How do tech startups go from idea…to iconic? Greymatter’s weekly podcast features interviews with VCs, founders, and tech visionaries on everything from AI, marketplaces, and cybersecurity as well as company-building strategies like growth, finding product-market-fit, and recruiting top talent. Founders and VCs offer candid accounts of success, failure, and adaptation. Greymatter delivers practical insights mixed with the bold optimism that defines entrepreneurship.
Oble, native apps and businesses like Instagram, Snapchat, and Uber, emerged and gained popularity several years after the app store's launch. Utilizing new capabilities like GPS and the camera to serve our needs for communication and transportation in entirely new ways. As we examine the current AI revolution, similar patterns are emerging. Ports are all the products adding conversational AI interfaces to their existing platforms. With varying degrees of value. AI toys are also everywhere. Some may disagree, but I believe most text to image generative AI tools like Dolly are toys, delivering mostly on novelty for enthusiasts, but not yet durable utility. That said, in rare cases, products that may be considered toys can sometimes evolve into native products with staying power, but this requires meaningful core utility. We're currently at the phase of the AI tech shift where there is still ample opportunity to build the native transformative products that will define this generation. The last phase is where the most innovative, impactful, and enduring products are built. These are the founders I'm looking for. They will build the future. Having previously led product and growth for 1 of the iconic native apps of the mobile era, I've developed a perspective on some of the best and worst strategic and tactical decisions that companies can make during tech shifts. It starts by asking an important question. Are you building a novel solution to an enduring need enabled by new technology? Using lessons from my hands on experience at Snap, and my observations across other companies, I hope to provide a framework for evaluating your product and company to ensure you are building a transformative product and not just a toy. These guidelines are not intended as exhaustive or definitive, but I hope these help founders answer some of the key questions during the various processes of building, designing, testing, and launching your product. First, it's important to have
Is which platform is going to enable because AI is going to be indispensable part of any technology solutions that would be built for the next year coming years decades. So which platform would allow these large, you know, organizations, enterprises, individuals to go and build those applications will accrue that value. Yeah. It goes to this blog then, obviously, thesis. You know, I've talked about, the new Moats originally 6 years ago, which kind of shortly after the InstaBase investment and then the new new Moats update that you read and help review around where's the value accruing. For sure, there's value accruing to foundation models, to the GPUs. And then I think potentially, you know, I talk about the system of intelligence, the system of engagement. Right? So clearly, you know, I I look at Instabase and you guys have built maybe the definitive system of intelligence company out there for an enterprise AI. Right? And, which gets me excited is that could be an area that does accrue value. So I'd love to understand why you think this area that InstaVates plays in is the right market for start to enter as opposed to build another foundation model or build, you know, a cloud hosting provider to run AI models. And, you know, those are all right kind of technologies to build. I'm not saying people shouldn't build it because more foundation models, more, you know, competition there. Foundation models will get better. That's great. But I think there would eventually be only, you know, some 4 or 5 foundation models that would end up winning. And the reason why is, like, all of these foundation models have their own quirks. They have their own, you know, way of how do you do the prompting? How do you optimize things? And when you look at, you know, sort of like the the layer that is above this, it's very, very hard to support more than 4 to 5 underlying platforms. Right? And so tell me about where this idea for Instabase came from. Maybe, start at the beginning. MIT, you're a grad student at building something called DataHub. So tell me about DataHub and tell me how that was
Where the modes, where and how can startups attack big cloud? And it kind of comes full circle that 6 years later, you want to revisit this thesis of systems intelligence in light of all the technology changes the past 2, 3, 4 years. And how would you characterize the way AI is being used right now? So I think how we interact with AI has changed dramatically from kind of a very stilted, you know, Apple Siri kinda got it right, kinda didn't get it right, to holy cow, GPT and Bard eerily know what I wanna talk about or can kind of answer my thoughts or questions when I kind of access kind of very abstract concepts. I think there's 2 ways AI is being used by applications. Number 2, under the hood, these large language models that kind of power these chat bots like ChatGPT, this would be GPT from OpenAI or Palm from Google. These large language models are actually pretty game changing in terms of really problem solving reasoning engines. They actually solve problems and help us kind of be more productive. So either 1, change how you interact with the world. Or number 2, use these large language models to actually build something net new. That's pretty powerful. So would you say that AI is an option for companies to use? Or is it becoming the something that is essential to even running a business? Yeah. I think we're kinda joking around AI or die because it had this kind of nice little rhyme, but it's more like evolution of what companies and technologies are. These are the the great waves we're seeing. So I think in the early phase of any kind of technology wave or new platform shift, you're like, yes. It's a mobile app or mobile first or it's web first, web only, or it's cloud first, cloud only. I think we're the early days of saying everything's AI. And, really, what we mean is like, AI is going to permeate enterprise software, security software, consumer experiences, your self driving cars or home automation, your healthcare. So in some small way, just like the internet is
Will be amplified by artificial intelligence, and that's just the beginning. Mhmm. Right. Now you announced a little over a year ago that you'd cofounded Inflexion. At the time, you described it simply as an AI first consumer products company. There hasn't been much else said since except that inflection is attempting to flip roles for humans and computers instead of us trying to understand how to talk to them. It's the other way around. Now bring me up to speed on where you are on delivering on that goal. What has inflection built? One way of thinking about it is that for as long as we have been creating software, we have put the onus on us as humans to try to understand the language of machines, and we've had to learn their programming languages, their interfaces, And that's been a huge constraint, and that is all about to change because for the first time in history, computers are actually able to communicate with us in the same natural plain English language that we're using to communicate with one another now. And we think that that's going to completely transform what it means to have a digital experience. Everything that you currently do with computers is going to increasingly feel like a conversation, a back and forth, a dialogic. Right? Your AI is going to ask you questions, is gonna clarify, is going to try and sharpen your understanding, and through iterative back and forth, you'll be able to convey your real intent. In fact, you'll be able to share how it is you're feeling, not just what it is you're thinking or what you need to find, but actually your emotional state. And that will create a very, very different experience to the kind of thing that software and technology has ever been able to do before. And I think part of looking at this in terms of what we're doing with inflection is to say, how do you help people navigate their entire lives?
Out there. And so we're super excited about how this develops, but for sure, there's one layer that is the foundation models. OpenAI, Cohere, Anthropic, StabilityAI, companies out there like Adept and Inflexion might play in that space, but for sure you're gonna see startups and big companies play in the kind of the the low level. You're gonna see the top level application innovation. So Jasper, EvenUp in legal, tome in presentations, Harvey in legal AI applications. Just so many companies out there taking advantage of these, you know, advantages in AI to build applications. Close to the Greylock portfolio is something called Insubays that started document understanding AI around there. They're really expanding on top of these foundation models. But this middle layer in between is kind of a an evolving middleware stack. So we're seeing companies and technologies like LAMA index that's really defining, the category for data ingestion, data indexing for data querying is like this memory data layer around these AI apps because these these large language model apps need some level of data integration and and memory. We have, technologies like Langchain and Fixy that are really pioneering how you think about building agents, doing prompt development and prompt operations, And then vector databases. Right? Vector databases have been around for a long time, but also the idea of storing and querying these vector embeddings is becoming a key piece of all these applications. So you're seeing comes like Pinecone, Weebiate, Chroma, all kind of take hold in terms of this kind of embedding stores. And then when you have a new stack, right, just like when you build mobile apps or cloud apps, you break a bunch of things, you gotta fix these things, you gotta monitor these things. We're seeing starts like Helicone and Honey High around Elm monitoring. Right? And it's saying there'd be there'd be more monitoring tools, security tools, and management tools kind of all around this new application stack. So, look, if we had kind of LAMP stack years years ago, I don't know what the stack looks like the next 3 or 4 years, but we're seeing evolving, you know, setup.
