Journey of a Product to Enterprise Level and Building Trust with Big Enterprises

Host Victoria Guido interviews Krish Ramineni, founder and CEO of Fireflies, who provides insights into the evolution of Fireflies. This AI meeting assistant transcribes and summarizes meetings in multiple languages. He explains the rapid advancements in AI models that have allowed Fireflies to expand its language support and improve its transcription and note-taking capabilities. The conversation shifts to the broader AI industry, where Krish comments on democratizing AI technology, making it more accessible and practical for various applications beyond just number crunching. He emphasizes the importance of AI in enhancing productivity and enabling small teams to achieve large-scale impacts.

Victoria and Krish explore the future of work with AI, discussing the balance between job creation and replacement. Krish argues that rejecting AI is akin to dismissing essential technological advancements like email, highlighting businesses' need to adapt and embrace AI tools. They also touch on Fireflies' journey to enterprise-level adoption, addressing challenges like data security and compliance. Krish shares his optimism about AI's potential to augment human productivity and creativity, positioning AI as a transformative force that can empower individuals and organizations to achieve unprecedented efficiency and innovation.

• Fireflies
• Follow Fireflies on LinkedIn, Facebook, Instagram, YouTube, or X.
• Follow Krish Ramineni on LinkedIn.
• Follow thoughtbot on X or LinkedIn.

Transcript:

 AD:

We're excited to announce a new workshop series for helping you get that startup idea you have out of your head and into the world. It's called Vision to Value. Over a series of 90-minute working sessions, you'll work with a thoughtbot product strategist and a handful of other founders to start testing your idea in the market and make a plan for building an MVP.

Join for all seven of the weekly sessions, or pick and choose the ones that address your biggest challenge right now. Learn more and sign up at tbot.io/visionvalue.

VICTORIA: This is the Giant Robots Smashing Into Other Giant Robots podcast, where we explore the design, development, and business of great products. I'm your host, Victoria Guido, and with me today is Krish Ramineni, Founder and CEO of Fireflies. Krish, great to be here with you.

KRISH: It's great to be here. Thanks for having me.

VICTORIA: Wonderful. Before we dive into what Fireflies is all about and start talking about AI, just in your personal world, what are you learning right now?

KRISH: Well, I'm traveling this summer, and I've always wanted to speak multiple languages, both for functional reasons and to be able to actually enjoy my travel experiences. So, I'm trying to learn Spanish. I took three years in high school, but I forgot everything, and I'm trying to pick that up. I'm also trying to learn Hindi. We have teammates that are in both Latin America as well as India. And so, both of these would be really valuable [chuckles] to learn. I probably need to get a tutor, something that I'm working on right now.

VICTORIA: Yeah. How are you learning? Are you using an app? You said you might get a tutor.

KRISH: Yeah. I started looking at Duolingo. I started doing flashcards. There are online instructors. So, I'm just trying to learn the quickest way possible so that I can get just the basic, common phrases down that I could understand so that I can ask questions and understand what people are saying when they're giving me directions on the streets. So, that's the plan. I don't expect to be a fluent speaker. You know, I always wondered, too, like, since we work in this AI space, if we could build a tool that, in real-time, could translate what I'm saying into the local language and the local language into English using my voice. That would be pretty cool.

So, I think our whole mission is around like, eliminating communication barriers. But as I've been trying to learn new languages, this is something I realized is...it's a big world out there, and a lot of people in the U.S. only know one language, whereas people in other countries know multiple languages. And yeah, something that I didn't really appreciate growing up or being in high school. But now I'm realizing, like, the immense benefits of being able to speak multiple languages. So, I'm trying [laughs].

VICTORIA: Right. And I think the benefits even to your brain health and your way of thinking is really exciting. I also learned Spanish from a really young age and grew up with it. And recently, you know, I moved to San Diego, so I have a lot more exposure to people who are just speaking Spanish all the time and getting to overhear little bits of conversation, or at a restaurant, even though right now I usually end up ordering incorrectly [laughs] and getting a little bit of surprise. But it's, like, really sweet to be able to connect with people in the community at that level.

And last year, I went to Japan for a conference for Ruby, and I learned just a little bit of Japanese. And it just made me so happy, for some reason, to be able to say even a couple of words and a couple of phrases and to have other people, like, say that I was doing a good job [laughs]. You know, like, it's just really nice, especially if you're traveling a lot, and you want to actually connect to people to be able to share that language.

Yeah, it's interesting about AI translating there. I will say the translators that we had in Japan they may be caught about 60%. Like, you know, and then with context, it was quite difficult. So, yeah, I'd be curious how AI could address that and even get more personal and being able to use a voice and added more information into that, so you get that full translation.

KRISH: Yeah. Local languages and, like, the common phrases. So, for example, the Spanish that's spoken in Spain is going to have different phrases than the ones spoken in Mexico or in other places, right? So, that's also really interesting to think about how local dialects, accents all play into it.

Growing up, I used to love watching Bollywood Indian movies, and I would need subtitles. And I slowly started to get to a place where I can still understand what's going on without subtitles. It's really interesting that some of those jokes and some of the things that are said don't really translate exactly into English, right? Like, someone that's a native English speaker wouldn't quite get it. There's a lot more to language than just the words that are used. It's like the culture, the phrases, the people. And so, that's the beauty. That's the beauty of this world. There's so much diversity.

VICTORIA: So, I'm curious. As a founder of an AI app that takes recordings of people's meetings and turns it into summarized language, are the models based primarily on English, obviously, right away? And how are you thinking about incorporating other languages into your model?

KRISH: When we started, it was primarily English. Fireflies would take notes in English. It would transcribe English meetings. And then, this past year, we started support for 60 different languages, including Spanish, French, German, Hindi, and so many more. And on top of the transcription, we now can also do AI note-taking in some of these other languages. So, if you have a meeting in Portuguese, the summaries and notes will be in Portuguese.

We have a big global presence today with Fireflies. It's used in over a hundred countries and lots of different languages. And I would say the foreign language segment of our market is growing incredibly quickly. And we also hear requests from people where they have people that speak different languages because they have global teams in meetings. And it would be super helpful to be able to translate and transcribe and so that when they look back, they can get help and understand or clarify certain things.

Yeah. I think language when we started, and most of these LLMs (large language models) were primarily built around English, right? Especially transcription and speech. But there are companies coming out that are now building these models that give better representation to other languages. And we will have AI that will be able to understand and speak many different languages. And just the rate at which this technology is changing, I'm super impressed.

I read somewhere that they were building a model back in the day before the whole ChatGPT, where they were using reinforcement learning and transfer learning, where they were able to teach it one language. And it was able to quickly pick up another language, even though it wasn't taught to them. So, AI works in very magical ways [laughs].

VICTORIA: That's really cool. I wish that I worked that way with Portuguese because I was like, oh, I know Spanish okay sometimes. And then, I was like, but Portuguese when I read it, the words make sense, but then hearing it, the pronunciation being totally different, it's like [laughs] a long way to go. But that's really interesting. And you've already started to talk a little bit about the changes in the industry and what you're seeing as the new capabilities for AI. Can you tell me more about that? What other changes do you see in the industry in the last, like, year or even, like, a couple of months?

KRISH: At least in the last two years, people's perception of how hard it is to deploy AI has changed. Before, you needed to have a PhD. You needed to write a lot of code, and the AI was not practical. Now, AI is just a few lines of code. You don't even have to be technical to deploy AI. And you can ask it to do a lot more than crunching numbers, and that's what's so powerful. And we are getting these generalized models where, in the past, if you had, like, an AI model, it could do one thing like classification or sentiment analysis.

Right now, I have AI that can give me French poetry. It can generate images. It can summarize things. It can help me have a conversation with it and learn how to improve my speaking skills. AI is trained on the web, right? And whatever is on the web, it's a reflection of that. So, that also comes with the good and the bad. The good being that it knows what most humans feel and think and can relate to. And the bad, though, is there's a lot of nonsense on the web, so a lot of the bias, a lot of the information that it's getting.

AI today can, with confidence say the wrong answer and believe that that is the right answer. So, that is one of the risks. Some people call this hallucination, where the AI goes haywire and wonky. But I'm hoping that with time, that does get solved; we have better guardrails and parameters. Some people will say that hallucination is a feature and not a bug because it's letting the AI be more expressive. But everyone's understanding of truth should not be, I think, different. Like, I think there is one set of truth sometimes, and you don't want the AI to misinterpret that.

So, yeah, I think it's an exciting time. And more people like our company are embracing and adopting AI into their core products. And it's causing incredible productivity gains. But it's nowhere perfect. People talk about this AGI, (artificial general intelligence). I think we're a little bit away from that, but we're moving fast. Like, this stuff is happening at an exponential rate.

In technology, there was this Moore's law, right? With the number of transistors and how amazing and exponentially better the chips got. We saw that with storage, right? The cost of cloud storage when it first came out was so expensive. Now it's super cheap. If you remember, back in the day, you got, like, a USB card where it could probably store, like, 10 megabytes. Now it can do, like, 10 gigabytes to, like, one terabyte. And the cost is, like, super affordable.

If you think about TVs that came out in the past, right? Like, getting a 60-inch TV was super expensive. Like, a 40-inch TV was super expensive. Now everything is, like, LCD. You get, like, 60, 70 inches. And the price is the same as what a 40-inch TV was back then.

So, AI is all of that and some more. It's moving at a rapid pace. Like, technology, as an industry, like, it's moving so quickly, and AI is moving more quickly than what most people can keep up with. So, that has pros and cons. We can dive into that more. But, yeah, things are changing on a weekly basis, not on a yearly basis right now.

VICTORIA: Right. And there's a few directions we can go in from there, I think, that are really interesting, right? There's, like, the future of work with AI because I can relate to a feeling of fear and anxiety about what is this new technology? Am I going to lose my job? And when I talk about it with people I'm mentoring, I try to position it more as this is going to change the way we work. You're still going to need people to do stuff.

But if you're rejecting AI because you think it's just a fad or it's just silly, like, I think it is fundamentally changing the way people are going to do their jobs if you pursue that. And I think if you're capable with using AI as a tool, you're going to be more powerful than you've ever been in your job in most cases.

KRISH: Rejecting AI is like someone rejecting email for faxing and sending paper mail by hand. You just cannot compete, right? Imagine if you were a business that said, "I don't believe in AI. I'm going to do everything old school." You'd be like, today, okay, cool. You should do that. And imagine if you're a business today that says, "I don't use email. I will physically mail everything to you handwritten." So, that's what it's going to be like in a few months to a year. Like, this stuff is happening quick.

And I always like to say that AI will it create more jobs? Yes. Will AI replace jobs? Yes. But the probability of someone using AI who will replace you is far greater. So, AI isn't going to replace you as much as someone using AI is going to replace you. It's a skill set that we have to all learn, just like how we had to learn to use a computer, to use the internet, to use the smartphone. This is the same thing here. Like, we're going to all have to learn to use it and learn to interact and gel with AI in the workplace.

VICTORIA: Absolutely. And how does that relate to what you learned in your journey with Fireflies and talking to people about AI? How have those conversations gone forward?

KRISH: Fireflies at the core is this AI meeting assistant that joins your meetings. It takes notes. It helps you remember what was discussed before a meeting, during a meeting, and after a meeting. It helps me recall any information that I talked about. If we met six months ago and I'm meeting you again, it has the notes for me. It lets me search back through it. It lets me ask it questions about what you talked about. What were the next steps? What were the action items?

So, it's giving me structure to my life because a lot of my life is having meetings with lots of people and having many conversations, and then recalling those conversations and staying on top of that. It gives me structure in terms of what I do day in and day out.

I always believe that work originates from conversations. Meetings are some of the most valuable conversations that we tend to have. It's also very expensive for an organization to have meetings. Because when you get four people in a room who are all making six figures and spending an hour having meetings, that information, whatever is discussed, can have a huge cost to the business. But it can also have a huge potential to move the business in the right direction. So, organizing all of that knowledge that originates from meetings was the initial vision of Firefly.

Before all of this AI and ChatGPT hype, that was what we'd set out to do. The LLMs and AI help us do that job better: summarize the meetings better, generate better action items, create meeting outlines, allow you to search back. Instead of searching by keywords, you can now ask specific questions and talk to AI. So, this is what AI enables people to do, especially with Fireflies, is you can now interact with Fireflies like you would with a teammate, and that has changed the way people feel and use our product.

And people don't come out and say, "Hey, you're replacing secretaries. You're replacing the intern that I've hired to take notes for me. Like, you are replacing the job that the new hire has to do because it's a rite of passage." 95% of people will not make that argument because it's actually silly because your assistant, your interns, your new hires have better things to do. And these are the mundane, monotonous stuff that you should delegate to AI. Obviously, you can have humans review all of that, have their own, you know, take on it, generate reports. But it's actually leveling them up to be more productive and be more valuable to your organization.

So, I think there's a lot of pieces of AI that will do the same. You know, other technologies like, for example, AI that generates images or graphic designs that's not going to replace the graphic designer. It's going to allow the graphic designer to be able to create many iterations, be more creative. Like, if they don't have the technical skills to use certain pieces of software, it can help them ease that barrier to entry and give them more assets to work with. So, I think of AI in the workplace as how can we augment human productivity by giving each and every person a superpower?

VICTORIA: And you started this eight years ago now. So, you were really, like, ahead of the curve in terms of all these AI companies coming out. I'm wondering, what challenges did you have early on, and how did you overcome them?

KRISH: When we started, this was not obvious, like, that we should be doing this. It sounded obvious to us. We felt like every person in the workplace deserved an AI assistant that takes notes, not just the C-suite who, has a secretary or a business admin. And it felt like it's so obvious. It should exist. We should build it. And we need to create the experience like an assistant that follows you around.

But when we started, there were so many uncertainties. Can this technology work? Can this technology scale? Is the transcription going to be accurate? Can you actually even summarize things? And does that stuff make sense? It's a new behavior. Are people willing to entertain AI assistants and meeting assistants? So, every step of the way, there's a technology risk, a go-to-market risk. You are doing a sales risk. Like there are so many like pieces to the puzzle that you have to figure out. And you have to peel each layer of the onion and get to the core. So, I think it's been quite a journey.

We've been lucky in a few ways, right? Because I do believe that luck is sometimes about being at the right place at the right time. But those that always keep showing up are going to be able to get lucky from time to time, right? If you take a thousand shots, at least one of them will make it. That was my philosophy. We tried. We built seven or eight different products that all somehow worked or utterly flopped. And eventually, we got closer and closer and closer to the truth of what customers needed. And that led us to build the version of Fireflies that exists today.

So, it's definitely not easy, but there were three core phases to Fireflies or three core movements that allowed Fireflies to exist. One is speech recognition and transcription fundamentally got better. It got more accurate and more affordable. Before, it was ridiculously expensive. It would take a dollar per minute of transcription, and you needed humans to do it. But these AI engines, speech engines, got better.

The second thing is when we launched Fireflies, the pandemic happened a few weeks later. Everyone went remote. Video conferencing became more mainstream, and people were actually having Zoom fatigue and way too many meetings. And they needed a way to organize all those meetings they're having, jumping from one meeting to the next. And Fireflies got pulled forward, and a lot of people wanted to have it in meetings and help them around. And that helped us grow exponentially, virally. To this day, Fireflies has taken notes for over 16 million people across 300,000 organizations.

And since the launch in January 2020 to where we are, the first two to three years were trial and error, right? From 2016 to 2020. We built our product in 2018, 2019, launched in 2020. The pandemic accelerated the adoption. And then, you have this new LLM wave that comes out at the end of 2021, which allowed us to make the product fundamentally more valuable. And everything got better from the notes, to the summaries, to the search. Everything got better. And we crossed the chasm from where people thought, "Huh, this is a cool idea, but I don't think it's going to work," to "Holy crap, this is one of the best use cases for generative AI and LLMs."

And yeah, like, it was luck in terms of being there when this movement was happening. I think a lot of AI companies can say that. But it also took a little bit of fortitude to be able to be doing this several years before the stuff came out, right? Once a gold rush occurs, everyone's going to want to go in and then build something. But if you were already there, and you were searching and searching, and you were very close to something, and then you discover the gold rush, you're going to have a head start, and that's what happened with us.

VICTORIA: Yeah, you said 7 to 8 product iterations, and I was like, uh, you really had to go through an emotional roller coaster, I'm sure early days. But you were lucky enough to be in the right place at the right time and have a good picture of what the problem space was. It's really incredible to hear that.

MID-ROLL AD:

Now that you have funding, it's time to design, build, and ship the most impactful MVP that wows customers now and can scale in the future.

thoughtbot Liftoff brings you the most reliable cross-functional team of product experts to mitigate risk and set you up for long-term success.

As your trusted, experienced technical partner, we'll help launch your new product and guide you into a future-forward business that takes advantage of today's new technologies and agile best practices.

Make the right decisions for tomorrow today. Get in touch at thoughtbot.com/liftoff.

VICTORIA: Why don't I move forward a little bit into where you are now, where you have GDPR and SOC 2 compliance, and you're, you know, really doing well. Like, what were the challenges in getting that product to enterprise level?

KRISH: We let the customers pull us in the direction that needed for us to go. A lot of times, we try to see, okay, what is every friction point along the way? What would it take for larger organizations to adopt it? There is incredible product value. People have been saying it. But I need these sorts of features and capabilities in order to deploy it inside my organization. And we are handling meetings, which is sensitive. And so, we have to be able to give them more access controls, give them more admin features.

You know, we have a policy at Fireflies where we say, "We do not train on your data by default." So, most AI companies they're using their customer data to train models. We do not do that. So, we made that explicitly clear. CIOs love hearing that because when they look at us and other potential competitors, those competitors are bragging outside, saying, "Hey, we built this amazing model training on all of this data." And we say, "We don't do that." So, unless you want us to build something custom for you, we will not train on your data by default.

The other thing we said is, "Look, you own your data. If you want to delete it anytime, you can. You can request to have the data deleted. If you were a participant on a meeting and you don't feel comfortable with the data, you can request the host to delete it, or you can come to us, and we'll delete that data for you." Like, you have rights to that. And we put everything in a very customer-centric worldview, and that usually aligns with the big enterprises. That aligns well with a lot of the folks that want to use your service.

Because when you're using a new technology, the first question people are going to have is, "Does it work?" The second question they're going to have is, "Is it safe?" And with AI, a lot of people think about the safety of using the tech. And when you're building for a B2B enterprise, we had to make sure we put in the hard work to tailor our product to the needs of those customers.

VICTORIA: That's really interesting. And maybe you could say more about why would a company want to train a model on their own data and create an LLM like that?

KRISH: Are you talking about customers wanting us to train something for them?

VICTORIA: Yes. I've heard this idea from a few different people, actually, where they want to be able to build an LLM and build a model based on a company's own knowledge and their own information. So, maybe you could say more about that.

KRISH: I think it's really around fine-tuning and personalizing the AI. Now you can train on models. You can do fine-tuning. You can do other parameters. But it's really giving everyone their own personal experience with AI. We can do this today even without training just by understanding your preferences, and we want to continue to build towards that.

So, yeah, we believe that every person inside an organization should have their own personal AI note-taker, and no two meeting notes will be the same because each set of notes is unique to you, your meetings, what your team wants. And so, that, to us, is like a vision we try to build towards. AI can bring about insane level of personalization, and that's one of the reasons why people would want to train their own models based on their like, knowledge graph, and information.

VICTORIA: How do you think about the cost of building and running these AI tools from an infrastructure cost perspective? How does that translate to your cost for your customers, that kind of thing?

KRISH: AI is expensive. The unit economics...I think a lot of people are taking for granted that it is insanely expensive to run these models to use a cloud provider of these AI models. Some people are spinning up their own models. It is insanely expensive. But the good news is the cost is going down at an accelerated pace, and it's just up to whether the pace of the cost decrease will outweigh the amount of spending some of these startups are doing. And that's why some of these companies are raising tons of money as well because they don't really have a monetization strategy. They have no revenue. They're making lofty goals that "This AI is going to do this. It's going to do this. It's going to replace this function in your org."

But who's going to pay for it? How are you going to make people pay for it? Is it going to be subscription-based? Is it going to be utility-based? How much upfront cost is going to be there to train these models? And what if you do all that work, and then you deploy an LLM; you're an infrastructure provider, and no one cares? What if you're an application layer, and you're giving all of this stuff away for free and then eventually realize you can't get people to pay for it?

So, there are so many open questions for these companies where the technology is changing quickly. The cost is changing quickly, and consumer preferences are also changing quickly. We'll have to see. Only time will tell because there's a hundred companies out there, all raising a hundred million dollars. We know that all of them are not going to make it, a few are. So, it'll be interesting to see what happens once the dust settles. But I think people should take that very seriously because you can't always expect to be bailed out by investors if you don't know how to utilize AI and how to build for cost.

And I think a lot of investors tell startups to not worry about that. They say, "Don't worry about the cost. You know, as long as someone's there to pump you money, you just keep building, like, the best product out there." That works for some companies. I just don't believe it should be the only strategy that someone should take.

VICTORIA: What if you build it and no one cares? It'd be so heartbreaking [laughs], but it happens, yeah.

KRISH: That's 95% of startups that die is because no one cares.

VICTORIA: Right. Yeah. And I'm curious, like, what other use cases do you see as being the most relevant for AI? Like, what problems does it really solve very well? I mean, note-taking, obviously, one of them.

KRISH: I'm really excited about all of these AI tools that can write code for you. And maybe they can't replace a software engineer, but could you make a developer 10x more productive? And could today AI start off as a copilot for writing code for you to eventually building you full-fledged apps, right? And imagine what that would do in terms of reducing the barrier for so many people to be able to create their own personal apps and tools. Easier said than done.

But I think what's really working really well, whether it's with GitHub or some of these other AI tools, is, can it actually write code for you? And I think that's a wonderful use case. It'll still need a lot more fleshing out, but I am bullish on that use case for sure.

VICTORIA: Yeah. I'm hopeful that companies will figure out how to use AI to level up engineers because right now, we have the problem of the flattening of the middle where you have really senior people who are very in high demand. And then, you have a lot of people with very little experience who really want a career in technology. So, I see that as an opportunity, but also a risk that some people will create things with AI code and sell it. And it'll just be a hot mess [laughs]. But, you know, that's kind of the risk it is even if you're paying real developers at the same time, so...

KRISH: Yeah. I think AI will take a C player and make them a B player, maybe a B player into a B plus player. And then, it can take an A player and make them, like, A plus. So, I think it just levels the playing field a little bit, eventually to a point where everyone in the org is going to get a little bit more productive.

And I also think that small teams are going to be able to do incredible things. You, as a small team will be able to compete at a larger scale with some of the bigger companies. You know, Sam Altman said maybe there's a chance that a 10-person company is going to build a billion-dollar market cap organization that goes public. So, all of those are possibilities, too.

I love the idea of solopreneurs and people that run their own, like, small businesses, you know, three to four people, super lean. Obviously, I'm in a venture-backed world, so I can't necessarily run that, but I am very excited by that potential. And I like those types of people that are entrepreneurial and don't need a lot of CapEx in order to get started. AI will allow a lot more solopreneurs to thrive. If social media created a market for people to have, like, a full-time job as influencers, I think AI can create a market for people to have full-time jobs as creators of products, goods, and services that can be managed with just, like, a few people.

VICTORIA: That is really interesting. I'm curious if you want to...let's say you're meeting a founder or an entrepreneur, and they're AI-curious, but they don't really know where to get started or how to step their toe into the water. What advice would you give them?

KRISH: I think the best place to start is by building and building something for yourself that you yourself would use. Try all these different AI products that are out there. Look at what's trending in the news in terms of which...every week, some new model is being deployed, some new changes are being rolled out. Google is rolling something out. Facebook is rolling out something. OpenAI is rolling out something. So, try to keep pace. It's going to be tough.

And then, go play around and tinker with these tools. Like, you should be a tinkerer first. You should like to build things. You don't have to be an engineer to get started, but you need to be able to go and get your hands dirty, roll up your sleeves, and play around with these tools. The belief and conviction comes with you yourself gaining experience through understanding these tools. You know, you can't tell someone, you know, how to make a music video or make a movie without ever having used a camera before, right? So, it's the same way. You've got to learn how to use the tools first.

VICTORIA: And are there any yellow or red flags you would tell people to watch out for if they're thinking about AI or thinking about using a new AI product?

KRISH: I think for those founders that want to build large venture-scale businesses, and they're trying to bite off way more than they can chew, you should consider focusing. These are the sort of folks that maybe are not making a sequence of bets. They're trying to throw a hundred darts and see what sticks. And I usually think that's a strategy that will fail.

You need to understand why you're building, what you're building, who you're building for. Don't just build it because the technology is cool. You know, not to pick on any products out there, but there's a lot of hardware devices coming out recently that have AI backed into them, right? And you wonder, why the heck is this a hardware device? Couldn't this be just an app on my phone? Like, why do I need to go spend $200, or $600, or $1000 buying this device that has a lot of limitations?

The reason you built it because you thought the technology was cool. But by the time it got to production, it has a lot of faults. And you're trying to get people to change their behavior and take money and pay for this? That's tough. And I think VCs are falling for that as well, like, in funding tons of this money into these sorts of companies. Some can argue that it will get better with time and iterations.

But I personally stay away from hardware. I don't want to touch anything related to hardware right now because we don't even know what the new form factor is going to be. But the hardware people should ask themselves, "Should this be a standalone device, or could it just be something on my iPhone as an app?" That is something that's really, really interesting.

The space that I'm most excited about outside of AI for the workplace is robotics. And I've been seeing a lot of really cool products where they're trying to build these AI humanoid-like robots that can do a series of tasks. They're not like the machines in, like, an industry or a factory. But they can make you coffee. They can clean the dishes. They can cook you some food. I think the market for that is massive. Like, if that stuff works, people are going to be able to pay a lot of money for it. Like, the amount you'd pay for a car, you would pay for a utility-based robot inside your house and, like, with nice financing options and stuff. So, whoever cracks that is going to be really, really successful.

There's people companies that have raised a lot of money solving that. While I'm generally not bullish about hardware little devices, I am very bullish about, like, these general-purpose robots that I think the potential is immense. Like imagine every household having one or two of those; what that means for domestic productivity, like, someone's folding the laundry, someone is cleaning up the house, taking out the trash. These are jobs to be done, yeah.

VICTORIA: Well, then what would my husband do [laughter]? I'm just kidding. I don't want to replace him. No, I think it's interesting especially just, like, thinking about elder care, and having someone in the home, and watching, and cleaning up, and all of those tasks and being able to live independently. I could see that having a huge potential. So, also, obviously, I think robots are cool. It's the title of the podcast. So, I'm very pro-robot [laughs] in most cases, not all cases. Yeah. Well, that's super interesting. Let's see. Do you have anything else that you would like to promote?

KRISH: You know, besides embracing AI and using, you know, these tools and services, I would really be excited to hear about people's ideas on, like, how they're using AI in the workplace. Everyone has so many creative ways to go about it. So, each week, we discover new ways people are using Fireflies, right? Some people use it for taking notes. Some people use it to be able to take customer quotes from calls. So, they can literally ask our AI, "Hey, go through these, like, past two customer calls and pull out all of the nice things they said about us, and then turn that into a soundbite that I can share with my marketing team so we can run a marketing campaign on that" So, there's just so many interesting use cases.

I do want to say that voice is going to be a great form factor for AI. We work in the voice space. Like, I love talking to my AI during the meeting. So, I think that's going to be something that I would say is if you are an end user in the workplace, think about how you would use voice to get work done and turn your words into AI. And we're trying to solve that at Fireflies. And if you are interested in that space, we would love to talk to you. And if you have some interesting use cases that you want to see for Fireflies, please send them our way.

VICTORIA: I love that. And it's interesting when you bring up voice. One thing I was surprised about with my parents, actually, obviously, a generation older, I got them an Alexa Dot that I got from a conference. I didn't think they would ever use it, but they actually use it all the time. They're, like, asking for recipes, setting timers, and doing things like that. And, yeah, if you have, like, an AI voice, like, "Send an email to this person" or, like, "Open this task and do it." Maybe I would actually get some more tasks done [laughs]. I could just do it over voice. Sometimes like, the keyboard and the screen is part of the delay. That's really interesting. Thank you so much for being on the podcast.

Do you have any questions for me before we sign off?

KRISH: I'm curious to hear your thoughts on what are the biggest risks with AI you foresee for people, and what makes you more skeptical about AI?

VICTORIA: Yeah, you touched on a little bit earlier when you said about the cost of AI and the cost-benefit analysis; I don't think is always there for every single use case, right? There are some use cases where it is so clear there is a benefit for that. Note-taking is one of them. There's a million professions, I think, that would benefit from having AI note-taking apps.

I think the risks which we've already seen that impact people, you mentioned the biases, and things like people getting denied health care, getting longer prison term sentences. You know, the way that they might blindly incorporate these algorithms into decisions that really reinforce biases because of this historical data that it's based on.

I think whenever someone asks me about the risks of AI and, like, people losing jobs, or, you know, rogue AI taking over the world, I always bring it back to that some AI is already hurting people, and it should be stopped, and people should be educated on it. Like, the big scary AI conversation is almost a distraction to what's really going on, and we need to all be smarter about it.

At the same time, I love using AI. I think it really can, like you said, get your productivity up 100%. In some cases, like, you can just do so much more so much faster. And I see that potential. And I think that there's always that balance, right? Like, you have to be able to be aware and embrace both if you're going to stay current. But there are some people who still send faxes and still do everything by mail. But, you know, it's like technology never really dies. There's just more of it in different ways, right?

KRISH: Absolutely [laughs]. That's awesome. Well, thank you. This was great.

VICTORIA: Wonderful. Yeah, I really enjoyed our conversation.

You can subscribe to the show and find notes along with a complete transcript for this episode at giantrobots.fm. If you have questions or comments, you can email us at hosts@giantrobots.fm.

This podcast is brought to you by thoughtbot and produced and edited by Mandy Moore.

Thanks for listening. See you next time.

AD:

Did you know thoughtbot has a referral program? If you introduce us to someone looking for a design or development partner, we will compensate you if they decide to work with us.

More info on our website at: tbot.io/referral. Or you can email us at: referrals@thoughtbot.com with any questions.

Sponsored By:

• thoughtbot: Now that you have funding, it’s time to design, build and ship the most impactful MVP that wows customers now and can scale in the future. thoughtbot Lift Off brings you the most reliable cross-functional team of product experts to mitigate risk and set you up for long-term success. As your trusted, experienced technical partner, we’ll help launch your new product and guide you into a future-forward business that takes advantage of today’s new technologies and agile best practices. Make the right decisions for tomorrow, today. Get in touch at: thoughtbot.com/liftoff

Support Giant Robots Smashing Into Other Giant Robots