Casber Wang of Sapphire Ventures joins Nick to discuss How AI Reshapes Distribution Models, What Startups Get Wrong About Product-Led Growth, and the Real Use Case for Agents. In this episode we cover:
- AI’s Impact on Business Models and Capital Efficiency
- Defensibility and Product Market Fit
- AI Fatigue and Market Conditions
- Fundraising and Investor Expectations
- Cybersecurity and Competitive Intensity
- AI Agents and Enterprise Value
- Glean’s Success in Enterprise Search
- Market Conditions and Investment Strategy
Guest Links:
The host of The Full Ratchet is Nick Moran of New Stack Ventures, a venture capital firm committed to investing in founders outside of the Bay Area.
Want to keep up to date with The Full Ratchet? Follow us on social.
You can learn more about New Stack Ventures by visiting our LinkedIn and Twitter.
0:17
Casber Wang joins us today from Menlo Park. He’s a Partner at Sapphire Ventures, where he focuses on growth-stage investing across enterprise AI, cybersecurity, infrastructure, and broader enterprise software. Casber has backed several breakout companies that have reached unicorn status, including Huntress, Cyera, and Glean.
Before joining Sapphire, he was a tech banker at B of A Merrill Lynch, working on major IPO and M&A transactions. He also co-founded Etch.ai, a personal CRM incubated at Berkeley Skydeck.
Casber has been recognized as a Rising Star by Business Insider and named to Silicon Valley Business Journal’s 40 Under 40.
Casber, welcome to the show!
1:00
Thanks for having me here. Nick!
1:02
It’s such a pleasure, sir. I hear it’s a very nice day in Menlo Park, so I’m jealous, but talk through you know your backstory and your path to venture?
1:10
Yeah, certainly. And it’s absolutely a gorgeous day here in Menlo i So, like you have laid out already. I went to Berkeley for college, actually, at Berkeley, started the company called edge. That’s a personal CRM sort of management software where you and I, who can use this as way to track our contacts. And maybe before the podcast, the idea was very much, I can talk to Nick about three top, top topic topics, and what should I talk about, and actually raised a bit of money then, but we launched an app store, and, you know, honestly, it didn’t really take off, so had to make a pivot. And folks don’t really know what to do in college, they tend to go to banking, and that’s kind of what I did. And, you know, and like a lot of folks in banking, it’s not really their full time career for the rest of their lives. And I was fortunate enough to be on IPO of a company called MuleSoft, and CEO introduced me to sapphire and photos. Who don’t know MuleSoft was big position of sapphires back then, and the company went public, and Salesforce bought them for, think 6.5 billion. And yeah, it’s been a little over seven years and have a look back. It’s amazing.
2:22
One of the very first interviews I ever did 11 years ago was with Anne Winblad. Oh yeah, summer windblad. And on the show, she talked about her fair at the time, fairly recent investment in MuleSoft. So that was pretty cool. Tell us a bit about you know, the approach at Sapphire. What do you focus on, and what is the thesis? Yeah,
2:42
for sure. So we are a growth stage generally. Growth stage is a pretty big term. The way we define this as post product market fit, right? So not just three folks in a garage, sort of product market fit, but really we see some sort of repeatability from a consumer, customer, buying behavior perspective. And the other thing that we focus on is really B to B, right? So we focus on businesses that are selling your business. It could be B to B to C, but it has to be, you know, B to B to begin with. So we focus on sort of, you know, one is growth stage and 2b, to B. And the way we think about backing our portfolio companies is we really try to back company of consequence, companies that today are sort of the category leader of that category, or they are emerging category leader right of an existing category that have the potential to be one of the greatest software companies of all time. And that’s our model. And we try to be active on your cap table too. So we tend to lead 75 80% of rounds. Not dogmatic about it. We want to be involved with the right companies. But we really try to be, you know, the most valued added investor on cap table. How do you feel about
3:54
categories that are known in our like existing budget items in the enterprise versus category creators, you know, very disruptive and brand new categories, you know, how do you think about that at the investment stage and sizing up markets and making investments? Yeah,
4:11
Nick, it’s one of the hardest questions for any investor you answer. Frankly speaking, the way I think about this is, if there’s an existing budget, you know, there’s a clear target that you have to go after. Generally, the path to, you know, building the IR, or a really big outcome is more laid out, right, versus category creation. And oftentimes, when you’re going after incumbents, and the way I think about this is the pro, or one of the pros, is certainly the target is a lot more clear that you have someone to go after. On the flip side, the issue is you have this 700 800 pound gorilla in the room, and it’s creating sometimes what I would call consumer surplus problem, enterprise software, right? What I mean by that is, you know, you can’t stay in my room. But I have a big TV here, and you look at TV prices right from the 1960s to the 2000s this would have cost like, $100,000 I think in the 20 years ago now, it’s probably, I don’t know, like 10 grand or something. And the problem, similar to you with an existing category, is you have incumbents who are pricing products, right? You know, the Microsoft rule of the world at a pretty aggressive rate. So no matter how strong, how good your product is, you have a benchmark in terms of how much you can charge from a velocity standpoint, on the flip, flip side, a category creation story. I would say the issue with that story, oftentimes, is the category is not here. The puck you the puck is coming, but it’s not exactly here yet, so you have to be patient around this, right? Sometimes it’s a product issue, sometimes the technological issue, sometimes it’s go to market issue, but once the switch is flipped and you’re the only sort of game in town, it actually is a really good dynamic, because, in comments, don’t think your category is big enough. Well, they don’t think your category is real. And you can sort of have this time to really define and, you know, in some ways monopolize the entire category, right? So, you know, there’s no one way to think about it. I think we backed, you know, companies that look like Looker, who are more in the business now, intelligence side of things, very much existing category, you know, new approach to go after an existing budget and we back companies are sort of more more on a category creation side of things, right? So there’s no one size fits all approach to what we do, and we kind of value it on a case by case approach. I want
6:39
to talk a bit about AI and how quickly companies are expanding. You know, we’re seeing these growth stage companies now hitting 100 million of ARR, previously a very difficult milestone to get to, and now we’re seeing it happen inside of months instead of many years. And also we’re seeing that happen without sort of the corresponding massive burn and huge, you know, staffs that were once considered normal. So are we going to continue to see capital efficiency metrics accelerate with these advances in AI? Yeah,
7:16
I would say, first of all, I think AI certainly makes all business models more efficient, right? And that’s the promise of AI, and I think that’s playing out in a real way. That being said, I do think the capital intensity question is much more of a business model question versus Hey, whether you’re AI or not, right? The way I think about this is if your distribution model requires you to sell to JTM, Bank of America, Lockheed, Nike, you had a wine and dine. The key decision buyers. No matter whether you have AI or not, you’re going to spend the expense. You’re going to wait until the procurement cycle is ready. And that’s just the more expensive sales cycle. On the flip side, once you have those customers, if your product is real and good, it’s sticky, and your lifetime value is a lot higher and the quality of revenue is higher. Right now, a lot of AI companies like you mentioned have taken off in a massive way, and there’s word kind of has gone out of favor. But you know, three, four years ago who were chatting will be chatting about plg a lot and and in some ways, those names are more, you know, plg targeting consumers slash prosumers, right? And that’s just a very classic and very different distribution model, right? And I’m sure we can touch upon this. But you know, those are folks who are spreading credit card coming to your website clearly see real product market fit by paying but you know, they could exhibit very different behavior than a jpm who’s buying an ELA on your platform, right? They could be turning next week or next month. So there are different trade offs to it, and there’s no doubt, again, AI is making things more efficient. But I do think whether you know a business is capital intensive or not, the better predictor is actually distribution model versus whether this business AI or not. What about a company
9:10
like a bridge, for instance? I don’t think you’re invested, but you know they’re they’re doing a lot in healthcare. I just had a podcast with samesh dash over at IVP about them, and, you know, they’re selling into enterprise, into healthcare, and they’re working with epic and it’s like historically long, bloated sales cycles with some cloud software, some on prem. But they’ve also reached that 100 million dollar mark, I think inside of three years, it’s been pretty fast.
9:38
Yeah, totally. I would say, like again, capital intensity and velocity, two things that AI has triple charge, right? And we’re not investors in a bridge. So I, you know, wouldn’t comment on the specifics there. I would say, you know, there’s certain unlocks as I think of the story from outside in, right? Whether it’s the tech partnership they’ve established with epic, right? That’s, I think. Charging if I were a buyer all my days in Nappa, it’s pretty easy for me to adopt this. And on the flip side, like we talked about earlier, AI as a budget item, is so top of mind for CIOs and users from a tops down perspective. So if the entrepreneur is able to build a product that actually resonates with the end user, and you already have tops down approval. It’s a very sort of magical moment for the company, right from a velocity standpoint, because you have tops down buy in for folks who are motivated to come in and buy the product. And you know, if you actually build a product that users love, it’s sort of the right and the best combination both worlds, and that’s where we see sort of increased velocity of things
10:45
earlier, you talked about pricing benchmarks, and then you also mentioned revenue quality. Is there a risk to pricing durability with AI startups, you know now that they can be copied and competed with more readily.
11:00
I think 100% yes, if that’s the thesis, right, if the thesis is, hey, you build a product that could get commoditized, I think 100% it’s probably easier. But that being said, I do think AI companies and AI software companies, B to B companies, just like non AI B to be companies before them, if you just stop evolving your products, you’re gonna face competition, right? You know, the famous saying, your margin is my opportunity. I think that’s happening real time here in Silicon Valley and outside the valley, as well as more and more, you know, entrepreneur start companies. So I do think there’s a dynamic here where, you know, regardless of whether you’re AI or not, regardless of whether your first product actually hit that, you know, magical Product Market Fit benchmark or or milestone, you still need to keep evolving the product to keep up with competition, because the velocity of competition to avoid Nick is just as easy as Ever to build a company. So I do think again, going back to my point, like, is the mode in AI worse than before, not necessarily, in my opinion, but if you don’t evolve, I do think you know, the pricing power that you have will certainly be completed out. How are you
12:15
thinking about defensibility these days? Right? We see, we see a lot of startups grow fast and then churn fast. And I would venture to guess that most of those are because the product is insufficient in some way, but, but in some cases they’re they’re being displaced by competitors. So you know, what are your thoughts on defensibility?
12:32
It’s really hard and really broad, crushing Nick because sometimes I think defensibility is also crushing around. Why is this company growing fast, right in the first place? Is this because of true product market fit, or is this because the business is selling $1 for 90 cents, and sometimes the story is murkier, right? I’m sure on the consumer side, we’ve all had experiences where you used to be able to DoorDash a meal and have DoorDash pay as a coupon, right? And back then, the unit economics is probably not that great for the company, but lo and behold, now we’re in 2095 right? The business is a really profitable compounder in the public markets. So I do think you know, the most important thing is to figure out whether the business has true product market fit, right? And how durable that product market fit is. And in DoorDash example, that’s probably, you know, a scale question, right? The tech back end, those are scale question. They scale to a certain point where they’re in all the major markets. In a lot of BB software, it could be a brand thing, right? True. We’ll talk about cybersecurity later, but it’s the entire cybersecurity selling is built in trust to some extent. So if you hit some save velocity, it’s much easier for you to drive network effect through trust. It could be, you know, certain key partnerships and choke points that you’ve unlocked, you know, through partnering with certain players. But I would tend to say that there’s not necessarily like outside of the five big labs out there, like there’s not a quote, unquote, rocket science technological problem that nobody else could have thought of. And the competitive advantages of this era, in my opinion, it’s actually pretty similar to the classic software winner stories. In my opinion,
14:20
you know, when you’re you mentioned true product market fit, you know, versus selling $1 for 99 cents. Can you unpack for us some of the metrics outside of growth that you prioritize in order to assess true product market fit?
14:32
Obviously, turn and we’re not backward looking. Is it’s good to see. But I think the number one thing I really look at, and this is not something that, by the way, I paid that much attention to pre AI, is really engagement of this tool, right? I generally don’t. I mean, I care about average for sure, right? If you average looks more like a consumer product, that’s fantastic. And in some ways, that’s, you know, the gleans case, right? But if it doesn’t have to be right, because I. I do think the things I care the most about is one power users, so people who love you, like are they on this 724, seven, and then the other pocket folks who are going to be more turning is what I call, I would call it non consumption, right? Imagine, you know, if you’re coding tool, you have great power market fit, and now you have all kinds of non engineers coming and build, yeah, of course they’re gonna be more training, because building a tool on top of the platform still takes mental energy, still takes time. And if I were not an engineer, I probably would probably take a break week two to week three before I come back again. But I think seeing sort of how that user group behave, right? Like, sometimes is there smart smiley curve here, meaning they turn early on and they come back. That’s something that I really try to dig in. So one is just the engagement of power users. And two is, is there a smiley curve, amount of user group? Because I expect churn, by the way. Like, if you build a great product, yeah, it almost like you’ll have churns. You have all kinds of people come in who are not your ICPs, who are trying your products. But I think the the more key thing is, is there a smiley curve but they’re coming back later on? That’s that’s almost more important than if they’re turning or not in the first place?
16:15
What’s a healthy percentage of power users relative to total?
16:19
I think the 2080, 20 rule is, you know, sort of accurate in a lot of cases. You know, sometimes you have, you know, the editor slash reader, sort of dichotomy, right? The editor who are in it, who are constantly tuning and fine tuning things, those are probably more than 20% of players versus 80. So that’s kind of the benchmark I go through. But in general, it’s hard to say, right? Really depends on from a case by case standpoint. What I would say is, like, if there’s a certain critical mass of folks who really love the tool, I do think there’s something, something there, is how I phrase it. And I try to dig in more.
16:56
You know, we’re seeing some AI fatigue setting in across the enterprise sector, you know, forecasts are being revised down, and yet still lots of demand in the private markets, in the ventures sector. What do you what do you think we should expect from the buyer set, you know, on the customer side, and then, and then also the the investor set, going forward?
17:19
It’s really good question and also tough to answer. I think from a tops down perspective, it’s just so hard to predict where and when. Right, we’re gonna have that moment. I think it’s gonna be more gradual than say when chat GB, he first came out. You know, we’re gonna have this ai demand, meat supply, magical moment, I would tell you that you know, as somebody who spent a lot of time in software, the way I think about this is, so you look at S, p5, 100, which is the 500 most important companies out there, there’s probably single digit, right, not counting Microsoft, with those conglomerates, single digit of pure play software companies within that index, right? Which is just feels low to me, like just, I think about impact right, from productivity perspective, from from, you know, a lot of different angles, there should be more S, p5, 100 software companies getting created over time, and, and so, so I think, like, from that tops Down perspective, I’m certainly very bullish on that. And secondly, Nick your point about, hey, the dichotomy being private and public, right? I think one thesis, and why hypothesis I have is a lot of the AI innovation is truly happening on the private side. You know, there’s obviously a lot of other, you know, undercurrents around, hey, companies not going public at 100 million AR anymore. So you have more more companies staying on a private side. But I think the truth is, you know, whether it’s from talent perspective, whether it’s from a, you know, innovative dilemma perspective, there’s just more great AI, strategies, companies, products on a private side, then the public side. So that part, you can also see why, you know, sort of there’s a dichotomy between private AI valuation and paths to public valuation when it
19:11
comes to talent. Do you find that it’s increasingly more difficult to attract top AI level folks in this environment to companies, or is it becoming easier? Part of the reason I ask is because the supply side of top end talent is, is thin, right? It’s, it’s, it’s not expansive. And then also the public companies, there’s a lot of incumbents out there that can pay huge dollars for those, those types of folks. So how are you seeing your companies attract talent, and has it been more challenging or easier than expected? Yeah,
19:49
so this will be a pretty, you know, pretty gross generalization, my opinion, but if I just draw a line right so we have the true AI talent. We’re building large language models. I don’t think I telling my companies to go after those people, right, in rare cases, maybe. But, you know, we’re talking about a very small number of people who understands, knows, LLM, every lab on the world, you know, OpenAI, meta, Google, everybody’s going after those people. And in general, unless you’re building a model, which I would say, like for most private companies, it doesn’t make any sense to build a model you generally doesn’t make sense for you to go after those people. So there’s that dynamic there, and there’s a second group of folks who are traditional, great product people, who know how to incorporate AI into the design of your products, and it’s much harder to tell who are those people, because, you know, you can’t just ask and put a job description, asking for five years of LLM experiences. You know, chatgp came out in November 22 but I think like finding those really killer product people, who knows how to integrate, who knows what, where the product is going. Who knows, like, what models they want to integrate in your product is more important than finding the first group of folks right now, the other point that you touched on, Nick was, hey, all these very big firms are throwing dollars at those talents. I do think a very interesting thing, I think, like that, is where you probably see some personnel changes, right? You saw that at Gemini. Saw that, you know matters AI organization, a lot of times the tension is you have a VP layer, VP level people who’s been at meta Google for 17 years, and they’re more than qualified, by the way, great people probably don’t know them, just outside any observation, they came in to this completely new thing, right? And they’re making much more than the people who are actually doing the real work on LLM, and that’s a real tension, to be honest, right? Versus like you go into an open AI, that’s much more, much flatter, right? So that’s, I think, you know, from a talent war perspective, between the matters, the Googles, who have a huge balance sheet, and it started that’s building models. I think oftentimes, you know, that’s a very attractive thing for the talent, because those organizations a lot flatter than an organization where, you know, again, it’s really hard to throw a way to sell your skill, right? It’s almost like a sports team. You know, you have this rookie who’s playing a completely different game than a veteran player who won you five Super Bowls, like you’re just not going to cut this person who’ve done so much for you and paid his rookie more right from a salary scale standpoint. So I think that’s actually a very interesting heroin conflict that’s happening real time on the AI talent war side of things. What sort
22:40
of advice would you share with the founders out there that you know are running an AI startup? Maybe they’re trying to raise, planning to raise in 2025 you know, how should fundraising narratives shift and and, you know, amidst these investor expectations that are kind of evolving on the fly. So first
22:57
of all, I think all you need is one, one term sheet. Okay, so that’s number one thing. I wouldn’t go in too deep into, but all you need is one. There’s no point of collecting a bunch all units. One. Which makes that one so much more important, right? Which is where I think the unique thing about AI, and that’s my personal hypothesis, I don’t know how this will play out, is, I think the company would look very different at 1 million, then 10 million, and at 10 million, 100 versus the prior gen of SaaS, right? Like, you know, like, and that second time, much more shows up. Build the same product, next gen of sector that they’ve done before, and the company looks exactly the same. You know, at 50 100 down the road, this is a very, ever evolving thing. So I think, like, my best advice is probably pitch something that you truly believe in, right pitch the vision that you really believe in. You don’t need to pitch exactly what you have today. Pitch something that you really believe in, and find someone who truly believe in the vision, because I do think this journey is going to be more turbulent than again, a traditional SaaS journey, where investors show up quarterly and they care about five metrics and and afterwards, just just, just going off through races, right? I think like there’s going to be a lot more iteration, especially on the earlier stage of things, and given a velocity of where, where things are, I do think like it’s important for every entrepreneur spend the time and find folks who truly believe in who are truly bought in to their vision as part of the board and as their lead investor. Do you typically
24:27
establish that fit in the intro meetings and the the early meetings with the founders? Or do you think founders can identify investors who are bought into their vision from afar. You know, maybe by reading their content or or or other things. You know, what do you suggest for founders that are looking for investors that might, you know, share this, this vision? I think
24:52
those are great ways to ramp up right, reading about investor from outside and doing some references outside. And I. Don’t think there’s anything that replaces in person, interaction, shaking hands, more old school, but sit down, share a meal, talk about things that are not work related, right? And especially like, are they Steady? Steady enough, right? One of the things you know, in my opinion, the feature and the bug of private markets is, once you’re in, there’s you don’t have liquidity, right? It’s not like, Oh, you can get out. I mean, there’s secondary funds and whatnot, but, like, generally, you’re in for the long haul. And whether the company has a good quarterback quarter, that’s the person you have on board. So do you have a steady person or steady voice in a room? I almost feel like that’s that’s more important than you know, again, it’s harder to assess, but that’s probably more important than how you feel right in the moment of signing the term sheet, like whether it’s a big brand name versus nod. So that’s kind of, that’s kind of how I would think about it if I flip the table and sit on your side.
25:52
So Casper, let’s, let’s talk about cyber security a bit, right? It remains in Evergreen category, but with threats growing faster than budgets. You know what? Advice matters most for the cyber startups?
26:04
Yeah, I would say, hey, the items I said evolve matters in cyber. The two features and bugs of this gallery is really one is lower velocity, right? Which makes sense. You know, I again, I think cyber is more and more important by the day, but it’s not more important than your business priorities, right? Like, if you’re in the way of building your products, CISO, going after, you know, building products, you’re always going to fail as a CISO. So I think that’s why it’s a low velocity category. And people really value trust, and people really value sort of network effect, right? Like, all my peers use this product or not, so I think, like, that’s a very important thing to sort of remember. And two, is just the catalyst, right? In this category is also important, because oftentimes you show up in cyber people don’t rip what they have out. They’re doing a layer cake approach, right? Like, oh, we’ve got this thing and, and, oh, instead of written this thing out because they have 20% value, they could potentially, you know, preventing me from getting reach, let’s layer on new thing on top. So then the question becomes more, why now? Like, why is now the point for you to be be there, right? And, you know, RSA is around the corner in two weeks. And I would encourage everybody to go and walk around the RSA floor. You know, your chances of finally picking a great company is just as good as anybody who doesn’t have the context here, because there’s so many companies out there. And by the way, CISOs have such a such a hard job, right? They’re putting out fire every day. And you know, they’re not thinking sitting here pontificating on what the future of cyber is. So I do think that the company, the solution, has to show up and have a really compelling why now reason, right? And usually the why now is created by the product, by the category, but also by, you know, old school go to market as well, right? So again, just a very unique, idiosyncratic category in software and and I think like being aware of why now is very important for entrepreneurs to know. You know, speaking of
28:07
why now with with AI happening, of course, and in the cybersecurity, cat and mouse game always happening. Are you seeing certain methodologies or certain approaches and frameworks winning in cyber right now, and conversely, you know what? What are some maybe older methodologies in cyber that that could become obsolete? My
28:28
view is actually, the more things change, the more they stay the same. I talk internally a lot about sort of, my personal focus now is actually back to basics. I think, you know, instead of sort of chasing an AI application that’s going to be built maybe 345, borders down the road, who, by the way, your engine probably doesn’t even know what the architecture really looks like, let’s actually focus on shoring up, you know, the key things that we have today, right? The identity, the parameters, cloud security and make sure that everything that we should have cleaned up is cleaned up. And by the way, Nick like you’d be surprised to know that how many things that are out there that just not cleaned up, right? They’re just not threat vectors, but they could be a very important threat vectors. Vector as as AI, because more more prominent as attackers become more sophisticated. So I think this idea about getting back to basics is is really important to me, and I would probably not use use the word ops lead to describe some of the more traditional companies. Now I would say, you know, there’s probably too many tools on this layer cake that a new vendor could show up and potentially consolidate some of those tools, right? That doesn’t mean you have to show up and pitch a completely new thing that CISO hasn’t even heard about. In fact, I think that would you know like cyber buyers tend to be the most cynical people ever because they’ve been pitched so many times on things that didn’t work. So I wouldn’t actually encourage folks to go in and pitch AI all the time, right? Because for every good. AI tool, there’s probably 10 that didn’t work, or 100 that didn’t work. So I think that’s also another pretty interesting dynamic here as well. I would say, you know, if I say one area that’s becoming more and more important and interesting, it’s probably the data security side of things. Ai needs data to work, and majority of the time, you know, the enterprise data security posture is not where the age of AI needs them to be. I’m biased. I’m I led the last round of a company called Sierra that helps, you know, companies address, you know, data security risk and remediate some of the data loss issues. But, you know, I do think, like that’s, if I had to pick one that’s certainly, you know, an ever increasingly important issue in the age of AI love it
30:46
Casper in growth stage investing, one must weigh metrics, pattern recognition, market forces, you know, balance everything with gut instinct. What are the most critical factors at the growth stage that consistently signal outsized potential to you. And I know you’re going to have to generalize, but, you know, give us your thoughts. Yeah, we
31:05
had, we had to talk about this honestly. My answer would be, I do think a lot about competition and competitive intensity. And again, if I were early stage investor, I wouldn’t be thinking about it, because I feel like you, you know, you want to take the glass half full view as a growth stage investor, I do think it’s super important, right when I get in, not only I want a company to be 100 AR, I want a company to be 100 and growing fast. And sometimes the other players in the room kind of determine the ceiling of where this company could be, right, at least the first core product could be. So I do spend time thinking about the combat competitive intensity of the product, right? Oftentimes, like when I think about, you know, great companies have great promises that didn’t really achieve, again, still great outcomes, but didn’t achieve what they really wanted to achieve. Generally, there’s some sort of competitive, you know, intense competitive issues in a broader space, and sometimes consumer surplus, right? Like you have, you know, people going out here to handing out the software new category, essentially for free. Like, it’s hard to maintain a strong level growth. You know, if that’s the intensity you’re facing again, it’s not a static view. It’s just something that we consider, right? It has to be factored into other other things, products, founders and whatnot. And then the other thing that we had touched upon is because of this, and that’s my thesis too, is a lot of big winners actually not obvious at all early on, right? Like, you know, figma for design, or even get up back in the days for, for, for developers, right? Like, the long wisdom is, the wisdom that’s been around is like, you can make money from developer tools and and because developers are not willing to pay right for for a lot of tools, they’ll just go and download some open source software and so, so I do think, like those are the factors that perhaps made that sector last track it right? Well, the sector’s not defined creation, like nobody knows, so it’s hard for a team at Microsoft sit down here and be like, I gotta build a better product than Oh, what is the sector again? So I do think, like, that’s why some of the biggest winners are in sectors where, again, it’s either it’s infancy, it’s not defined, or it’s a category that people don’t think is attractive. And I do think that has a point linked back to the competitive intensity point as well. How
33:24
do you measure competitive intensity? You know, during a diligence process, I
33:28
do ask a lot of customers to talk to you again as part of the process. Process, we talked a lot of customers. You know, I do try to find who the end buyer is, and try to ask them, like, what exactly are you willing to pay? Right? Because I think you know the obvious answer I’ll be getting is, you know what you’re willing to pay, right? Like, if we two extra price, would you turn or not? But the last obvious answer sometimes is, is, Aspire the actual user who’s experienced the pain, right? Sometimes you those are two people, the people who are using the tool, who experiences the paying are different than the buyer, right? And that’s, you know, in some ways, again, plenty of common examples of why this could work as well, but not as ideal. If you have one person in the room who has the budget and who has the paying at the same time, right? Because they usually be like, yes, we will be willing to do it right, versus you have to go convince someone in the different department to buy the tool here. Because, you know, on the way down, right, when budget, you know, sort of contracts. You don’t have the issue of somebody who’s fighting to keep your tool doesn’t really have a budget. They have to convince somebody who has the budget right. So, so I think, like, the second level thing I try to tell is, if this tool has, you know, has this dynamic of the buyer also has the paying as well. Yeah,
34:51
it’s like the workflow issues versus the spreadsheet issues, right? Like, in a previous life, I sold an analytical device, and my buyer. He was, you know, plant manager for, like, a water facility, and he wanted much better data, cleaner data, accurate data, GPS, stamp, time stamp. But the users wanted ease of use and efficiency. You know, the the old model was very inefficient, and so it’s almost like you have two different customers for something like that, and it can complicate the process. That’s
35:21
right? It doesn’t mean that you can’t win. It’s just just more complex. One more thing to take into account, right? So let’s talk about AI agents
35:29
a bit. You know, feels like every other article I read is about AI agents these days. So, you know, big storyline is AI agents embedded at the application layer, are kind of the next big unlock, you know, give us the bull case for that thesis and and how does it reframe where enterprise value will accrue in the AI stack?
35:50
Yeah, so the way I think about this is, first of all that the common narrative today is AI agents could do things that human doesn’t want to do, or the toil that people don’t want to touch I think that’s sort of one you to look at this, because a lot of the software apps today, right, are those quad apps where you have database, you have a middle layer middleware, and you have an UI on top right, and then going point and click, upload my Gone transcript into Salesforce. That’s something that every sales rep probably needs to do, but maybe you can automate that in a better fashion, right, tagging it somehow. So I think that’s one promise of agent, which I think is playing out real time. Now, I actually think what’s more exciting is what is actually outside of this crowd app paradigm, right? Like, what about a 90% of things that’s not part of the SaaS app, that I can’t just point, click and drag, right? Like, for example, if you and I were working on a project that evaluates the competitive intensity of our competitive products, right, maybe you have your own SaaS app to go in to download, you know, sales data. I have my marketing data I have to download and today. How does work is I email you before I log off tonight, and you get a screenshot of, you know, the BI dashboard. I just screenshot it, and you will slack me again saying, Catherine, what about this bar? What does it mean? Right? So those are 90% of work, in my opinion, right? In an enterprise really, truly unstructured. So how do you evaluate, how do you use agent How do you build agentic workflows around the this and unlock those? 90% to me, is actually more interesting. Now, I do think it’s a harder business problem if you don’t try to define it in certain verticals, right? Let’s say you’re just customer success. You’re just sales versus a horizontal thing that does everything under sun. But I do think like that the bigger value, in some ways, and the value that’s hard for incumbent to attack, right? Like, think about how hard it is for Salesforce to try to get, you know, some sort of data access within servers now and then. First of all, they wouldn’t even ever, they wouldn’t build that product to begin with, right? So those are probably a bigger pocket than the 10% of you know toiled work that people don’t want to do. So
38:12
So glean is now a unicorn. You mentioned glean before, and they entered the Enterprise Search category, which was already packed with incumbents, but they still managed to find success and break out. Casper as an early investor, what signaled that it could not only survive but lead in kind of this noisy space?
38:32
Yeah. First of all, I do want to say enterprise search has been around for a long time, but it’s not really a sexy category, right? It is today, but it was not at all when glean was started. And keep in mind, glean started before 22 right before the LLM days. It is, I would say, like a universally recognized problem. But there just hasn’t been any solution that really sort of, sort of like scratch the itch or solve, you know, the bias problem, right? So I think one of the things that people, oftentimes, you know, don’t recognize about Glink and some of the other startups, right? We have a company called moveworks that’s pending but a $3 billion transaction to service now that people didn’t recognize is they are AI companies, but some of the core technologies they built were not AI related, right? They benefit greatly from the rise of LLM. But in glean case, for example, the retrieval architecture about, how do you get the right data information from the online apps? Those are not those are pretty hard to build right building that middleware to retrieve the right information real time at the right place, while looking through who is authorized to glean what information. That’s a very, very hard problem to solve, and to me, AI is more of unlock an accelerator to their problem versus a direct head to hack. Petition right now. To your point, like, you know, search was one of the greatest use case ever for internet. Of course, the first set of use cases that are gaining traction in in in the consumer and enterprise AI space is search. That’s why you’re seeing more and more folks coming. But I do think, like one of the things that you know, maybe a little more hidden from the general public is the fact that they have a really great, you know, retrieval architecture, and they spend a lot of time and energy building that prior to the LLM era,
40:30
amazing Casper, how are recent marketing market conditions impacting your investment strategy? Well, we It
40:36
doesn’t. It doesn’t right. It doesn’t because we take a 10 year view, and we want to partner with the entrepreneur to very meaningful exit event, right? But it’s, you know, IPO, or hope, you know, hold the the stocks post public for many years, or a very meaningful M A event. So we take a very long term view and, and so the recent event, you know, just like, I don’t know, every year there’s something that’s that went on, right? Like 2021 this COVID Back then, 2022 2023 right? Folks, remember, software is that, right? So there’s all kinds of events going on, and we try not to to pay specific attention to one thing or the other, but that’s sad, right? We do, you know, we’re investors. Your entry price does matter, and we do take the view that, you know, in a given time and in a given sector, the broader macro condition and public condition has to matter in our decision making process. So we do factor those in when we think about pricing the deal and whatnot. But, I mean, there’s not a single sector that we would say, hey, like, we just don’t touch because of tariffs, I would generally say we do. We do not touch secular losers, right? We do not touch the sectors. We feel like a losing because AI, those are the textures we don’t touch. But, you know, I look around software, I feel like every sector will be in some shape or form, reinvented by AI. So it’s hard to say, like, Hey, here’s a sector that you won’t touch, right? If anything like, we probably won’t touch certain companies that are not well positioning those sectors.
42:07
Are there any fundamental changes in the way that you price deals now versus maybe earlier in your career? I mean, pricing always fluctuates. You know, there are multiples in the public markets that some look to and but I’m looking for more like fundamental lessons or learnings that really inform your pricing strategy today.
42:27
Man, great question, but also very hard I do think again once case by case, right, it’s easy for me to kind of put out a one liner to you Nick and just say pricing actually doesn’t matter. If you pick right or pricing really matters, because enterprise always matters. But I think the truth is probably somewhere, somewhere in between, right? And, you know, just, just by law of physics, pricing has to matter. But on the other hand, if you partner with the right company, there could be multiple points or multiple rounds in the company’s life journey that you can, you can make money, right? And and, by the way, the other thing that people don’t talk about is venture investors are terrible, terrible people. From a pricing perspective, it’s hard to price, right? We compare. It is not great at pricing, which is VCs, right? I myself as a culprit, I’m not good at pricing, versus your PE. Folks who are modeling everything down to a te or public hedge fund, folks who have liquidity, right? They’re like, Oh, now your stock is 12 times. I gotta sell this venture. Folks are not great at pricing, because you are the marginal price taker, right? For some of the really great, yes, right? So we try not to get overly Cute, right? But in some other cases, you can find rounds where we would internally call money rounds, where we feel like, you know what this is for this company, which will rise. This is the best round to get in. Sometimes when companies 2 million are sometimes when the company is 50. So that’s where we spend more time thinking about we’re just saying, like, you know, here’s gonna be coming in. Here’s a waterfall chart. Let’s plug in three numbers and the pricing come out. Because we have to be aware, right? Like we talked about, we had to be aware of competition around us. And, you know, myself included, my colleagues, my partners included, I don’t think VC is good at pricing at all. And then, by the way, that’s not a bug for what we do, right? That’s more of a feature in our ecosystem. In my opinion, 100%
44:27
Yeah, the marginal benefit of 20% probably doesn’t matter if you pick right, but if you’re picking wrong, you’ll be out of business either way. Casper, if we could feature anyone here on the show, who do you think we should interview and what topic would you like to hear them speak about?
44:43
I would feature a public investor to talk about their view on what is interesting to them from a private perspective. And you know, if there are companies that are going public today, like, what are the entry names? What are the interesting traits that we’re looking for, right? Like, you know, obviously, there’s obvious ones, like Shri per Databricks, but what about everybody else? Like, what are the interesting names that they are excited about, sure, you know, holding shares in and why? What are the key characteristics?
45:15
Are there any public investors or analysts in particular that you follow closely?
45:19
Yeah, there are. I certainly have friends, and we’ll follow up after the show. Perfect, interesting names, very good
45:24
Casper, what book, article or video would you recommend to listeners? For
45:28
software, I’ll just stick to investing, because for software, one of my favorite book of all time is Larry Allison’s autobiography. It’s got soft war. You can probably get a reprinted version on Amazon or somewhere, pretty easily. And the thing that I learned the most about BBB is everything that had happened today, right, the MCP that folks are talking about that’s so hot in AI, the idea about coordinating different systems, the idea about single pane of glass, all those things were tried, you know, 2030, years ago, right by all kinds of different companies, and they, you know, they’re right or wrong for all kinds of reasons. So I think, like digging deeper on the history of lessons of what happened then, is super interesting. And you get a sense of, like, what happened, right? Like workers stock was down 4050, after quarter. Like, what actually happened there, and I think that’s a very good way of getting a glimpse of history and little more about software perfect
46:27
Casper. Do you have any habits, tactics or behaviors that are a force multiplier? Oh, I don’t.
46:33
I don’t have an unconventional one. I wish I’m doing, you know, I don’t know intermittent fasting or something. I’m not doing any of that. I would say, I think the biggest thing I remind myself this says, is I just try to take it one day at a time, right? I do think it’s impossible to not have a bad day, but the most important thing I want to avoid is having a bad day but not being productive doing that bad day, right? I do think those days count, and I believe in momentum, in startups, in life. So you know, if anything, I would just say, hey, take it one day at a time, right? The long term vision is important, but you got to take it one day at a time. And sometimes there’ll be a lot of this woman, if your goal is is grand, your vision is grand, but you got to treat every day as a separate, new day and and take it one day at a time. I like
47:19
it. And finally, here, Casper, what’s the best way for listeners to connect with you and follow along with Sapphire?
47:24
You have websites, email is just Casper at Sapphire ventures.com. You know, I love investing. I love software. Love talking about, you know, the history of investing in history of software. So if you just want to chat about investing different forms, I didn’t have to venture at all, because that’s what I do every day. But you know, from top of Publix, talk about privates, talk about cash flow businesses, I’m all yours.
47:47
Perfect. Well, Casper, this is a huge thrill for me. Thank you so much for joining us and sharing your insights with the listeners. I really enjoyed it. Thanks for
47:55
having me here. Nick. Really appreciate it.
48:02
All right, that’ll wrap up today’s interview. If you enjoyed the episode or a previous one, let the guests know about it. Share your thoughts on social or shoot them an email. Let them know what particularly resonated with you. I can’t tell you how much I appreciate that some of the smartest folks in venture are willing to take the time and share their insights with us. If you feel the same, a compliment goes a long way. Okay, that’s a wrap for today. Until next time, remember to over prepare, choose carefully and invest confidently. Thanks so much for listening.
