that question is going to cost versus if you want like simple answers, simple models, then we're going to give it to you almost at like base price.
00:52:01
Speaker
The goal around AI credits and AI utilization is so that people have the freedom of choice to pick what they want and how they want to use it. And we'll continue to work on that, continue to make the models better, make AI credits more affordable over time. But this is one way we can offset that whole discussion around
00:52:21
Speaker
Oh, someone's just about to do a really difficult query. Like, should we just prevent them from doing that? No, we want to let you can do whatever you want. Um, just know the quote for it or the price for it. But like the query I said, like, go analyze my last 50 calls and pull out feature requests. If I had to have a PM do it, that might cost me hundreds or thousands of dollars for the month. Whereas if I ask fireflies to do it, maybe it costs me like $2, right?
00:52:45
Speaker
So the margin of utility is so significant that we want to go with this pricing model so that you only pay for the quality of the queries that you're
00:52:58
Speaker
So this you're talking about the Ask Fred product, like Ask Fred is your product where it's like a personal chat GPT for the enterprise trained on their meeting data. Also for AI apps, it'll be for both Ask Fred and AI apps because AI apps will run on the same sort of underlying technology. It's just that it's more automated, right? Like Ask Fred could be that pull out the budget for this meeting, I go in and ask it.
00:53:22
Speaker
versus an AI app could be after every sales call, pull out the budget, go into Salesforce, fill that out, and then send me a report of how many deals we closed. Okay. And these AI apps, have you built any so far other than Ask Fred?
00:53:39
Speaker
Yeah, we've built custom notes for sales calls, discovery calls, board meetings, recruiting calls, BC meetings, podcasting. So we built like a bunch of custom notes apps. We built a app that can pull out magic sound bites for meetings, the highlight reels. So before people would have to go into a meeting, highlight a chunk of the transcript and say, turn it into a sound bite. Now I can just go and say, hey, create a sound bite on all of the complaints.
00:54:06
Speaker
and it'll go in and just chop it up and automatically edit those little sound bites into a highlight reel. We're going to have more apps to follow as well. Maybe I want an app that will look at a candidate's resume and then see what they're saying on the call and give me some sort of score if, hey, does what they say match with what they wrote on their resume or were they making stuff up? So these are the sort of things that I think would be interesting apps of the future. There's so many.
00:54:33
Speaker
I could think of an app where we're having a call and you ask about pricing and I share pricing, and then the app suggests to me, well, actually, you probably wanna recommend this product or this tier to them, right? So yeah, the apps are gonna be like creating whole new product categories with this stuff.
00:54:54
Speaker
Amazing, amazing. And all these apps are going to be priced through AI credits, like based on the complexity of the use kit. Some apps may be free, like that doesn't require AI. Some apps may require some AI credits. Some apps, depending on the complexity, may require more AI credits. So Ask Red is like a free app as of now.
00:55:13
Speaker
as Fred
AI Credits: Flexible Pricing for App Usage
00:55:14
Speaker
uses AI credits, um, and then you basically purchase some amount of AI credits to start with. Uh, it's nominal like $5 and you get like 50 credits. Uh, that usually supports most of the queries people do. Uh, and then it scales. So, you know, why change your pricing philosophy? I mean, with the core product, you're not,
00:55:40
Speaker
charging them based on how many meetings or how long the meetings were. People don't need to think of how many minutes of conversation. They just pay a flat monthly fees and it removes the friction. So why change that philosophy for the AI apps? We've done a lot of optimization on the transcription.
00:55:58
Speaker
And we've done a lot of work in making it affordable so people don't even have to think about it. Even to this day, there are many platforms out there that limit the amount of transcription, amount of things that you can do. So I think we've done work there. And anytime we do work, we want to share the benefits directly back to the customers. In the world of LLMs, it's still very new. And candidly, the technology is getting better, and intelligence will get cheaper. But we're still very early on that technology curve.
00:56:28
Speaker
So.
00:56:29
Speaker
The question that people have to ask themselves is, do we refrain from giving them that technology? Or do we give them that technology, but also know that, hey, some of this stuff is going to be having a utility-based price? So we said, let's give them the technology and get them to use it as quickly as possible without having to worry about the variability of cost. We didn't want to do the musical chairs and make something cheaper in quality just to lower the price. So we didn't want to do that. We said, you as a customer can pick.
00:56:59
Speaker
The other thing is apps have a lot more dynamic range. So with transcription, it's just the duration of the call and how much audio you're processing. But there could be simple apps and there could be complex apps. And the app utility will vary. And how do you price for that? There is no way to do pricing based on, oh, well, this app or this query is super simple versus this one is super complex.
00:57:27
Speaker
And that's a headache for customers to want to figure that out. So instead of limiting the power that they can have, we're giving them unlimited capacity to do whatever searches, whatever apps that they want, and just tie it directly down to the actual literal cost of the LLMs and then tie it to that. And now they know that we're not BSing them. They know what the cost of the
00:57:55
Speaker
cost of computers and what it comes down to for us and over time we're going to scale and it's going to be affordable and it's going to be better. But the same thing could be said maybe about like some of the other technologies over time where back in the day the cost of storage in the cloud was super expensive till Dropbox came around and made a lot of refinements and then over time it became more and more
00:58:18
Speaker
commoditized and affordable. So it's the same way. I like to look at companies like Zapier. They're fantastic examples of when you build zaps and workflows and you pay for the zaps, right, and the volume per month. Ultimately, you pay for utility. It's just much more transparent.
00:58:37
Speaker
I've also heard that GitHub, which has a co-pilot product, they charge $10 per user per month, but they're losing $20 per user per month. At Microsoft, they can afford to do that, but AI is really expensive.
00:58:53
Speaker
The short answer to the question is no one has figured out how to just give AI away for free because there is a real unit economic and cost to it. And most companies that used to build SaaS products in the past that did not understand that, this is something new for them. But we had to know about this from day one because we had to deal with this already for telephony, for transcription and summarization. So this is not new to us, but it's just now leveling the playing field where all these other people realize software is not free.
00:59:23
Speaker
Amazing, okay. I'm guessing your cost must have gone up significantly because you decided to get rid of what your own algorithms were and adopt OpenAI. OpenAI is costlier than what you would have in the house, I'm sure.
00:59:41
Speaker
Yeah, I mean, we plan to spend millions of dollars with LLM companies, whether it's OpenAI or others over the next couple of years. So here's the other thing, right? When you talk about competition and this sort of market, it does take a lot more startup resources and funding and investment to do it. Whether you're building your own models or using OpenAI or fine tuning your own LLMs, the reality is it takes data, it takes money, and it's costly.
01:00:08
Speaker
So we believe that as long as it has a net ROI of doing this, we're going to continue doing it. And if people are willing to pay for that value, then we'll continue to build towards that. But yeah, you're not going to be able to make money without spending money on infrastructure and these sort of things.
01:00:27
Speaker
I remember when we first started and we got our first bill, which was a couple hundred dollars for cloud to now, right? It's millions of dollars of negotiations that we have to do. It's a completely different ballgame. And you grow into it, you learn about it. And sometimes you have to pinch yourself and say, say, really? Wow, like million dollar like infrastructure bills. Are we really approving these budgets? But that's what it comes down to.
01:00:55
Speaker
Amazing, amazing. What kind of revenue will you close this year at or do you have like a revenue goal that you're chasing?
01:01:03
Speaker
Yeah, I mean, in terms of revenue, we don't share publicly the numbers. But what I can say is that we've been very capital efficient. We did our seed round in 2019, which was a $4.9 million seed. And then in 2021, we did our Series A, which was $14 million. And we've been able to scale, been able to build really good profit margins on the actual
01:01:29
Speaker
economics of the business. And we're in a place where we're looking at profitability and scaling with that. So for us, we always want to grow like a rapidly growing SaaS company. I think this 2023 was definitely a transformative year for us in terms of how we scaled. We 20xed in terms of meeting signups and all of those sort of things. And that obviously correlates into revenue as well.
01:01:56
Speaker
But yeah, it's a continual battle. But I think what's fortunate is that we can control the direction of the business and not be resource constrained to have to raise capital. Whether the market has a downturn, whether it comes back up, everyone knows AI companies are raising money pretty easily in this market. But other companies tend to struggle. But in general, to be a company where you don't need to raise capital is the best position to be in.
01:02:25
Speaker
And we're doing everything in our power to be there. So long-winded answer, but I think it was a good perspective of the numbers are pretty then 2023 has been very kind to us. So you don't need to raise more capital, you're on your way to becoming profitable? Yeah, that's amazing. Amazing. And could you give some indication of revenue like say by which year would you look to cross $50 million ARR?
01:02:54
Speaker
Yeah, I mean, it would be great. Like a lot of times that people look at IPOs and set like the hundred million ARR benchmark, realistically, that's a while away in order to get to that level of milestones. But I think that if we continue at the current trajectory we're doing,
01:03:13
Speaker
It's not too far, it's a couple years away if we continue to execute. Again, people also say the benchmark for going public is different. So some people say it's 100 million ARR. In India, for example, companies are going public with a 50 to 70 million ARR like gold, and other companies are at like 500 million ARR and still staying private.
01:03:35
Speaker
But I think for us, we want to continue to grow and scale this business as effectively and efficiently as possible. That's the key, right? Like efficiency. For a venture-backed business, having that discipline is rare. And I hear that all the time from investors. And yeah, I would say in the next couple years, who knows? Maybe if we're lucky, in the next four to five years, we can get to that holy grail, or maybe soon.
01:04:05
Speaker
I'm guessing you would be like between 10 and 50 million as of now. Yeah, somewhere north of that. Yeah. Okay. Got it. Interesting. Tell me what
Seed Funding Insights: Traction and Technical Showcases
01:04:15
Speaker
your fundraise journey was it a challenge to raise your seed round or, you know, what was some of the learnings from fundraising?
01:04:23
Speaker
Yeah, for me, what I've found is that people don't give you money when you ask for it. People give you money when they feel compelled by what you're doing. So the best place to be is doing something really well.
01:04:40
Speaker
and not having to raise and letting others preempt or wanting to invest in you and get excited and get behind you. Obviously you need money to get started. So show some sort of traction, whether that's building a product, whether it's showing some
01:04:56
Speaker
traction on R&B, but they're showing user growth metrics. Just having conversations and saying, I want to raise money. And if the VC asks you, what are you going to do if you don't raise that money? And your answer is, you don't know, then that's not good. So the best founders, the best entrepreneurs will try to make ends meet no matter what, like figure something out.
01:05:19
Speaker
But again, you don't have to have built a business with a million users before you go ask for capital. We were first-time founders, so we didn't have the opportunity to just go and say, hey, give us money. There were a lot of founders during the heyday of 2021 where they're like, hey, I'm an ex-Google PM, ex-Microsoft PM. I want to work on something in SaaS, give me money, and people were writing checks without even a deck.
01:05:43
Speaker
uh the world is different now and people want product people want traction people want progress so i do think that in the long run uh you need to be able to have a good understanding of what you're building and again it can only just be you and your co-founder working on building a prototype but making sure that like
01:06:07
Speaker
prototype works well. Better if you have customers, even better if you have some paying customers. I think that's the way to go about fundraising is every time an investor talks to you, they're going to compare you to your traction the last time they talk to you. Are you moving up and to the right? It all comes down to that execution.
01:06:27
Speaker
And from our own journey, we knew our seed investor two to three years before we actually raised money and officially asked for money. It was more of, hey, we're doing this experiment. It hasn't worked. Doing this experiment, this hasn't worked. Hey, we're finding some traction here. We're continuing to do this. Let's talk more about this. Is this something interesting? Or how would you go about it? Do you think it's a good time to raise capital? And then you start those conversations out. So sounding as cliche as it is, and as much as we need it,
01:06:58
Speaker
being needy is always the biggest turn off for investors. It's funny, but that's the reality. That's the truth. Yeah. Interesting. What kind of, what did you show when you raised your 5 million seed round? I mean, 5 million is a pretty significant number for a seed round. So yeah, what times were different. And I think it was different for us because we raised it directly from an institutional investor rather than just a bunch of angels.
01:07:26
Speaker
For us, we had a product that was working that was in market. We had a group of hundreds of beta users. We had a clear understanding of how we were going to go to market. We de-risked the technology stuff around transcription accuracy and cost.
01:07:42
Speaker
And reality was that, hey, if we can hire a few more people and scale this up, we can, in theory, get to a certain level of going to market and production. So we de-risked a lot of the technology risk to go to market. Risk was not 100% solved, but we at least had clarity and thought and conviction in what we were trying to do. And I think those were important. Because at that time, this technology was not trivial.
01:08:08
Speaker
That's very difficult. And they were like, wow, just these like two people have figured out so much on their own. I think it's something that we can.
01:08:16
Speaker
Wow. There were just two of you at that stage when you raised your seat. Two of us, and maybe we had some contractors here and there, but yeah, it was just a small team. Amazing. Amazing. Amazing. And had you started monetizing? Was it priced or it was like pre-used? We had in that beta a few people that were paying, but we actually said we're going to just try to get as many people to use it because investors at that time were not as worried about monetization. They were worried about, are people using this and is there a plan? Like, do you have traction and usage metrics?
01:08:47
Speaker
And for getting your paid like, say first 1000 users, did you use like paid marketing channels and all that? How did you? We've done zero paid marketing in our entire journey of Fireflies. It's just this year we're starting like now to experiment with a little bit of it just for fun, more than anything else. But most of our paid customers came through the viral loops and organic traffic. Wow. Amazing.
01:09:11
Speaker
So, but you would have needed to reach a certain minimum threshold for the viral loops to kick in, right? Like, or that Yeah, then that's hard because there have been many companies that came before us and after us trying to do the exact same thing and that they couldn't reach that level of virality. So that honestly, the viral loop took more than a year or two years for it to really compound. It's not easy.
01:09:37
Speaker
Okay, so you were just like, patiently chipping away on making the product better for those two years, you didn't try to get customers in through any other means, like, you were only wanting to rely on. We lived and died by organic traffic, and that can be pressure.
01:09:56
Speaker
We really didn't do any marketing. Everything was around the viral loops. And then if people came to us and said, hey, I want to learn more about this, we would obviously share it with them. Really started with our close community of friends and other coworkers and peers. And through that, it started growing and then people searching for it.
01:10:15
Speaker
We are in a true definition PLG at its most core element because we did not use anything to supercharge marketing in the early days.
01:10:27
Speaker
wow amazing and you've started to spend now what is the reason that is it to get more enterprise customers yeah we felt like we wanted to run a couple different experiments just to make fireflies have like multiple channels to explore we were also reaching that critical
01:10:47
Speaker
period in time where the brand is recognizable. And so the ROI of investing in brand marketing, even performance marketing can help us understand. Again, nothing is going to be as good as viral and word of mouth and organic, but these are just experiments that we would want to run to see like what else have we not tapped into. And we feel confident like that the product is at a mature enough stage where this could work.
01:11:14
Speaker
Amazing, amazing. And who's your closest competitor, like someone who would be looking very similar to you from the outside?
01:11:24
Speaker
I've mentioned a few of the companies like we don't directly compete with, but it just so happened we get a lot of their customers our way. One is like Gong and the other is Chorus because they sell basically fireflies, but for salespeople. And the price point is like $150 per seat versus the $10 for fireflies. So we tend to see that more and more as like, wow, we're getting a lot of customers from the Gongs of the world.
01:11:51
Speaker
There are some other platforms out there that have started, that have been working on it for some time. Like I would say Otter is something that comes up commonly in discussions. And with Otter, they started with a mobile app and we're more focused on prosumer users. When it comes to integrations, utility, and the depth use cases, businesses want to use fireflies. And so that's been a way we've been able to focus our energy around.
01:12:20
Speaker
basically spend much time on prosumers or consumer use cases. So I mean, if a podcaster or journalist wants to use fireflies, great, but our whole focus is on B2B and businesses. Okay, amazing. So any final advice for aspiring founders?
01:12:39
Speaker
I think that the world is changing so quickly. This is like
Advice for Founders: Embracing Generative AI Era
01:12:43
Speaker
the best time to be working on something related to AI. Even if the market conditions are crazy, we enter this golden era every like few decades. We entered this when the internet came out. We entered that era when mobile came out, cloud came out. And I think this is that other inflection point with
01:13:04
Speaker
all of the magic that's happening around generative AI. So if you have strong conviction around those technology tailwinds to go and pursue that and otherwise just keep being vigilant about understanding customer problems and what you can make better for people. Like that's probably the best thing about if you're starting a business is trying to understand a problem that you fall in love with. Don't fall in love with the solution, fall in love with the problem because the problem is the one that's going to make you money.
01:13:34
Speaker
And that brings us to the end of this conversation. I want to ask you for a favor now. Did you like listening to the show? I'd love to hear your feedback about it. Do you have your own startup ideas? I'd love to hear them. Do you have questions for any of the guests that you heard about in the show? I'd love to get your questions and pass them on to the guests. Write to me at adatthepodium.in. That's adatthepodium.in.