An AI Deep Dive with Lightspeed and Typeface: Embracing the Potential, Addressing the Challenges

Great to see you all. Thanks so much for joining us today. I'm Alex, partner at Lightspeed. Been at Salesforce Ventures prior to this for about years where I actually worked very closely with Julie. I'm excited to be at Lightspeed where I'm helping us double down on SaaS. What we're going to talk about today is a new framework that Lightspeed has developed around how to think about this new era of SaaS 4.0 and what the opportunities are for startups, how to think about incumbents, and then really excited to have Abe, who's the founder and CEO of Typeface. One of our portfolio companies. He's a perfect person to have this conversation because he just started Typeface last year and his background will speak for itself, but you'll see why it'll be a very rich conversation. So without further ado, you probably know Lightspeed, but for those that don't, we've been investing in generational companies for about 20 years now. Our roots are actually in early-stage enterprise investing. Today we also do consumer, healthcare, and fintech, but the real bread and butter of the firm has always been enterprise investing. And that's why this community is so important to us at the firm.

As Anik mentioned, we've been investing in enterprise software for more than 20 years now. And I think what's been remarkable to us is really that over those 20 years, enterprise software has been incredibly resilient and continues to be so today. So why is that? Think about it. JP Morgan today has more developers than Salesforce. L'Oréal, the beauty company, has over 1,000 employees with data in their job title. And John Deere, the agriculture manufacturing company, has over 4,000 developers as well. So every single business strives to be a technology-first company today, and enterprise software really went from just being part of the IT budget to touching the entire organization— sales, customer success, operations, and HR. And that's why we're so excited about it. Plus, cloud applications, a huge important part of this whole sort of growth, is still growing super, super fast today. Over the last year alone, cloud applications, public cloud as part of the IT spend has tripled in one year. That's incredible. And that's why Alice and I, we spend so much time in vertical SaaS thinking about disruption of not as digitized industries like supply chain, for instance, and workflow automation as well. But the real elephant in the room is obviously generative AI and the impact of that shift on the SaaS community as well. We see 3 buckets, and that's at the risk of oversimplifying a little bit, but really excited about A, the model layer, and we do think there's gonna be more than one foundation model here. Secondly, the enablement stack infrastructure so that actually enterprises are able to adopt these technologies safely and also efficiently. And then lastly, where again, we spend a lot of our time, the application layer and vertical SaaS is gonna be a big part of that. But the point that I really want to drive home here, and that's, I think, really important, is the real opportunity of generative AI is really enabling industries that so far have been very slow to adopt software and have been very services-heavy. Think about everything we're seeing right now in legal tech, accounting, and healthcare. That's where we see the real shift towards a new era in SaaS that Alex is going to talk about.

Part of our view, by the way, is that There is an incredible moment of opportunity right now with everything happening. Julie's laid out the broad trends behind this, but so much of what's the history books will say about this era is being written right now. And so we've taken a step back at the firm and try to think about how to piece together where we are. And a big part of that is looking at this at a historical basis. So taking a step back, the first era of cloud was really back in '99 when companies like Salesforce where both Julie and I used to work, pioneered this whole shift in cloud computing. Of course, CRM was not a new category. There were existing databases that did this. They just did it on the cloud— sorry, on-premise. And so the real innovation was taking that traditional database and pushing it up in the cloud. And so we think of that as the 1.0 era, where SaaS was really about the system of record. You move one over in the SaaS 2.0 era, That was really the system of engagement. And what we saw there was all of a sudden it wasn't just recording and storing the data in a database. It was how do you activate that data? How do you take advantage of it? How do systems start talking to each other? Actual applications connecting. And so companies like Slack and Twilio on the communication side, companies like MuleSoft on the API economy, and just a whole plethora of that next generation of software companies coming on stage. Then maybe about 7, 8 years ago, we had the early innings of AI, the first sort of automation of, hey, how can we drive more intelligence? And in our view, that was the real precursor to this GenAI movement. That system of intelligence was really focused on how to drive a bit more automation in the core application. A great example of that are companies like Moveworks on the customer service, IT support side. And then Gong on the sort of call recording and engagement side on top of Salesforce. And now finally, we're in this fourth era, which is really exciting. And we're calling that the era of cognition, where it's not just about actual automating a small part of the task. It's really about creating an entirety workflow where work gets done. And to us, we still don't know who the category winners are gonna be. You can see it's a big white empty box on the right, and hopefully some of you here will go fill that.

Obviously, it's also important to talk about the implications for software companies as we shift into this sort of fourth generation of SaaS. And I'm talking about both large enterprises as well as young and small startups. The first part is creativity is being completely democratized. What does that mean? We completely redefine what it means to create in the first place. What is creativity actually? It breaks down the entire barriers of creating, and that's why we're so excited about it. That means it touches music, film, art, but it also means how we, or completely changes how we build products and how we build user flows. And I think there's a lot of creative potential that is being unleashed here. Secondly, we've seen a sort of an evolution from being data-driven in our recommendations and next best action to a more sensing approach, as we like to call it. What we mean by that is instead of suggesting the next best videos to watch based on your preferences, why don't we start with actually creating that video that matches your preferences the best from scratch? And then lastly, this is truly the era of personalization. Obviously that means for founders building that you really have to— the user expectations are higher than ever. You really have to match a huge level of personalization for every single user. But what excites me is the opportunity here in education, healthcare, Also the content ecosystem. It could be that finally now we're getting close to a point where personalized healthcare is becoming much more available to a much larger part of the society. With all of this opportunity obviously also comes a huge responsibility. This is an ecosystem that is still moving incredibly fast. We're having a lack of ownership to some extent, and sometimes even don't really understand what's happening under the hood technology-wise. And so I think as company leaders, as organization leaders, we really need to ask ourselves, what are the implications of such personal data being used and shared in that, in this large amounts for our employees, for our organizations, for customers and users. And I think we see actually a big stack of technology infrastructure also emerging around that topic. But something that we love to spend a lot of time with is, as Alex already mentioned, that we're also gonna discuss in our panel after this. Is how the dynamic between incumbents and startups has shifted by the advent of generative AI. What we mean here is that so far it's been always enough to be very agile, fast moving in a new space, and the first one with a small company to go after it. That has shifted now. Today, the ability to access large amounts of proprietary data and have huge distribution power have become so much more important. And so surely we'll discuss a little bit how that changes the playing arena for startups and incumbents alike.

So one of the benefits as venture investors that we have is we get to meet a lot of companies. And actually the flip side is if you're a founder building, one of the hardest things is you're within your four walls, really focusing on building your business. And so with the last, the benefit of the last year and having seen a lot of companies, both in our portfolio and companies that we've gotten the chance to meet, we've taken a step back and thought, what are the big dimensions that when we think about backing companies gets us excited in this sort of new era of cognition, as we're calling it. And there's 5 areas that we've identified and we're going to go through each of these. So the first one is data and is it proprietary data or is it data that's publicly available? Our thesis is that it's a lot more valuable if you can take advantage of data that's proprietary and private. Great example of this would be if you're building an inventory optimization company, What's more relevant? Is it just the weather patterns in terms of predicting when someone might buy a t-shirt or ski gear? Or is it actually being able to integrate into the actual transaction history of the merchant, seeing all their historical patterns and actually driving insight for them that way? And so that's what we mean by private data here, not just what's available to other companies as well. I think the simple way to think about this is if you have access to it in the public domain, incumbents, large technology companies also do. And so really trying to think about where are the sort of unique datasets that you can develop with your customers to drive more of that insight and automation. That's number one. Number two is frequency of interaction. The beauty about software is that it really powers workflow, but not all workflows are created the same. And our hypothesis here is that frequency is going to matter a lot. There are some tasks that happen once a week. Once a month, once a quarter, maybe even once a year. There's other tasks that happen literally on a daily basis, at least weekly. And our excitement in terms of backing the next generation of software companies are those that are really focused on that frequency of interaction because we think it can drive a lot more engagement and a lot more value for customers. The third is what's the actual product that you're building? The comparison of starting a company today and building product versus 20 years ago 20 years ago, you had to raise $20 million just to get an email server up and running. It's incredible how much now is available. And that's true of the GenAI movement as well. We've all played around, I'm sure, with ChatGPT and OpenAI and just the amazing technology that's at our fingertips as consumers of this. But product gravity matters a lot. And just having a thin wrapper around an existing workflow, we don't think that's going to drive a sustaining advantage. It's a lot more about the deep integrations that are required into traditional systems. It's more about the workflows you can drive. It's the community you can build around that product. So in our view, the gravity of what the sort of depth and gravity of that product matters a lot. And so if you think about depth, try to explore not just the initial use case, but where you go from there. And that's one of the big areas that we've been excited about in the companies we've backed. The fourth, this is an amazing opportunity to drive 10x improvement in terms of what actually happens from a use case product collaboration perspective. I think today decision makers in software and IT, they're inundated with tools, companies. There's more than, on an average company, about 4,000 SaaS applications. And so ask yourself, how can we drive a 10x improvement? Whether it's through the product and actual experience, whether it's through the business model, whether it's through the community you can drive, this is going to be a tidal shift, I think, in terms of what can actually be built. And those companies that drive that sort of significant improvement are going to be the winners. And then last and not least, this ties a little bit to the SaaS 3.0 versus 4.0. What is it that you're automating? If what you're automating is a sort of nice to have for a customer, sort of more in that low-level intelligence where yes, it could help their business, but it's not going to move the needle. That's going to be a very different outcome for the business and for you. Than if you're automating what's a more, what we call high-level automation that drives meaningful ROI. And so as you think about building in this era, really go back to customers and what's in it for them. Why should they pay attention to this? So that's our framework. It's clearly a super exciting time to be building in software. We've been investing as a firm for over 20 years. I've been investing in software about 15, and I have never been more excited about what's to come, but obviously it's also a challenging moment because there's so much innovation happening quickly. Incumbents are taking advantage of that as well. And so as startups looking for opportunity, this is a really unique moment. With that, let's actually dive into a builder that today is at the forefront of this, is at the, in the trenches of building with GenAI, and is actually disrupting himself. Abhi, I'd love it if you can please join us on stage.

Great to be here. Thanks for having me here and really good framing. I think you covered a lot of the themes that certainly we are experiencing as we're building the business. I think for me, it's actually a lot of times when you have these kinds of discussions, sometimes the expectation is to answer with, oh, there was a well-laid plan and was very clear from the beginning. Actually, the reality is very much a lot of coincidental moves. I happened to always love software, was self-taught in programming, ended up in the first job where there was a project that didn't even have a name. And that became Java, which was a programming language. And I didn't even know I was working on something. And so a lot of it is a series of kind of coincidental moves. But probably the one common theme across all of them is I've always gotten more excitement and energy working on what I call flag planter type of activities rather than road builder, where there is no well-defined path, there is no well-defined market and product, and you get to go chart some new territory. And so that's probably the one common theme. It doesn't always work out, but that's probably the fun part of it.

Yeah, I think by the way, that story is very common. This sort of like accidental, some of the best founders we've backed have that sort of, they, they go into domain and it, to them it's obvious, but to the rest of the world it isn't because of that path. Where did the idea for Typeface, maybe you can share a little bit about your vision around the product and the category and how you thought about what you wanted to build there.

Yeah, I mean, it's interesting sitting here today at Saastr when GenAI is all the buzz and everything is centered around GenAI. Around November of 2021, I was still at Adobe as CTO leading a lot of the products. We were already investing quite a bit on AI, but just not the GenAI form. And then I happened to get an early view into some of the transformer-based kind of diffusion models, GPT-1 and 2, and they were not quite there. But if you could extrapolate, you could see that the world of software is going to fundamentally change because in some ways the last decade with the cloud, and you talked about the SaaS 1.0, 2.0, 3.0 waves, the world of data has gone through a complete transformation. If you look at all the unstructured data, big data companies like Snowflake, Databricks, amazing companies have been built on the side of data. And then at Adobe, we always sat at the intersection of creativity and kind of science. And what was clear to me that the world of content in the enterprise is just completely stuck in the dark ages because there wasn't a 10x breakthrough, as you were talking about. And GPT, even in an early form, was very clear that if this continues, the hardware acceleration, software acceleration is going to get to a place where entire world's content architecture for companies is going to go through massive inflection point. And so for us at Typeface, really the genesis was the same way when original printing press happened, there was only one font. And until typefaces were invented, you couldn't really personalize and tell your story. And so our thesis was if this underlying GPT and foundation models are the AI equivalent of printing press, you need a personalization layer for every company, every person to tell their story in their own voice, in their own brand. And so we said, let's go build that software.

Is that where Typeface comes from?

That is the name, actually.

I didn't realize that.

That is not a story I usually— yeah, but that is the reason why it's called Typeface. In some ways, the rate at which the industry is moving, I have been fortunate enough to be part of cloud transition at Adobe or mobile. This one is at least 10x faster in its way it's moving. We actually officially started, I started a company May of last year, so roughly 15 months ago, which in some ways is super short. We are still obviously super early, but I remember ChatGPT, Stable Diffusion, none of the things that happened. And when we started the company, And we started talking to customers or some early investors like Lightspeed. One of the questions was, but what is the platform you're going to use? Because none exists and there is no platform that you can build a company on. To Lightspeed's credit, Ravi, who is founder and CEO there, certainly saw the arc of where it could land. And literally when Lightspeed and I started the journey together, there was a thesis, there was a conviction, but ChatGPT had not happened. And so we thought it's going to take 3 to 5 years. And yet here we are within a year, everything has changed.

That's amazing. I, I'm curious, a lot of founders, I think, looking at the space that you're building would have thought incumbents are just going to go and do this. And you've now with your career have been on both sides and you've seen plenty of startups. You've been in really the iconic companies of technology. How do you think about spaces that are interesting for startups to go and explore? And where do incumbents actually have the advantage? As you maybe in your own journey with Typeface, but even more broadly to the audience today, because there are obviously distinct advantages and disadvantages.

For us, it really, I would probably start with what's common across incumbents and startups, at least for me, that over the years I've seen is whenever there is a disruptive shift, you have, at least for me, you have to start with a deep product thesis. Can you build something that is a deep technology and product moat? I think you called it the thick or heavy application layer. Actually, that's very similar across both these companies. If you look at companies that I was fortunate, like Adobe or Microsoft, there aren't many companies which have products that are 20, 30, 40 years later are still considered gold standard, like Photoshop. And the reason they are is they have a deep technology moat combined with an exceptional product user fit where the users love the product. So first I would say that's actually something we think a lot about at Typeface, which is what is that product truth and what is that product mode? Because no other— there can be a lot of hype, there can be a lot of capital, there can be all kinds of PR, but if you don't have that product truth, that I think is true across both of these. But then where it starts diverging, one of the things I saw consistently at big companies, and I think companies like Adobe are still better at transforming themselves over the years, but when you have existing businesses and existing customer base, your ability to disrupt your own business takes a significant amount of commitment and discipline, and not every incumbent can do that. And even those that can do it, they cannot move as fast enough or adapt fast enough. So one of the advantages, if you have a core product truth, then your ability to adapt and fundamentally make offerings in the market that the incumbent will never make because their economic incentives are just misaligned. In the short run, I think is one of the fundamental things. At Microsoft, I worked on Azure version 1, and when I was there, one of the challenges was, could we or should we offer Linux-based VMs when you have a $60 billion Windows business? And even though we knew technologically that was necessary, it's not easy to, when you're a public company. And so I would say that economic incentives, incumbents can leverage that to their advantage. When incumbents actually have it misaligned in the market. So that's kind of one of them. But I think also speed of change. I fundamentally think, and this is what keeps me up at night, is our only weapon at Typeface is the rate at which we can change with the market has to be 10x faster than the rate at which an incumbent can change. Because they have 10x, if not 100x more capital, more reach, more customers, but their ability to change is the one thing we can do faster.

Super interesting. I don't know if you both agree with this, by the way, but I found that whenever folks that haven't worked at a big company talk to me about a big company, they think of it as a monolith. They're like, oh, a Salesforce. Why does Salesforce do or not do this? Or why does Adobe do or not do this? And actually in the case of Salesforce, there was really 6 or 7 different business units.

Yeah.

And they're, each of them, by the way, would be like one of the largest public SaaS companies on their own. As a result of that, it's not like there's, of course there's an overarching strategy, but there's a lot of sort of resource allocation. That has to happen within those departments, domains. And to your point, it's like finding that thing that you can go fast at, faster than anyone else with that focus. It might be something interesting to the big company, but it's not something that today they're prioritizing.

And the people who have that DNA, there are people like that in those companies. I was one of them for many years in big companies, but those people typically have to fight the internal kind of status quo to get these new products out. And they do, and they try and they succeed a lot of times. So I do think you're right. It's not a monolith, kind of one giant fabric.

I think when we talked about the session and prepared a little bit, you mentioned, you talked about the term data uniqueness and how important that is also when you're building a business in this ecosystem. How should we think about that? First, as an incumbent, you might have more access, but then also as a startup, How do I get there, right? When I have this closed-out problem a little bit.

Yeah, this is, first of all, that's a fascinating question. And I know you talked about the public versus private datasets. And this, I would say, even for us at Typeface, we are through that learning journey ourselves. So I don't think the final chapter on this is written yet. But the one thing that is very different about GenAI, not only is the technology different and it's fundamentally about algorithms, not just procedural code. Is for the first time, software is not just about codifying your intellectual property through bunch of automation and algorithms, but it's actually about extracting unique patterns through datasets or deep domain specialized kind of verticalization. And that's actually one of the hardest things for incumbent because at least most incumbents, they have a lot of that data. They have a lot of those assets through their customers, but they didn't build their businesses with that architectural pattern, but they are sitting on that gold mine. Now, if you're a startup, your problem is you don't actually have access to that data. And so the best strategy, at least in my mind, is one of two things. Can you in fact partner with these incumbents? And I know we'll talk about that a little bit because they are sitting on that gold mine, but may not always have ways to extract value out of that, at least right away. But then there are a lot of end customers who actually are looking at GenAI Like we are working with one of the largest consumer electronics goods companies. They operate in 90 countries. They sell TVs, refrigerators, washing machines, all appliances, and they have tremendous amount of data. They just don't know how to connect that data to the world of generative AI. And they're very nervous about handing over those crown jewels to OpenAI or Microsoft or Google because they're worried that their crown jewels then suddenly will be in the hands of these incumbents. So they are actually extremely excited about working with startups because they want to push the limits of their assets by working with companies that are not as big a threat as incumbents are for them.

That's fascinating.

So it's almost like the fact that you're small, new is an advantage in that case. Wow.

How do you set yourself up though to be able to get access to the data, let's say from security perspective, and especially if you're just super young? All the requirements that large corporates have to work with you. It's, it's a lot.

Yeah, it is. And there are days I wish I had a simpler answer to that question. But one of the things I will say, and maybe this is a little bit from the trenches, as exciting as GenAI is and as exciting as the pronouncement at SASTR every session may be that this is going to change the world overnight, I actually think it's going to take a long time. I think this is an undeniable fundamental shift. But I will remind people, when I worked on Azure version 1 in 2008, 2009, and 15 years later, cloud is only still 30% deployed. The world is still 70% on-premises. Doesn't mean cloud isn't the right direction, but these things take time. So first I would say the notion of compliance, copyright, safety, do you say kind of have ethics and bias, all those considerations. Are as much important as any innovative demo or a feature or exciting, cool new thing you can build with GenAI. What we are finding, if you look at Typeface, which I did not think when I started the company, we just hired a world-class chief security officer within the first year of the company's existence. I don't think that's what most companies would've thought they would need because the enterprise dialogues are not just about the positive aspects of GenAI. They're all about, tell me how you're going to indemnify any IP copyright violations of these models. Tell me how my IP is not going to leak back into these public platforms. And so I do think if you're a startup, back to your question, which I really think is very insightful, which is you may think that is not an advantage an incumbent has, but that is actually a fundamental advantage incumbents have with these enterprise customers because they are used to speaking that language. They know what the traps are. They know what the CIO cares about. And they don't even have to build any features. They can just freeze you out based on those aspects, which I think is again, something incumbents have a strength on, but startups can actually go there. It is a real investment. We are investing a lot in that in our engineering roadmap. I wish we didn't have to, but if you're selling to large companies, that's kind of part of the journey. One other thing, the way GenAI space is going, I think incumbents are not going to stand still. This is too big an opportunity. The economic prize is too big. I don't think it's going to be one winner take all. As much as OpenAI had an early start, amazing work, every big player is going to end up owning their share of this prize in the coming decade, whether it's Salesforce, Adobe, Microsoft, Google, they're all, they are not dummies. And so the question then becomes, if you're a startup, can you just ignore all of their investments, their strategies, or can you somehow capitalize their interest and their excitement around this space? Like this chief security officer, within the first 9 months, we struck deep partnerships with Microsoft, Google, and Salesforce. And all 3 of them are actually deep technology partners. But I do think distribution is the new product mode in the GenAI. You cannot just build a product. I know there is a lot of, I'm sure, talks here around PLG and how PLG, you can just win with PLG. Yes, that's great. I happen to believe you're ultimately going to need distribution advantages like data. And these big companies have massive distribution advantages.

I'm just curious on the time commitment that took to pull that off in terms of how involved are you on the partnerships? I think it's an area that a lot of founders haven't done. And as they think about what's outsourced to like a BD person on their team, how involved you were. You talked about a few of the largest companies in software. I'm sure it took a bunch of your time.

Yeah, this one is unfortunately, and like anything else in startup, it's all a matter of trade-offs. First of all, I do think I was personally the program manager of these partners because at the end of the day, when you're a small company, it's hard to get access to these dialogues. They're not going to listen to, they'll barely listen to a founder and CEO. They're not going to listen to somebody within a company that they don't know has any kind of influence. And yeah, in my experience, for first couple of years, you can't delegate that, which is unfortunately the real risk because you are taking valuable cycles away from product building, go-to-market, getting your own customers. To trying to land these somewhat speculative partnerships. And if you can't land them, you've just wasted all your time and energy. So I do think you have to be mindful of that. For us, for the first 15 months, I was the BD person myself. Yeah.

Amazing. Look, thank you so much. Thank you so much for joining us. And Abhi, on behalf of Lightspeed, thank you so much for your insights.

Thank you. Thank you both.