Vertical AI Deep Dive with OMERS Ventures

The Cloud Returns Podcast covers all types of software investing, whether seed, venture capital, growth equity, private equity, debt, and even the public markets. Excited to welcome Omerz Ventures to the show. Marissa, how about you introduce yourself and then we'll let Taco introduce himself and tell us a bit more about yourself and your firm. Awesome. Thanks for having me and my colleague Taku here. We're excited to be here. I've been on the Omers Ventures team here for a little over, or almost coming up on 3 years now. I spend all of my time on vertical software and AI. I sit in our Bay Area office in Palo Alto and, you know, come from a variety or a varied background in financial services and market research. Started my career on the equities trading floor at Barclays. Ended up going into equity research where I covered large-cap industrial machinery, so John Deere and Caterpillar and Stanley Black Decker, and then went from there to cover private, you know, venture and tech landscape at CB Insights where I covered digital health and was there during the height of the pandemic. So it was exciting time to be there. So that's a little bit on me. I'll let Taku introduce himself and he'll tell you about Omers. Yeah, thanks, Marissa. So I've been at Omers for for, I'd say, going on 2 years now. Prior to joining the fund, I was on the other side of the table. So I was a founder, founded a fintech startup specifically in the wealth management space with 2 other co-founders. We did that for a bit of time and ultimately decided to go our separate ways. So one of the co-founders is still running the business. The other one has gone on to, to found another business in the blockchain space, and then I I found my way over to venture and pretty much since joining the fund, my focus, I think the focus was initially vertical SaaS and then we've since expanded that to vertical software. So, you know, I work very closely with Marissa and yeah, I was going to say most, pretty much all of my time is vertical software, vertical AI. And then a little bit of background on the fund. So OMR is the direct early stage tech investment arm of The Ontario Municipal Employee Retirement System, which is a Canadian pension fund with about $124 billion in assets under management. And what that means is that a huge portion of our LPs are our first responders. So the folks we invest on behalf are firefighters, police officers, nurses. And so there is that added sense of meaning and purpose for us when we're finding the best companies to invest in. We have a little over $2 billion in assets under management. So this is Omers Ventures separate from the pension fund. And we're currently investing out of our fourth fund. Fund is about a $750 million fund. And we have 4 thematic areas that we focus on, which are horizontal SaaS, vertical software, we've mentioned infrastructure software, and then fintech/insurtech. Last 2 things I'll say is, you know, we typically lead or co-lead Series A to rounds with check sizes ranging anywhere from $5 to $25 million with some flexibility to go, to go up or down. And then geographically, Marissa is in the Bay Area right now. I'm in Toronto, and then we have an office in New York. I think that covers it. I don't know, Marissa, did I leave anything out? No, I just thought maybe I would double-click on some of the areas that we're spending time in, in vertical software in particular. We will look at vertical software and AI targeting any industry, but our current thesis area is around software that's still defining resilience in 3 big areas. So first bucket is public health and safety. So this is obviously healthcare, life sciences. It could even be law enforcement, maybe a little bit of gov tech. Second bucket is industrial and supply chain. So anything manufacturing and broader supply chain, freight and logistics, transportation. And then the third bucket is critical resources and infrastructure. So this could be like oil and gas, power and utilities, telecom, et cetera. So any software that's empowering those organizations to be more resilient and respond to the chaotic world that we're in today. And yeah, respond to change and new datasets. That's what we're interested in right now. Great. And I know there's a wide range of investments your firm takes on. So maybe for your immediate practice area, what are some investment criteria in terms of ARR, check size that really fits with the the two principles we have today? We try not to be too rigid on the metrics front. What we really care about is obviously founders, and we can get into a little bit on what we like from founders, especially vertical software. But a team that really knows the space that they're building in has a unique kind of lens in which they are approaching the world, whether it's from a product or a go-to-market perspective. And the minimum criteria that we need to see at a Series A for us to get, you know, really interested in investing is clear evidence of product market fit. We don't necessarily need to see everything go-to-market sorted out. We need to have like a glimmer, a little green shoot that there is an inkling there on how they want to scale the business. But first and foremost, we need to see really strong customer love of the product, great testimonials, and obviously retention and expand, like expansion and land and expand are, are great opportunities to show that people are valuing the product and willing to pay more for more features or added capabilities. But there's a lot of different ways that we can slice and dice that and evaluate if a company actually is found product-market fit. It's not necessarily strictly just ARR-based, but in general, I would say, you know, at the A, we would love to see someone or a business that's around $2 million. I think it's the normal mark now these days. Yeah. Perfect. And where do you want them to get over your investment horizon? I know some people have like a threshold. We want to get to $100 million of ARR in 9 years. Or 7 years, or any context you can give there too on, on the scale you're ultimately looking for your companies to achieve? Yeah, I mean, I think that we're in the venture game, right? We're in the game of outliers, and what we want for any investment that we make is to have IPO potential. And what IPO potential meant several years ago was $100 million in revenue. I think that bar has changed. Maybe it will come back down, but we want to see a business that can at least get into the $100 million revenue mark and hopefully control their destiny from there profitably. But that's kind of the baseline of what we're looking for. Awesome. And this context is always so huge, right? When, when people are listening, having the context around your Series A firm and that's what you're playing for guides a lot of these discussions, right? Because if you're looking at something more niche that you could be a little bit smaller or slower, but like you guys are playing for IPO trajectory companies. Yes. Yes. And with that in mind, what type of founders do you prefer most? We'll go with that first. Sure. I'll start. I know we both have thoughts on this. In vertical software in particular, we have found great success and what we really gravitate towards is a combination of, usually it's a two-person founding team, but a mix of really deep industry and operating expertise. And then some, and then also complemented with technical acumen, the industry expert or experienced operator on, you know, one half and the technical co-founder on the other half. That's, that's what we like the most. And by the time they get to us, we also like to see a strong go-to-market expert on the team, whether that's a founding executive or not. Like I said, they don't need to have completely sorted out what the go-to-market motion is going to look like, but we like to have that expertise on the team so that when they're ready to really scale up and go after that opportunity, we already have confidence that they know the market well, both from the operating side, the technical side, and then now the go-to-market side. There are sometimes exceptions though. I mean, sometimes there's a founder who may not have strong experience scaling a B2B SaaS business but has a really, really strong industry, you know, background and a deep Rolodex that can get them pretty far on founder-led sales. And by the time they need to bring in a go-to-market leader, that person's job is less about getting like the name of the business out there and filling the pipe, etc., and it becomes more about how to build structure and incentivize the go-to-market organization so it operates more like a world-class SaaS business. So I'm not saying necessarily that like, and then in the founding team, we need to have a founder that both comes from and oftentimes very legacy industries where there aren't B2B SaaS experts and have B2B SaaS expertise. I think that person, that expertise can be brought in later in some, in some instances, but in general, we like the, the 3 ingredients there, the industry, deep industry expertise, usually having had direct experience working in the industry, technical acumen, and then like a really strong go-to-market person. Awesome, awesome. Taku, what are your preferences, if they're any different? I think Maurice and I are very much aligned on that. I think if there's one thing that I would add is part of our approach is when we are spending time and trying to understand the inner workings of a vertical, we do a copious amount of— there's desktop work, but then there's also foots on the ground talking to industry experts, founders, understanding some of the problems that are within this space. And I think You know, one of the things that gets me particularly excited is just when founders have a unique approach or unique ideas to how they're solving this problem. And I find that comes from really being a student of predecessors and folks that have gone before them and have tried to solve the problem. So having a deep understanding of what's worked in the past, what's not worked in the past, and you know, why the approach is different and why they believe that the solution can only be solved this way, I think is probably the other thing that gets me super excited. But, you know, I think Marissa and I are aligned on everything that she mentioned. And this is always helpful for founders because they have a lot of curiosity about your process. And sometimes it just helps to hear investors actually describe these things. Is how do you do due diligence on the founders? I wish I could say that it was the same way every time, but it's often, it's different. It's case dependent, but in general, it's a lot of referencing. Like it's, we try to uncover anyone in our immediate networks who has worked with this person. Whether they— maybe they're just socially related. I'm not saying we're having like hundreds of calls, but we try to get to the most high-intent, highly relevant people in our network and maybe our extended network that have worked with them, that have been board members, that have been, you know, bosses of them in prior jobs, maybe are former employees of the company, just to get a well-rounded perspective. And obviously talk to all these people independently just to see what patterns emerge. And then once those patterns emerge, we can address it head-on with the founder or with people who are closer to the shaping of that individual as a leader, whether it's, you know, an executive coach or one of the other board members or something. But it's a lot of referencing is like the short way I would, way I would, uh, describe it. Taku, I don't know if you've got anything you wanna add to that. No, I mean, like, that's spot on. We do a lot of referencing and, you know, I, I think you kind of outlined everything there. Great. And, and Taku, you made a, a reference to the predecessors and we had a, a prior guest on about, one differentiator they looked at in founders was people who've gone through the graveyard of failures. Because inevitably in any of these categories, something else has been tried almost always. And that the differentiator is the ones who look back 5, 10 years ago and track down a prior founder and got some unique insights. Is, is that a pattern you've also observed and kind of resonates with you? Yeah, I think, you know, The assumption that you make is that, you know, when someone has on their— when they're on their second or third company, that they've learned from prior mistakes, right? And so that's the assumption is so there is a natural bias to want to back someone who's had the experience and kind of gone through it. That's not always the case. You know, sometimes there could be market tailwinds that are in your favor. There could be a lot of things that necessarily helped you succeed that you can attribute to your own skills. So It's not always the case, but I mean, again, it's not always through your experience building a company. I think there's a lot to be learned just by being a student and understanding what's been tried and what's been done before. So I'd say a combination of both. It doesn't have to be one or the other, but just showing that you have spent the time to try to understand what has worked and what hasn't worked, I think is very valuable regardless of how you got there. Yeah, I think just to add on it, because Matt, I know you were going to ask a little bit about the Green Room, which is a space that I have created to help founders with their pitch. But this is something that comes up quite often, actually, is not necessarily learning from their prior mistakes, but coming with an awareness of the competitive landscape. We like to see that more and more because with AI now in particular, it's becoming easier to build businesses. And so spaces are becoming a lot more competitive. It's harder to differentiate on product. And so, and, and In our view, in a lot of vertical software arenas, it's going to be where you differentiate is a lot more on go-to-market. And so coming to that pitch and coming to the first introductory call with a VC and saying, look, I understand that there are 10 of my competitors out there. Here's, here's how I break them down. Here's how I compare to them. Like, is showing us that they're a student of the landscape and they're aware of where they fit in into that landscape. I think that's becoming more and more important these days than it probably ever has been in the past because there's just less white space to build. And so, yeah, I think someone who comes with that awareness is a, is a major plus. That's an awesome point. And maybe you could give an overview of what the Green Room is. Yeah, happy to. Sorry, I kind of jumped the gun there. Yeah, the Green Room, um, it's a space that I created for founders to practice their pitch in front of a small handful of VCs. It's sometimes industry leaders or their fellow founders in a low-stakes, judgment-free environment. So it's basically a closed-off Zoom environment with 3 founders and 3 panelists. Usually they're VCs, but sometimes I bring in industry experts. Strict confidentiality is enforced. It encourages the honest, blunt conversations around what makes a good pitch a good pitch. And also, you know, sometimes we get into topics that are beyond the pitch itself, and it's just a place for founders to see in real time several VCs and hearing the same pitch at the same time, get their reactions and kind of form some sort of consensus on certain aspects. Because one of the things that I was hearing over and over again from founders before I started this was that one of the most frustrating parts of fundraising was that they would create this pitch, the deck and the narrative, and go out and pitch it and then get a bunch of feedback and have to change it and then go talk to the next person and have to change it. And it was just, I mean, it still happens. It's always going to happen, but it's actually been pretty shocking and surprising to me to see in the 20+ sessions that I've done so far, we're more often aligned than we are not. And so having that real-time feedback for founders to see, like, okay, all three of these VCs who invest at different stages but all invest in the same sector, you know, are critiquing me on the same things. And so therefore, that's probably the top priority thing I need to fix in my pitch. So that's the intention of the space. It's unique because each session is focused on a particular topic. So it could be supply chain, it could be, you know, healthcare, or it could be fintech. And, and the panelists are selected for that particular sector. So there's, they're active VCs who are investing in that space and have probably seen all of the competitors that that company is, is pitching. And then it's unlike, unlike a pitch competition where a founder might get like maximum 5 minutes to give their pitch, this environment gives them about 40 minutes to do the pitch and get the feedback. So there's 3 founders in the room that all get to learn from each other and the feedback that they're getting on each other's pitches. So it's been a really special place, it's been exciting, and we hope to bring it to more founders going forward. Awesome, that sounds like a great initiative and a real need for that. And what are some of the common mistakes, or we'll put it more positively, things that improve as a result of being in the Green Room? Honestly, the, the most common areas of feedback are not all that groundbreaking. Like, there are things that you hear all the time, but people still make the same mistakes. It's a lot of times for pre-seed and seed companies, sometimes even Series A companies, they wait until the very end to talk about their team. And that the team is just such an important part of what an investor at those stages is making a bet on that we always try to tell them, you know, there's mixed views on this. People have different preferences, but I still think, I firmly believe that you need to talk about your team up front. We need to know right at the beginning, like, why we should be listening to you as an authority how we know that you know what you're talking about, because it sets the tone for the rest of the pitch. And I think the team is a really good signal of the type of leader that, you know, we're gonna potentially be working with. So I think setting the stage with that up front is, is very important. Not enough people do that. A lot of founders also over-index on talking about the problem, and this is going to vary, you know, depending on what type of VC they're talking to. They're talking to a generalist and they're building in a very specialized space, maybe it makes sense to spend a little bit more time on the problem. But if they're talking to a specialist in that space, that's been a very common mistake. It's, let's not over-exaggerate the problem or make it more macro than it needs to be. Let's really distill it down to the actual problem that your solution is directly solving. Like, if you can break that down, it makes it clearer where your solution fits into the, into the world, you know, in 10 years. That's, that's what we're all trying to get to, is understand like what the world looks like if your solution is adopted. And those focusing too much on the macro, where there's a lot of other externalities that come to play, makes it harder to see that connection. So that's a second one. And then third one, which is kind of related, is not enough founders like have figured out the mix between balancing the big vision and then the short term. So you have to talk about the big vision to get people excited about the investment opportunity, right? Because we're going to be in it for 10, 10 years. But you also have to be very specific about what your, the use of funds are going to be, because we want to know that you're going to be good stewards of the capital and that, you know, you actually, actually have a plan in the short term to meet the next milestone for the next fundraise. So there is a delicate balancing act there that you have to do, and a lot of times founders skew one way or the other, but you really do have to touch on both. And then the last one, this is kind of more tactical, but, and it kind of relates to the third one I just said, which is like not enough founders use timelines. Even if you're at the pre-seed or seed stage and you can't talk about like traction timelines, like just having a, a slide that says like, here's what our Act 1 is going to be, and here's what Act 2 is going to be, and here's what Act 3 is going to be, helps bridge that gap between like the short-term and the long-term. And I think just as investors, we're used to seeing things in trend lines. And so the more and more that founders can utilize that kind of schematic on their slides, I think helps illustrate what they're building. And not enough people use those, but maybe it's just me. Maybe I just love timelines. That sounds very powerful to me. And, and maybe to go a little deeper, is that even like a use of proceeds? Like if we raise this Series A, here's what the next 24 months look like. Totally. Yeah. I mean, it's not everyone is going to have that granular of a vision. I mean, like, especially for a pre-seed company, if they don't even have a product yet and don't know how they're going to monetize it, it's kind of unrealistic to ask them to put a revenue target on there. But you can say like, if here's our goal for, for Act 1, and if we accomplish that, let's just put some ballpark estimate out there on the number of customers we could go after and a rough estimate of what we think the value our, our solution provides. Like, here's the opportunity that we could go after in Act 2, not here's what we're going to accomplish, but here's like the prize that we could go after if we find product-market fit. Like that, I think Showing how act 1 will give you the right to move into act 2 and how act 2 will give you the right to move into act 3. Like that's the story that we want to hear. It's not necessarily always about the numbers. That's awesome. Taku, do you have anything to add to the fundraising learnings? Well, I mean, like, to be quite frank, I wish the green room existed when I was fundraising. I think I probably, I was on like version 60 of my slide and feedback. So I mean, like, I think Marissa's platform is incredible and She touched on everything, right? Like I think timelines is very big. I think helping people understand like, you know, where you expect to be because as investors, you know, venture capitalists, you're always, you're kind of working backwards and you're trying to figure out, okay, you take investment now, 18 to 24 months, where are you going to be at when you're potentially going out for your next financing round? So I don't have a whole lot to add there. Awesome. Awesome. And we will include links to the green room and opportunities to participate in all of the show notes and promotional materials. And then, you know, the real genesis for this was seeing your incredible vertical AI deep dive. I believe it came in at 152 pages. And yeah, I did not want to insult you, but I did immediately wonder, did AI help with that? AI definitely helps. I mean, it was not all of it. The benefit of working on a a big organization like Omers that invests across a variety of asset classes with $130 billion in net assets means we have a lot of internal experts that have expertise in, in a lot of these industries. So that was the starting point. And also just the— our team, we have an incredible team that's been investing across a variety of industries for a long time. So the starting point for all of this is like pure grassroots, you know, just learnings on the job from people that we work with, and then Putting it together and actually stitching it together into something that we could look at these industries comparatively based on qualitative aspects that we had gathered from our own research. Like, that is where AI was helpful for sure. We were like, you know, here's all of our thoughts on retail with when it comes to this criteria. Here's all of our thoughts on manufacturing when it comes to this criteria. Like, which one of these seems like it would rank higher? Like, we used it as a validating tool more than a, a writing tool. If that makes sense. For sure. And again, I thought it was phenomenal and it was just making a little light of 152 pages. That is, that is quite the production. But what I liked about it, and I think it's incredibly important for AI, is to get like very, very, very specific instead of generalizing and doing top-down and just assuming it might work or might not work and getting excited or getting disappointment is you actually have to go bottom. Up, and you guys did a great job with that. And I particularly like the way you categorize things into 8 core factors. And so would either of you want to just lay out those 8 factors first before we get even deeper into everything? Well, first I'll just start a little bit of context on why we did this, because we all know AI is drastically changing many job functions across a lot of industries. But we, having been students of vertical software for a long time, have known verticals aren't equal., and not all of them are going to be as equally ripe for AI, vertical AI, right? We're, we're talking about AI that is very, very industry specific, not like your horizontal, you know, document drafting tools. So we wanted, before we go and boil the ocean and look at every single AI targeting any industry, we wanted an objective view of where actually do we think is the most ripe from a development point for emerging startups to tap into existing data. Build something that's very industry-specific. This is not necessarily tying it to the value and the potential ACVs, or even if the go-to-market is efficient, right? This is purely from a developmental perspective, like where could vertical AI, where could those opportunities exist? So yeah, we, we spent a lot of time talking with the team about what criteria we thought made sense. Obviously it all starts with data, and not just unstructured data, but the complexity and the industry specificity of that data is very important. So that was the starting point. And then it was, okay, well, it's not just one thing to have the data available or the data be out there somewhere in the ether. A software vendor, an emerging entrant, is going to have to be able to access that data somehow, whether it's publicly available or it's available for licensing. There's no vertical AI solution to be built if you have no access to tapping into that data. So that was the second one. Then the third one was, there's another angle here, which is Okay, maybe there isn't an existing data pool for you to tap into, or maybe there is, but you might just by nature of having your product out there in the market in customers' hands, be able to generate your own proprietary dataset that's very unique to those users and their workflows. So you almost more metadata, but you, there could be things that the customers are contributing to your platform that aren't necessarily core to their strategic positioning or competitive positioning in the market. That they feel free to, to contribute to your model. So it could be things like, you know, supplies that they— in life science land, it could be like lab supplies or something, right? Like it could be a product catalog that they get from their suppliers. Like that's nothing that's really going to separate one pharma company from another if third-party developer has access to it. But that third-party developer, if they get that information from enough life science companies, then that solution becomes all that much more valuable to the industry. So there's that aspect of it too. So the third criteria was the software vendor's ability to amass and close off access to proprietary data. That's kind of what that situation would look like. The fourth one is the uniqueness or industry specificity of the workflows and the jobs to be done, right? The work of a, you know, a dental hygienist is going to be very different from the work of a paralegal, right? And in some industries, you know, maybe the receptionist job is very similar from industry to industry. So we looked at You know, how specific and industry-specific are the, the jobs to be done in, in the industry? The next one was the degree of regulatory oversight and burden of compliance. Obviously, because there's a lot of paper trails that are involved with this and very low tolerance for inaccuracy, a lot of fines that can come from that, and just a heightened focus on risk. We think that there's some interest industries that are going to be very heavily invested in AI for regulatory workflows. This kind of comes to the next one, which was like the degree of high-stakes decision-making, like where there's really low tolerance for errors, where humans might be more, more prone to making errors. Like we see opportunities for AI to lean in on that. And then the last couple ones we are, we didn't weigh as heavily as the, the prior ones, but we do think are, are still important to consider. It's the importance of personalization. So AI definitely has the capability of creating more personalized experiences than it ever has in the past. So importance of personalization. Second one was the prevalence of simulation and, and synthesis workflows. Next one was prevalence of exploratory analysis workflows. And then the last one was the importance of creative content generation. So, you know, all the video and image generative AI content that we're seeing, like that has a role to play in, in some industries more than others. So that was a long list, but hopefully me walking through it kind of illustrated how we were thinking about it when we came up with the criteria. I love it. And the type of people who listen to this show love it. So this is, this is a good host-guest market fit situation. Um, and diving back to the data part. Yeah. How important is the unstructured aspect of it to making AI like a differentiated reality, like compared to traditional software? Oh, I mean, I, I think it's a huge part of it, but I think when we're talking about vertical AI, vertical AI as opposed to horizontal AI, the thing that actually mattered potentially more to us was the industry specificity of it. But what often, what we ended up finding was that those things are often correlated because a lot of things that are industry agnostic have been standardized in, in some ways. So things like, you know, chemical and molecular structures or genomic sequences or medical images or CAD files for manufacturing versus architecture, like those are all examples of both unstructured and very industry-specific datasets. So we do see a correlation. I wouldn't say they're 100% one-to-one, but I think they're equally as important, unstructured and industry-specific for Vertex AI. And then two of the key factors were around the vertical AI's ability to access the data. Yeah. Like get the permissions, get integrated in a way that they're able— what are some learnings you've had from looking at successful companies? Maybe some examples of tactics that allow you to break through or like gain the trust of customers, something just so we can understand this one a bit better. Yeah, I mean, quite honestly, I think we're still in early innings of learning. I think there's a lot of hope that, well, the first thing I'll say is that there's a lot of companies and businesses out there that are now realizing that data that they've been sitting on and aren't using, and it's not, you know, necessarily core competitive advantage for them. It's now an additional revenue stream. So I do think more and more companies are going to be licensing their data for the purpose of AI training, right? It's, it's kind of a, a freebie. It's an obvious one. So that is one trend that I think will continue. As far as tactics in getting customers to contribute their data to your model, I think there are— we're very early in learning, and there's definitely industries that where that's more taboo. Obviously financial services, healthcare, legal, because a lot of the data owned by the law firms is actually isn't owned by the law firms, it's actually owned by the clients. And so any industries where there's like very strict, you know, security parameters, like that's that's a hard thing to get around. And I think we're still learning what that looks like. But in other industries, or maybe even in those industries, if I think it's about what data is being shared, like I said before, if it's something that's not, it doesn't have, you know, client confidential information or anything that would get you in trouble with regulators, it's not core competitive advantage to keep it to yourself, but it might contribute to the betterment of the model itself or or the, you know, the product that you're using to create a competitive advantage, then I think there's been openness to sharing some of that data. But I think we're really early in seeing what that landscape looks like. There's a lot of hope, but right now I think we're still in this world where anyone that's building a mo— not anyone, but a lot of companies are still licensing the data. It's like, we're willing to pay for the data in the meantime, and then hopefully we'll build up a good enough product that people contribute to it voluntarily over time. But I think it's very early. I don't know if Taku, if you have anything you want to add. Yeah, no, I agree. I think you had two good points there. I think the point on companies sitting on a ton of data that, you know, they'd not necessarily thought was useful. And I think the other thing is just data coming from usage, folks using different platforms and kind of generating that data. But we're definitely in the early innings at the moment. The, the one thing I will say that is a tactic, I don't necessarily know necessarily if it's like working. But one common tactic we see is like your stereotypical design partners. Like if you're able to get a few marquee design partners that are sitting on a lot of data, this is common in healthcare. So, you know, companies will partner with like a couple health systems that have enough data to represent a general population. I imagine a lot of creative contracting structure that you have to come up with to incentivize sharing that data and what they get back in return. Oftentimes it ends up involving an equity investment. But if you can land them, it's about making sure the incentives are aligned so that you can train on that data and that it's exclusive to you. But like we said, in the long run, I don't know how frequent exclusive data relationships are going to be when a lot of companies realize that they can monetize their data. That makes sense and sounds similar to your design partner, but are you seeing much give to get? I know a few where they have a you know, there's some value in, in the data product they're producing, and that your way to get the insights and the benchmarks of your peers is you have to contribute your data. I've seen some people use that as at the earliest stages to get around any data licensing problems. Yeah, I think we've seen it in some cases. I think it depends on the application and what data is being shared, right? But yeah, I've seen it in some cases. For sure. Awesome. And then this next one, we'll go back to the factor, understanding the intricacies of industry-specific workflows and jobs to be done. And when I read that one, I thought about founder market fit. What are your takes on how this relates to the founder and the management team attacking the market? Yeah, I mean, largely I, I say I would, I would agree with you. I think this is an example of founder markets. I think founders who possess a passion and the curiosity to deeply understand jobs to be done and the intricacies of industry-specific workflows are always uniquely positioned to succeed in building vertical AI solutions. And typically those come from industry insiders. There are some exceptions to that rule, but for the most part, it's industry insiders. And so yes, I would largely agree. Typically a founder-market fit dynamic. And I think, you know, when you think of it, like the deep understanding, not really, it goes beyond just allowing them to create tailored solutions, but it also helps them talk the talk and, you know, building trust with key players and stakeholders is hugely important. So yeah, all that to say is I definitely agree with you. Awesome. One that kind of jumped out at me, degree of high stakes decision making. That was a little bit of a surprise. Could you elaborate more on, on that factor and kind of some of the thinking there? Yeah, on degree of high-stakes decision-making, it was really about, again, this idea that AI, I mean, it is kind of a double-edged sword one, right? Like you don't want to necessarily give AI too much authority in something where there, it could potentially put someone at harm. I'm thinking in examples for like public safety, for example, like you might not want to have AI completely be the 911 operator. And so this is, I would say, more in a copilot context than in an agent context. We think AI copilots can be really, really great in high-stakes decision-making workflows. They're augmenting the ability of that person who is overwhelmed with the amount of data. They're probably getting more and more data each day that they need to filter through to make these really important decisions. Whether it's for actual, you know, health and safety, whether it's for financial security, or whether it's for regulatory reasons. Like, these humans who are human and make errors are tasked with really important jobs, and we think AI can help them focus on just those tasks where they really need to be fully on. So that's— I think about it more in a Copilot context than in an Asian context. I think we would be very cautious about an AI agent making those decisions completely autonomously. So we didn't write the report in totally like a Copilot versus agent world, but I think with this one in particular, it's very much more tailored towards the Copilot situation. Perfect. Yeah, I, I really appreciate that distinction and it makes a lot of sense. And then I'll read two more of the factors. Prevalence of simulation and synthesis workflows, and then prevalence of exploratory analysis workflows. And when I think about that, there's the obvious, like, founder market fit. You really need to understand what exactly is happening in these industries. And then the other aspect of this is, is the length of time it takes to really know that, even if you have a background, but like work with a diverse pool of customers and really build up that accumulated knowledge and understanding where the corner cases are. And with that in mind, right, it seems that some of this vertical AI development will take some time. And with that in mind, how long do you think it takes to go from seed to Series A when you start incorporating factors and development timelines like this, like where you're ready to scale? After establish the product? That's a good question. I mean, I have my take, Marissa, like I'm curious to hear your thought, but how long does it— I mean, like first to address your point on just simulation synthesis factors and exploratory analysis, like those definitely exhibit, uh, under-market fit dynamics. I mean, just in the case of simulation and synthesis workflows, folks that are industry insiders are really better positioned to appreciate the complexities that are involved with testing and experimentation and developing new models and systems. Similarly, industries where exploratory analysis and workflows are common, the founders with firsthand experience are, I would say, better equipped to uncovering patterns, trends, relationships, to kind of identify nuanced problems that vertical AI could address. In terms of how long it would take, I mean, I don't think I have a It's a good question. I don't think I have a defined period of time where I would say it should take this amount of time. I don't know, Mercy, if you have any— No, I think it's, I think it's really hard to say. What I can say is it's probably faster than it has ever been in the past. And I think these are examples where even someone that comes from industry, I mean, depending on their experience level, if they've spent 3 years in the industry, they probably need to still get deeply embedded with some design partners or some early advisors who have decades of experience in those workflows to be able to accelerate that product development curve. But it could also, I mean, if a founder themselves has been spending decades doing this themselves, they might not need as many co-design partners or, you know, advisors around the table in the early MVP period. So I think it's a function of what collective experience they're drawing from too. I think you really, really do need the industry expertise for these types of AI workflows. And so the design partners and having a strong advisory team can be really helpful in accelerating that curve in the early days. Awesome. I'll let my perspective here, and I, this question came up real time, but I was having a discussion with somebody yesterday, and basically the thesis was, look, OpenAI has been out for 2 years. That kind of kickstarted a wave where the incumbents start launching incumbent software companies start launching AI products 6 months later, kind of in a sprint that they weren't particularly ready for, and they're scrambling, and they still are. And that with what these products are really trying to do, it's so fundamentally different than core traditional software. And a lot of these understanding the workflows, the hard part of developing this, actually is going to take a long time, and consumer expectations have gotten ahead of the product development capabilities. And so we have this gap. I was just trying to think of how that might play at the earlier stages where you might be a bit leaner, you know, there's less alignment and less kind of incentives and conflicts around cannibalizing existing products. But the real challenge is that I, at least I see, is really, really knowing the details and how the work happens. And that it just— there's cycles, right? Like, users only give you so much feedback. You only get so many users in a given vertical or a given size. And it's in conflict. Like, smaller customers say this, larger customers say this. Well, what do you do? Because you have to serve both. And how do you do the product management? And so it's fascinating to think about like how quickly these companies can scale. And it's not always like the development. It's— Totally. Yeah. And the reason it's such a tough question was like, you know, it really depends on your industry and your vertical, right? Highly regulated industries traditionally don't reward speed. And so that is a significant factor. And I think that does make it a little bit tricky. Yeah. And I think it's kind of emblematic of You know, we are seeing companies— there's a conflict here too that I get in debates with people about too, because we're seeing the whole argument is like, oh, with AI you can get a product to market so much faster and with fewer people and more efficiently from a cost perspective, right? But yet, and these aren't even foundation model companies that we're talking about, but they're still raising huge, huge rounds, like successive amount of rounds, but they often are in these regulated industries like healthcare and legal. And the product might not be there yet that like fully lives up to the potential or the vision. But with a large war chest and the right people around the table, especially industry stakeholders like these large co-design partners, whether it's health systems or law firms or whatever, the combination of those things is really powerful. Like you can buy your time and build things the right way with the right nuances. You earn the right to kind of stick around and see how those workflows evolve. So yeah, I think it's a good question and I think we're still kind of waiting to see how it plays out. 'Cause you're right, it's only been 2 years since everyone kind of kickstarted their own AI development efforts. And I'll give you a really tough question then. Is the NPS bar higher or lower for an AI company than a traditional software company? Yeah, that's a good question. I would say it's— my gut would tell me it should be higher. Yeah, that's what I would say too. But then at the same time, if an AI product is doing a lot of the work for you, like maybe your expectations of what you have to do in the product is less. And so therefore, like, I think the NPS itself should be higher, but it's a question of like what meets like a great experience. Like what's the bar, I think is the question. And I don't know, I think that's a trick question. It definitely is. And like when you were making the point about the people who are aligned with the design partners and kind of have the trajectory where the, the product might not be there, but everyone at the table is aligned and just gonna take the bet it will get better, right? We know today it won't be optimized, but let's all work together, whether investing or customer partnership and bet on it improving that you could actually kind of underwrite underwhelming products. If you look at it in terms of you graded it as where it is today. And then the, the other part of what you said, and it's an interesting one, because so much of software in general and like value-based selling comes in like, how do you divide up the pie? Right? And it will vary product by product, right? Because some of that involves like, how much does the customer bring to the table? Like Asana or any type of project management is like great software, right? And adds a lot of value, but the customer creates a lot of that value, right? You don't buy Asana and magically have an organized business. You have it by checking off tasks and setting up the Kanban boards and doing all that work. And then to Marissa's point though, you'll have these scenarios where if something is fully, fully automated for you, like the work is actually done by the AI. Yeah, you kind of change the way you grade it, right? Because you're like, well, I was never actually making these account plans or market research reports. We always said we were gonna write the 15-page, you know, market study or whatever it might be, and we never did it. Now I had the AI do it, and it's kind of B-minus work, but there's something there, right? Yeah, some stats. We have a map. We have something. Right, and I'm only paying $40 a month or whatever it might be, like, how should I grade that? How should I think about that? And I think that's a good point about what value do you ascribe to it too? How much are you willing to pay? This is something that Taku's been thinking a lot about is like, what is the pricing model for a lot of these solutions going forward? Taku, I don't know if you want to add anything on it. Well, I mean, it's definitely like still an ongoing question because I mean, to your point, there's almost like a mind shift, right? Like I think traditionally, like the onus was on you to get the product and complete whatever task, but now you're almost switching where the service provider is completing the task for you. And so what does that actually mean when you're thinking of pricing this model? Does this throw traditional SaaS out the window? And so, I mean, it's an ongoing thing that I'm still thinking about, but look to see a blog post coming out in a couple of weeks. Awesome. Awesome. And then we'll zoom back out on all of this vertical AI, and we'll include this in the show notes, that you have a heat map where you've kind of ranked every, effectively every industry by their propensity to fit within a vertical AI software framework. Were there any surprising industries that once you went and did all of this work, you came back and, and were surprised at the result? Yeah, well, I mean, first and foremost, like, I want to caveat that this was a thought experiment and an exercise, right? We're not saying this is de facto, like, this is the truth. But it was interesting to go in with a completely objective, uh, bottoms-up view and then see what it spit out. And a lot of the ones that were on top were exactly what we would've expected. And we made sure that our process was completely objective, so we weren't skewing in their favor, and they still ended up on top. So like healthcare, financial services, and life sciences, we were— those probably would've been our top 3 going in, and they did end up on top. I personally was a little surprised to, especially given all the investment interest and a lot of the just kind of like macro tailwinds on legal that it didn't show up higher than it was. And I would say in general for a fund or for, you know, a large pension like Omers that invests very heavily in industrial assets and infrastructure and real estate, like I think we were pleasantly surprised to see some of these critical resources and infrastructure show up as high as they did, like oil and gas. I was not expecting power and utilities. I wasn't expecting them to show up as high, as highly as they did. But it just goes to show that I think this next wave of, of vertical software businesses, vertical AI businesses are going to target a lot of industries that historically did not fit the vertical software 1.0 playbook where you go after a really long tail of fragmented industries. Now it's much more like, hey, let's go after the enterprises. They're sitting on these large data pools. They are still very archaic in a lot of their workflows, and there's a lot of regulatory burden, and they feel the pressure to adopt AI or else they're going to lose, lose their competitive advantage. And in a market where, you know, the top 80% of the market share is concentrated among, you know, 4 or 5 players, like, investing or not investing in AI can like really make a difference. So it was interesting for me to see some of those kind of more legacy industries hard asset industries rank as high as they did, but we're hopeful and we're, that's, that's why I mentioned at the beginning, we're excited to look at opportunities in those spaces. Taco, I don't know if you want to add anything. I mean, I was, I was also quite pleasantly surprised to see those, those hard industries score pretty high. I think, I think selfishly for us, just with the, the under the Almost umbrella and the exposure we have with the different groups through our infrastructure group, our, our, um, our private equity group, our real estate group, I think there's a lot of assets that we hold there and just from the value that we can add as we spend time in those verticals. So I, I was kind of pleasantly surprised to see, to see some of those scored so highly. Awesome. Well, I think we've covered the report in, in great depth here. And as we wrap up, you guys get to work with like a pretty interesting mix of founders and executives. Marissa, what's a great habit you've learned from someone in your portfolio? Well, I'll tell you what a great habit is I wish I learned, which was we work with this amazing company called, called Mosaic. They're building a strategic finance, like, FP&A platform and incredible founding team. We love working with them. And the CEO, Bij, is just, I mean, I don't know how he does it, but he wakes up at like 4 or 5 AM and goes for a run like every day. And I wish that I could do that. And, and, and he still crushes it throughout the rest of the day. Never loses a wink of energy. So I wish I could do that. That's one habit I wish I had learned from one of my portfolio leaders. Taku, what about you? So I've got like, I've got an interesting seat. So I've learned quite a bit from just observing one of our portfolio company founders and CEOs. So Fred, who is the founder of Hopper, he's the founder of Deep Sky. And, you know, I've never actually had like any meaningful interaction with Fred, but what I have learned about, learned from observing him is just just perseverance and just sticking to something. And, you know, oftentimes if you not only work harder but work longer than folks, you know, you likely won't fail. And just seeing the journey of a business like Hopper and the ups and downs and some of the things that they've had to overcome, I think just from the outside looking in, I've had an opportunity to learn that and admire it from Fred just being resilient and sticking to something. And the reason I said like a unique seat is, you know, similar to Marissa, I have a platform where I get to interview entrepreneurs, you know, business leaders, you know, operators, investors, and I get to interact with a number of different people. And I'd say one of the big things that stuck, has stuck with me was, you know, most recently I interviewed just a CEO and a president of a major league baseball team, but He was speaking on leadership and building a culture where folks were bringing out the true potential in folks. And I think, I think the reason it resonated with me is, you know, when I was a founder, culture is one of those things I really just did not pay attention to. And I actually thought there was, was not a value add, at least focusing on it at the earliest, earliest stage. But I think spending time with him and realizing that the organization that you build is directly correlated to the culture that you foster in your business. And, you know, being very maniacal and obsessed with the folks that you're bringing into that culture, the type of people you're hiring, is extremely important. So those are the two things I'd say. And his platform that he mentioned is called HitVC. You want to put a plug in the show notes. He's doing a great job at it. Yeah. And maybe, maybe tell us a bit more. We can't keep these things too hidden. People need to know. To hear about it. Yeah, so I host this open series where, you know, as I mentioned, I invite entrepreneurs, investors, and business leaders to kind of connect with me one-on-one over a workout. And I think the best way to describe it— think of, I don't know if you've ever seen that show Hot Ones. Yeah, I'm aware of it. Yeah, yeah. So think of Hot Ones for workouts. The only difference is, you know, the goal isn't to make folks suffer, but it's really meaningful connections on a more comfortable playing field, right? And just kind of hear from these folks the tools, frameworks, the lessons that they've learned that have brought them to where at. And then I think the other cool piece is that I get to share relevant insights from subject matter experts. And so it's almost like, you know, I get to sit down with folks, understand what are some of the frameworks that they've used to get them to where they are. And then I share that things that I think are relevant to founders. So, you know, this last interview was surrounded on leadership and, you know, building team culture and performance, You know, I've had folks that have discussed venture debt. I have folks that, you know, focus on different aspects of AI. And so it's all these different things that at the end of the day, I think are great tools for a founder to have in their arsenal. But yeah, we work out and then we sit down and we have an interview and the platform is called HitVC. That's awesome. Well, we'll definitely include that in the show notes. And look, we've had you guys for coming up on an hour. I really appreciate the transparency and the insights. And I think we've got a great show here. Thanks for joining. Thank you so much for having us.