The New Data and AI Startup Architecture

Machine-generated transcript from the episode video, cleaned for punctuation and names. Speaker labels are not included.

[0:01] Welcome to the GTM Vault Podcast. I'm your host Rick Koleta and today we're joined by Ayush Khanna, the new ventures lead at super(set), a startup studio focused on building transformative data and AI companies. With experience founding and building four B2B SaaS startups across various sectors, Ayush has a deep understanding of the challenges and opportunities in scaling new ventures. In this episode we dive into his approach to founding startups, leading investments and shaping the future of data and AI. Ayush, thanks for being here. Hey Rick, nice to be here. We were just talking that you're coming off of a 36-hour fast. Why would you do something like that? And I don't think you'd bring it up. I wish I had a profound reason, but it just feels really good. I fast intermittently. I usually do a 16:8 split, which means fasting for 16 hours, eating in an ad window. My girlfriend said she's been trying to force to feat me breakfast. I get shocked to the system. Whenever I feel I'm getting to comfortable, a long fast helps me reset my body and mind. I sleep better, I

[1:10] have more focus. I have better focus, which is such a contradiction. I would recommend that you try it, for sure. Yeah, I've actually been experimenting with juicing myself, where I'll go without eating any food for two days, just juices, and I do feel a lot better. You look good too. They love the new haircut. So what inspired you to focus on data and AI when founding startups, and how did this lead you to super(set)? Yeah, well, few years back, I've always worked in data. I have a technical degree in computer engineering, and then when I was a consultant at ZS in Dey, my job was to help sales and marketing teams basically use data for decision making. I was business intelligence tools for them using MicroStrategy and Tableau, and it was really early in the teens, and use this data for compensation planning, for analytics and Le. I figured out how businesses, in this case sales and marketing teams, can use data as a powerful lever for making important decisions. Then I went for an MBA, and after my MBA I joined Salesforce on the

[2:21] platform team where I built a logging solution, our own logging solution, and I saw the vast amount of data that was being generated by the platform. And it wasn't just, was it, it was exposure to streaming data, streaming. I got exposure to cloud, unstructured data, search through Elasticsearch, and I really felt out of my depths. The whole paradigm had shifted from your transactional database using doing business to this new big data paradigm, and I was, I start out of my depth in initially, but I realized that this is the direction that I want to work towards and I want to learn. It's the same with AI. At Salesforce I work with a team building a recommendation system for our customers, and I realized how that self-reinforcing loop in AI makes your product better and the product stickier and reinforces the mode. So when super(set) reached out, it was actually very simple decision. There three reasons why I join them. So the first was there is a data ni focus to all of our startups. I'll give you an example, even BoomBox. BoomBox is a music collaboration platform, and when you think music you don't think D, and that's not the first thing that comes to mind, but believe me, it's the underlying

[3:27] business model and the value that SP is by its production and usage of data. And then, of course, to who are the founders. In general barers at super(set), they built Krux before that, so when I was researching their background in the product I realized Krux essentially a data product. They sold Krux to Salesforce in, I think, 2016. There was a billion dollar exit, and I just couldn't miss out the charts to learn from them and sit with them day to day. And then talking about startup studios, I think the speed of learning and execution at a studio is part anything else. And I'm extremely curious as a person, so learning is so important to me, and that's why I joined super(set). Can you walk us through your role as new ventures lead at super(set)? Talk us through some of your key responsibilities and challenges. Yeah, well, it's a very unique role. I love talking about it and people saying, whoa, that's so interesting and it's so unique. When I was interviewing and I was trying to make my decision, but super there was no reference there. I wouldn't talk to anybody about this role, so I had to understand the list and make a med on it.

[4:34] Intrinsically, in a nutshell, this role LS the best parts about being an investor, a founder and a builder. So I kind of get exposure to all three aspects and all three roles, and I get to talk to each of the people who do these things as that, as as them, so I understand more what thinking, more problems they face and how to talk to them. So what do I do as a new Eng's lead? So I walk you through the whole process of starting a new menure at super(set). So typically we explore multiple market theses, and these could be in different spaces such as devops and devops and tech at the same time. So roughly I'm exploring three to five market pieces, any in any time. I'm really trying to understand what the market is, what is the problem that the customers are facing, what is that persona, who are the players. After this I follow a funnel approach where I want to written down pieces into a few s ideas. So let's say I pick two ideas, as SN two ideas from all the thesis that have bu and all the markets analyzed. And in this ideation phase I'm trying to figure out if this is problem that is big enough, and for, do we want to solve this problem

[5:42] and how would we solve this problem, does this fit super(set), is this a data ni solution? And then after validating a few ideas and testing them with customers, with executives, I typically you put down them to into one startup that we end up launching. And I'm, as the new ages lead, I build the prototype, I build the MVP, I go market, and it's almost like doing Z to one as a founder. So I moved from being investor to a soundware role, and at the same time as I'm building the startup and building and going the market, I'm also building the team. So we hire head of product, head of engineering, key first engineers, designers, and then once the team is in place I transition the idea and the start with them, and then I repeat the whole process again. Now, challenges. It's a very unique role, as you can understand, so the challenges are pretty unique too. So the first thing I tell anybody who ask me about this job, I said there's no victory in this job. This role has no win. Starting a company is not a victory, right? That is the easiest part. So you really have to love the process, you have to love learning, you have to love exploring new ideas and connecting the dots, because if you find joy in that this world you have a lot of F this, right? You don't

[6:52] expect you chose the s, or you have a company exit, when the company exits you'll be known forward. The second challenge of this role is you have to go both deep and broad on multiple spaces, right? I'm synthesizing multiple ideas at the same time, a often that are conflicting with each other. I'm looking at data points, topic buyers and talk executives, and I have to make sense of all of this input, have to synthesize all of this input for multiple spaces at in a given point in time. So you have to have that ability to go and broad on in different spaces. And then, of course, don't fall in love with the idea. It's your baby until you're, it's not going to work out, and then you have to drop it and dump it in the trash. That's a bad example. You have to not fall in love with the idea because you have to able to slash and burn and cut your losses when you realize it's not a good idea, so as hard. I can tell you from experience. So that's the third challenge of the show. But doing this role has helped me build some very unique skills. The first one is, of course, I think I've gotten much better at building a pieces and validating it quickly. I Rel loed divers problem spaces, and the goal is, can I kill something in 3 minutes instead of taking 3 months to Val, save myself time and effort, and

[8:02] also have the option cost of not foret something else. And this is something that my friend M Fernandez, he's the managing director at Sierra Ventures, he talks a lot. He says the best thing you do is save time by killing bad ideas early. The second thing, the second skill I've developed through this mod is learning to speak to executives. CFOs, technology, revenue, marketing, ads, security. I had deep interactions from so with so many CXOs that it, I become better at talking the language and understanding the problems. So it's helped me learn about their persona, it helped me learn how to sell the executives, to build partnership with these executives, and most of all to excite them to join you, because you be successful if your team is very excited to work with you as a start. You're paying them all the people bus, and there got to be, you have to have passion. And the final skill that I've developed, I think, is discernment. As I said, after I start a company I bu the, I've they done, I think only 200 interviews for heads of product of engineering, and I feel I've gotten good at discernment and selection

[9:11] of people, and knowing very early on in the interview who are the ones that would make successful founders and co-founders. So I think these are all the challenges and the unique responsibility and skills through this SL. Can you talk to us about some of the experiences you've learned founding four B2B SaaS startups? So founding these four startups actually also made me a better investor today, and this is because, first, I've built a network of CXOs in security, engineering, marketing, revenue. I can speak the language, I'm much better at doing business development, I can establish trust. So it's important, I think, to know who are the customers of your customers, so you can be a better investor. Second is I've lived with founders and I lived as a founder, 24/7. At super(set) I spend my time with co-founders of other portfolio companies. I speak their language, I understand what they do day to day, and I honestly feel that makes me a better investor, because I understand how startups have built between the quarterly board meetings, not just that the three hour board meeting every quarter. Honestly, I've seen startups being built on Fridays at 10 p.m. and

[10:18] I'd like to give a quick shout out to my co-founder G at CA. It's an automated Q platform using G. One day before Independence Day holiday, I went back to the office after dinner, as I think 8:00 p.m. Everyone was home, except I saw this little screen in the corner throwing light at song. I went over and with G, and he wanted to wrap this one thing up for a customer before he left on a vacation, because he said he wouldn't be able to sleep otherwise. So that's the dedication, commitment that founders have, and something that most investors don't get to see. So appreciating it, I think, makes me a better investor today. The other one is idea selection. You have to be quick at selecting ideas because you can spend months or weeks on a bad idea. Even if you don't timately invest, it's a opportunity cost. So the way you kind of validate ideas as a founder is also good way to validate ideas and thesis as a investor, right? Are you solving a problem that is a Britten versus a painkiller? It should be a painkiller. You don't want to invest in vmin problems. And then, is the solution going to be data and AI solution, as the in super like to build? So the solution that you solve that to that problem, does

[11:28] it fit your area of competence? And also discernment, like I said earlier, I've interviewed 200 candidates so far of executive candidates, so that's MBE better at discerning very quickly who's the founder who will stick through the harder times. Yeah, so it's been a very intering experience over the last. How do you approach sourcing startups and preed and seed stages? Yeah, that's the billions and billions of dollar question, right? Sourcing is so important at in venture capital, especially in the early stages when you don't have a lot of financial signals, and there's so many more startups. So having seen both sides, being founded in VC, I prefer the inbound versus the outbound sourcing approach. And they're different approaches, they both have their own benefits, and I think you both have to do both as an investor. But if you building towards something, you have to build towards having opportunity and sourcing come to you, right? And the way I like to do that is what I call the three C's. I call them content, community and competence. So content, right, put your cases out there, right, put some comment out there about what you're thinking, publicize your thesis. And you do this for a few reasons. One, you want to test it, right? Sun is the

[12:36] best disinfected. If you're thinking crypto that Dogecoin will save the Federal Bank, you want to put your thesis out there and let people test in on it and listen to. I like the second is to educate. A lot of founders or potential founders are trying to learn about a particular space, and learning about the market, learning about the cers in that market or problems that buyers are facing, and if you can put content that educates them, then they automatically building relationships that I value. And the third one is, of course, attract, right? You want to attract founders and startups who are working on the problem that aligns with your thesis and aligns with your way of thinking about that thesis, right? So I think quality of sourcing is so important as compared to quantity of sourcing. So I would say publish high quality original content and make it so that the founders come you for advice. They understand that you know and understand the space so that they can come to you for actual advice, and not you hitting them up five colds a day. And then, of course, once they come to you for advice, one conversation leads to another conversation, as Tom says. And I have an example about this. I was

[13:44] building a thesis on retail media, and I published a lot of content around what I was thinking, and I got inbounded from this startup that was doing something different. They just wanted to know a small part of what I was doing, of what I was saying about he media, because wanted get into Thea, but they were actually a different product. And as I had that chat with them, I started deeper and talking to them about something else, about generative and advertisement, and I think they're a great company. I'd love to invest in them. I wish I could give them a shout out, I'd probably haven't asked them, so I wouldn't do that. But I'm still talking to them. So conversationally, conversation, good content, high quality visual content brings in incredible founders to you. The second C was community. So the community you can form, and I think you know this better than a lot of other people, that commun is can be so important. You can form it around a specific verle such as healthcare, or a function such as finance. You can form it around technology. Generative communities are huge right now. Are you a stage, right? Early stage founders or series A founders? Why do you want to form community? The first thing is you want to introduce founders to other founders, right? You want to build this community where people can come together, they can talk about the problems

[14:50] they're facing, they can bond. You can even introduce founders to their customers, which can be incredible. That's one of the hardest problems that founders have early on, is find their first customers, early customers to make a bet on them. So having that interaction face to face is incredibly value adding to them. And then you can also introduce them to experts. So if you have five data engineering experts or class data engineers in a network, just post an event community around them and let other people learn from them, and they also get exposure to other start things. So the way I like to do community and form community is through interesting intersectional topics. So not just one topic, but how does it intersect with something else? How does gener intersect with, right? And you can try different things. We, demos and spotlights and startups, and you never know how it pays off. It will pay off later. It's a slow process, but it's interesting things that happened when we've done communities. For example, this company that I'm friends with the founder and a friends with it, it's a company called nFactor. The founder came to one of our events at super(set) and we were just chatting, and we invited a few other people at events, and he met his other co-founder at that event. This happened

[15:57] three months ago, and met with them last Friday, and they just doing great because they such a good match. And that's not an outcome I had expected from our event, but I'm sure, as I work with them, they're going to be much more, they're going to be a much better company and a higher probability of success because of that match. And the final thing and the final C is competence. So I think competence comes from the foundation that you, no founder will want to talk to you if they don't think you're competent in what you're talking about, or the community that you're running, right? Founder time is the most important, the most valuable research. And so if you can play the expert role or the operator role and you can give advice based on your conference, based on experience, that is another way to approach sourcing startups from an inbound. I like that approach. Yeah, makes a lot of sense. Amongst all the startups that you meet, how do you evaluate which ones have high potential at an early stage? Isn't that the biggest problem right now with the hype cycle, right? The AI hype cycle means you got to have, it's causing even bigger discernment trou. So the first way I, discernment has to start with starting with us as investors, starting with credit mind, is a concept that, I

[17:07] think Accel came up with, and I really believe in it because you got to understand the problem, the market, the solution space before you even start evaluating opportunity. An example is we explored fintech for a few months at super(set) and we tested multiple ideas in fintech, instant payments, embedded finance, decision modeling for risk, and we didn't actually invest instantly in fintech. We shelved it. But then this startup approached us and they were doing embedded banking in nonprofits, and because we had done all the work in fintech, we already had ideas on their risks and challenges and strengths, so we were able to make a very fast decision and we made the investment decision. So starting with prepared mind first of all. And then how do you evaluate startups? So I evaluate startups first on the founder personality and then on the business. So the founder personality is that I love is what I call a lovable shark. So a lovable shark is someone, is a founder who their customers love them, their team loves them, they absolutely would die for them, right? And they're lovely people to be around, but under the surface they can make hard decisions very quickly and they can cut their losses if they need them. So they got to have a little bit of

[18:15] shock in them, and it's easy to test. You can ask them about the difficult decision they name, they can ask about their journey. You have to know that it's not been all rainbows and butterflies for them. So a love with shark personality is my archetype of founder that I love to invest in. The second thing I want the found is a key insight, right? They should understand the problem they're solving inside out. They should have one key insight that gives them confidence, right, that they are on the, they doing key solving this problem. They have to have a grand vision and they have to be able to explain that vision very simply. The third thing is what I heard this word, I heard this term a long time back but I love it, called stay in their edness. So the founders that I love have to have a stay in their edness, because they have to through difficult times, they have to almost thrive under pressure, because it will be hard. However awesome your product is, however unique your insight is, however much your customers and team love you, you will face difficult times. So you have to have a stay in there inness. And then evaluating the business, it's still early in the seed, in the pre-seed stage, so you don't have a lot of signals. So I like using the decision tree that Ulu

[19:22] Ventures uses. It has four parameters, market, product, team and financial, and we can go into detail another time. It's very thorough, there's a lot of parameters to assess these criteria for a startup, and it helps you essentially create a decision PR by assigning weightage. So it makes it simple, it quantifies something that's actually qualitative, but it's the easiest way to test the business and make the decision on this selection. You've conducted over 12 sector scans to create more architectures and opportunity assessments. Can you share your process on building these and how it informs your investment decisions? Yeah, so the way I buil thesis and make the investment decisions, actually it comes from product training. So I was a prod managed Salesforce, right, and I've understood and learned how to work backwards from the customer problem, right, and I think that should be a mandatory skill for investors, right, understanding the top key pin points for the customers of the startups that you're investing in, other market that you're assessing. So the first thing you ask is, what is the problem, what is the intensity of this problem, right? We talked about painkillers versus vitamin. You have to evaluate why now,

[20:33] right, what changed? And typically I think what changes are one of three things. Either there's a te shift, for example non generative AI, right, or there's an industry shift, for example there's a dominant player or there's fragmentation, or there is a customer shift, right, customers want something different. I give an example of a company started called super CMO. So it was an adtech company that used generative AI for creating different versions of ad creators and testing them out on social channels, and it was a close loop optimization, basically, of creatives, and that wouldn't have been possible without a technology shift. The why now was generative, right, and that opened up a whole problem space to us that wasn't possible before. So you have to understand the problem and the intensity of the problem and why now. The second thing you want to ask yourself is, what is the size of this problem and is it growing or is it slowing, right? You don't want to invest in a growing space, or whether this problem size is already so massive that it's attracted a lot of competitors. I like what Peter says, like you should have a market of 0 billion dollars and you start, because then you know

[21:42] you're the first one to solve it, and then it should be growing fast. The third way I kind of build my thesis and make investment decision is I interview customers who operators, typically CXOs, CDOs, and based on the comp starting, I interview these customers because I want to have a firsthand understanding of their priorities and how, what is important to that, and will the solution that I'm thinking actually address their needs. And then, of course, the fourth is, at super(set), because we invest in data and AI companies, I want to know the problem that we're trying to solve, is it solved by a data solution? For example, we would never build a Notion or a Slack because they're not data products. They are workflow, or not even work, they're more UX products. So I need to know whether the solution is the I on data n as a key differentiator. And then, of course, I also put on my consulting hat. I check, I test for market players. Are they dominant? Are the network effects? What are the barriers to entry into this market? How is the market size change? What are other startups doing? Not just the dominant players but other starters that are starting in this space. How they

[22:49] different, what are their approaches? And then finally the fit, I think the secret sauce. So I need to know what would be the secret source of the founder and the company fit that solves this problem, and not every founder who starts a company will be the right person to solve that problem. So product training, working from the customer backwards, putting all my consultant had, and then making sure there's a fit between the founder and the company is how I evaluate the thesis and understand and make my investment decisions. Can you talk to us a little bit about your strategy behind launching the super(set) community efforts? Yeah, well, that's been the most exciting thing I've been this year, I think, has been launching community efforts. And honestly, I, we didn't, because community is part of the P that it's good beh for sourcing. There at super(set) source talent, we source startups, and also joy, it gives us joy, because at our core we're all builders at super(set), right? We build startups, build companies, we operators. So we like being around other builders and sharing experiences and giving them advice and talking to them and having, listening to them and getting their feedback. So communities played a

[23:55] very important role and been an important pillar for success. I used to love going to events after work, and I think one of those events is where I met you, right, last TechCrunch just St. I still remember that meeting, and then, if you remember, we, I invited you to our office and we jammed on a few ideas. So I love going to events after work, meeting startups, forming relationship with founders, learning what's H in the market, what are the smartest people in the world working on. And then, as I was going to all of these events, I started thinking, why not bring them into our office, right? Why not do our own events? And so what I did was I first started small. At that point I was exploring F, this is last year, late last year, I was exploring the f space, and I had gone to a couple of events, and shout out to Sheila, Sheila and the F next community. So what I did was I hosted our first event, which is in collaboration with Sheila. We had a panel, we had sushi. So we started small, we leveraged an existing network to found that community of builders and your founders. And then what we did was after doing a few, it show these small EVS, we started doing more targeted outbound. So as we figured

[25:05] out who we, what are the kind of people we want at our events, we used LinkedIn and a tool called Rift to execute a targeted outbound campaign to invite the right people to these events, and we were extremely selective and who we wanted at these events. And once we have enough, once we had enough events under our bed, we encouraged them to subscribe to our event so they could be part, they could know and be aware of the next upcoming events and subscribe to our calendars on. And Luma is a great tool that we found to host events and engage with subscribers. And then, once we had these subscribers, we also developed some engaging content. We send our newsletter and blogs to all of our subscribers. I think we have more than a thousand now to our super(set) community. And this has helped us build a very curated community of people who want to be with super(set), engage with super(set), eventually be part of super(set), that we want to work with. So this has been our strategy in launching our community efforts. Now we have, the result is now we went from a monthly small event F, now we have three events every month, and I think we've been featured in the San Francisco

[26:12] event featured events five times. And we had the same founders and same pent coming to our events multiple times, so it's become a case for them to come and build relationships. Like I mentioned earlier, we've done co-founder matching inadvertently, but yeah, it's very, it's amazing how that played out. And you also need a lot of executive buyers to ask startups to, startups you want to work with. Our portfolio companies also benefits from exposures that they get at these events, so communities have been great for all of these reasons. Shout out to tool run by a friend, it's called Rift, get Rift, and it's great for target outbound, warming up your domain. They're not paying me for this and I'm no investment in them. Yeah, if anybody's thinking of starting an event and just starting a community, I just say start small, right? Just find your rhythm, find what works for you, find a topic that you're interested in and network that you get to talk about that topic, or BR the community around in that topic. And I also like to shout out to my teammates Casey, Maxwell and Cat. They actually do all of the work, not even most, do all of the work in our community efforts, and we all get the benefits of

[27:19] that. That's great. Yeah, and I want to just say thanks again for the phenomenal collaboration we had with you and the super(set) team when we did GTM Nights 3 at your office, and it was just a phenomenal evening. Thank you. Yeah, we had a great time too. That was a busy week, but I think having you guys, and I think you have a great community of GTM pros having come in, give us feedback on the idea, demo with them, and also just interacting with them. I still in touch with so many of those folks I met at your event. Glad to hear, glad to hear. In your previous roles at Salesforce and in management consulting, what key lessons did you learn that have influenced your approach to product and business development? Yeah, I'd say Salesforce and the consulting experience, they led me to where I am today. I think the newent lead role, a l of the qualifications are product and consulting. So as a consultant first, which I was as consultant first, I learned three things. I learned the customer focus. We call them clients and consulting, but just the extreme focus on solving their problem, understanding that for, talking to the customer and clients, and living their life, if was incredible. It was an incredible learning experience in consing. The second thing I

[28:31] learned in consing is how to roll up best leads. You don't have 50 people working for you, you don't have any engineers, you have to do everything from collecting data to getting presentations to delivering presentations to doing customer interviews. You have to roll up your as a consultant. And the third thing is how to use data effectively and how to draw insights and analytics out of that data. As I mentioned, you m, I think was family data and business intelligence, so I learned that, I learned this very critical skills that con. And then went on to Salesforce after my MBA. I started a couple of zero to one projects, almost as an, what we call an entrepreneur. I started at healthcare vertical in Commerce Cloud, and I also launched a feature by encryption Dr, which a lot of my customers are demanding, were asking for. And so this Z experience I got twice made me a better new ventes lead at super(set), and it helped me understand how revenue is developed, how do you do business development in a start in B2B size, what does the revenue machine look like. And I can tell anybody listening to this, Salesforce has probably the best well oil machine in revenue across any B2B SaaS company that I've seen. I, as a product manager, had to focus on a very narrow piece, and then the PMM took

[29:44] over, the general manager took over, sales enablement, everything was just so well oiled. I have to worry about any of that, so it made my life easier and long made launching new products easier. But I got to understand this machine very intimately by working with them when I did a few Z ones. And I also work with NX engineers. You meet, there's so many engineers in the area, but when you get the NX engineers you can see that it's very evident, and sometimes very hard to work with, but that's great because you get challenged and you feel you know nothing. So all of his shortcomings are very evident, but working with the 10x engineers, hardx engineers at Salesforce was a huge unlock, and it showed me how you can build products very effectively. So yeah, I think all of this together led me to super(set). And someone who has built several companies across different industries, how do you see the role of AI evolving in B2B SaaS, and particularly in the sectors you're focusing on? Yeah, well, I think now when anyone says AI, it generally means generative AI, so I focus on that. And we've been building thesis in gen AI for more than a year now

[30:50] at super(set). I, December 2022, when I came my first presentation. And it's since then, the primary use case I see today for AI in B2B SaaS is what I call the low risk high volume use case. So I like to qualify AI use cases the risk and volume, right? It's like a 2 by 2, typical BCG 2x2. And the highest that AI is making today is on the low risk high volume use cases. And then use cases, it's just customer success or creative tasks or analytics with human. So I see the impact of AI and need to be s in two ways. One, it's on the actual application, and the second impact is on how they B. And on the actual application, the impact of AI is only additive. It hasn't changed the workflow, you. And as founders, I think this is important, don't short change the workflow, don't short change the customer needs. AI is not going to solve all the problem. AI should be at top, not be the product itself, right? So in applications, what I've realized with AI now, data is becoming even more important. So you have to

[32:01] figure out how do you capture data, how do you use data effectively, how do you use data for reinforcement, and this is all in making your AI better, right? New products will be built around collecting data. And I have a hard take, I think ChatGPT for consumers was actually meant for free data collection. I think that thumbs up thumbs down, every they have had the to the responses, they were able to collect free feedback from 100 million users that help them improve the models. So it's just surprising, people now pay $20 a month. But I think Sam Altman is smart, he said let's launch this product and use free labor to get, make our CR better, instead of paying $2 an hour that they pay to the DAT supply chain, the. So the first thing that changes is not data has become more critical. The second thing that changes is a human workflow will become the new way of designing a workflow. So AI will be embedded in the SaaS application workflows, and there will be a human in the loop to perform certain tasks, right?

[33:13] Humans won't be eliminated completely. Yeah, will do 100% end to end. A human in the loop is my thesis around AI in the new B2B s using AI. For example, let's take two examples. Let's say applications that are analytical, right, like logging, logging solution like Datadog. So AI will perform 75% of the task but you'll still need human oversight and human editing, right, because the AI output is still probabilistic and human will need to check the output of AI. So in log analytics, what I've seen is you'll have AI synthesizing multiple LW line data coins and compiling all of these we signals from strange signals from logs into a human readable text format, saying it looks like your performance is decreasing or your CPUs are overloaded, right, and then they will surface this insight in natural language to humans, and humans will have to make, will have to dive in if they want to, if there's something critical they have to dive and understand the go to the root cause, go to the actual data points, but they don't have to continuously oversee all of these Nores and try to analyze that data.

[34:24] But if you think about creative tasks, for example producing content, I think what will happen is there will be human direction at first, right? A lot of what we call prompt engineering could be PR engineering, reinforcement, or giving data, giving the right data to AI, or creating the first draft of a certain creative. So I think for creative applications humans will give their direction but AI will be used for scaling, right, taking that same content and personalizing it for different audiences, translating into different languages, creating 10 version of the same content. That is what AI will be used for. So I think analytical workflows and creative workflows will have different ways to use AI, but that this is, it'll definitely be a human in the loop kind of a workflow. Now, we talking about AI, I know, and AI is different from agents, but I also want to give a quick shout out to what I think agents, the role of agents. Agents basically replace human task. They do an action through reasoning and use LLM in the back end to look at it and do the reasoning. I think agents will reduce the cognitive and the physical workload on humans, so that you

[35:32] know humans can not have to press all the buttons or type all the keys. Instead humans can play that human the loop, the oversight, editing kind of a role, right? So it's not just AI, but I think agents itself will play a huge role in the new we s applications. So what does this mean? This means reinforcement learning will become more important, right? So I'm working with this company called nFactor. What they do is AI for security. So you want to respond to RFPs. If you're in the sales process, you understand that you know that huge problem of having a lot of RFPs you need your security needs to respond to. They need to fill out all these questionnaires, and it's a huge block be your sales center, and nFactor is resolving that using a reasoning LLM that looks at all of your data, all of your security and monitoring datas, and using agents to fill out these responses, right? So at the end day you can see that these a and agents, it's all a real problem, if they're not just. But I think a human in the workflow will be CRI. And then the second point was how startups are built. So not ding too deep into it, because it's a lot about the tooling, but I think SaaS application, B2B SaaS applications, they be easier to build using AI, right? Everything is

[36:42] getting extracted. Your application development is getting extracted with coding platforms that write code by itself. I don't know if you've seen the new Claude demo, but you can make simple apps such as Splitwise, or you can make financial apps, or you can make marketing apps just by writing simple text, and it spits out the code and even spits out the app. So I think making B2B s application, making any application where B2B SaaS as well, becomes so much easier, that even AI is going to be abstracted through AI. So there will be a lot of point solutions for every problem, right? So it'll be tougher for the buyer to understand which is the right solution for them, but this is an opportunity for tooling companies, right? Tooling companies, how do you make developing application, how do you make develop B2B SaaS applications easier, how do you make moving data around easier, how do you make integrating AI into B2B SaaS applications easier? So I think this whole B2B s, the transformation with AI, also affects how they bu, and there's a huge opportunity for, you can call them devops and mlops tools, to come in and help B2B size platforms go to market

[37:50] and build applications fast it. And do you have any advice to founders who are looking to build and scale in the B2B fast space in today's competitive landscape? Yeah, so kind of taking forward what we discussed, since it becomes easier to build B2B SaaS applications, it becomes harder to differentiate and harder to scale. So there are a few successful strategies that I've seen farmers a do. The first one is building in public. You want advocates. The more advocates you get, the bigger the better the flywheel, and it's easier to scale. I've seen a lot of startups now B in public. An example of that is my friend P. She runs a company called Butut. She founded a startup called Butut which helps you build good looking websites within a few seconds, you gener, right? Another startup is r2b. People have very different opinions of that company and the founder, but one thing you can't ignore is R to be l. So using content community is whatever, like you want to use, but building in public, being out there even before your product is ready or perfect, that's a huge huge way to differentiate. That's create success strategy. And I

[38:58] think last I heard from froma about B, they had thousands, and I think hundreds of thousands of websites they had launched using Bon, and this has been less than one year. The second advice is do more with less, right, because it is getting easier to build. You don't need a 20%, 30% engineering team or sales team or marketing team, right? The best products will be Pi by a small team that is a. Right, you want to constantly find the problem, you want to find different solutions to that problem, you want to test the solutions and you want to be able to pivot, right, until you land on that fit where your market is pulling the prod out of you. And you need people to scale. Don't hire too many people. It's very easy to build an application these days, so you can do a lot more with this. And I watched this talk by Mark Zuckerberg when he said he didn't many started Facebook, he did not know how to build a million, million people community that he learned it as he built Facebook. He started off very small, three people in a d building something just for Harvard. So that's the approach that I think founders need to, even for B2B SaaS. Related to that is just build, don't wait for perfection. No, perfection will

[40:07] come from irration. And there's a study given once that was done once where two sets of students were asked to paint a drawing, paint a painting, and one was asked to paint 10 times the same painting and improve, the other one was asked, was given 10 days to make the perfect painting, and guess what, which one got the best painting at the end, right? So just build, right. Don't wait for, just go to market and then you can iterate once you're in the market. And I'd say something that sounds different. I say ignore AI at first, even in a B2B s startup. Do everything by hand, because as soon as you start integrating it and making that the cornerstone of your product, you focus so much on AI, you forget the workflow and the problem. What you want to be doing is testing the problem and building the workflow very cheaply. AI is going to be the easiest bit that you do, honestly, right, unless if you're building a foundational model, which B2B s companies, most of them aren't. You want to solve a problem for the customer. Your AI can be your co-founders sitting in the garage at night puning the recommendations manually. Follow that at

[41:12] first, build the AI once you know you're solving a problem. So that's, I think, what I've seen be successful in building and scaling B2B startups these days. I know you talked a little bit about your excitement around AI agents. Looking ahead, what other emerging trends and data and AI are you most excited about? Yeah, so here we talked about the 2 by 2, right, the list versus the volume of task. So I think now AI is getting better, the models are getting better, foundation models are getting better, people have figured out different ways to train in and fine-tune these models, so the task that they're allocating the AI is moving from lower risk to higher risk, right. And the volume that, they were the use case that were focus on high volume, noway use cases are changing the even lower volume costs, right, because AI is cheaper now. The processing cost going is so much lower than when, two years ago, that yeah, had can found lower volume task cheaper. So now why not just have a everything? For example the RFP process, right? It's low volume, responding to RFPs, not as high volume is analyzing loss, right? Find is ends cheaper, and why not

[42:23] just have it ends faster? So I think that's the overall overarching trend that I'm seeing. But three very specific trends that I'm very excited by, or three spaces that I'm very excited by in data and AI, are data engineering, AI agents for security, and AI for creative task. In data engineering, I'm seeing a developer iation or developification of data engineering. So data engineering is basically how do you bring data from different places into different place, how do you move data, how do you manipulate data, transform data, and how do you use the data, right? So I'm seeing a deification of data engineering, and this is something that has happened to a few other problems before data engineering. It happened with developer operations, devops, right? Companies like Pulumi, companies like capard IO have made it easier for developers to do devop and not worry about it. The same thing happened with security. Panther allows you to write security rules in B and it abstracts all the complexity behind that, so developers can now manage security of their applications in the language they understand, right? They don't have to hire a security team. I

[43:34] think the same thing is happening with data pipelines now. Manage your data without being a data engineer or a data devops engineer. So companies like Quil, data out hand, I'm talking about both of these companies, they're doing fantastic work, and as I understand what they do, I think this is a trend that's going to accelerate, because data is becoming even more important in the new G world. The second trend that I'm very excited about, the second PR space, is using AI and agents for security, similar to what nFactor does. It's using AI and agents for sec. So it's different from security for AI. I think when AI started, everyone was worried about how do I secure my AI, how do I not pass data that is sensitive to AI. I think that's still a problem but not as much as it was made out to initially, because people have gone more comfortable, solutions have come up. And if you think about the security, right, if you just think about the security team, the security op team, they're so overburn with work, right? No sec op team says we have a lot of time on a. So they have to deal with a high signal than M ratio, a low sign ratio. I have to figure that out, that the speed of that action is very ENT. So what

[44:42] AI does is that it synthesizes security data, and agents helps close vulnerabilities much faster than human would have. Of course you need a human in the loop, ear point, a human in the loop in the work for, to make sure AI as there human oversight. But I think using AI and agents in security is going to be a very exciting transformation. I know all the existing vendors from Ling the secur and looking at AI, but I know there's a lot of few startups that are also making it easier, for example. And then for creative tasks, right? Yeah, for creative tasks, creative tasks have very few rules, right? There are more like guard rails, such as, if you're talking about advertisement, then what is the brand guideline, what is the creative guidelines. Other than that there's no real ru. You need something that is imaginative, right, between the lines. So using AI to create content, to transform content from one format to another, for example images, text, to scale this content, how can you take one source piece, how can you take a podcast, make it into a

[45:49] video, make it into clips, make it into blogs, right? I think that's going be incredible instead of hiring 20 people to do that, as you do today. You can have AI do that, and also test, right, because testing is a huge part of, especially if you're not really sure of your messaging, testing messaging and testing content becomes easier with AI. It's a huge part of finding the right audience and message and channel for you. So I think these three, data engineering, AI e security and AI for creative tasks, are the trends I'm most excited about right now. And a final note, I want to ask you, if you could wave a magic wand and change the way venture capital functions, what would be on your wish list? Oh man. So I've been on both sides, and as a founder I have fed with some amazing venture capitalists who very smart, who delighted to work with through the entire process. But I've also kind of figured out some ways to improve the experience as for a founder. So the first thing I'd like to tell my friends is stop being SDRs, stop sending generic cold email, stop just grabbing coffee at Blue Bottle with founders. Founders are humans. When I'm

[46:57] founding a company, and when VCs reach out and they just want to talk for 15, 20 minutes, I'm making a decision on my time allocation, right? So make a connection, right, cultivate a relationship. You don't have to talk to them about investment when you want to talk about investment. Just see how you can add value, right, how you can help, how can personalize the whole experience, how can you make the experience of a direct of working with a VC as a founder more enjoyable. The second is, the second way I think we can all improve is by having a prepared mind. Be like a heat seeking missile, not a cluster WM. Don't just spray and pray and see who, what you going attract and where you can make, where you can make and make a de. You should in fact have a thesis, have a point of view, and then you're finding the right TI, the right sound, right comfortly, based on your thesis. And that makes it better for both. You is that it reduces your cost, your oper cost, the time you sp, but all makes it better for the founder, because the founders know that you're not somebody who's new to the space. It

[48:06] also becomes easier to pitch when you're finally ready with the startup. You pitch the startup, your investment team, it makes it easier because you have prepared mind, you already have the facts, and then you're just playing in the company. So there the I have, which is around the invested company F, then we have the private market F, we have the message market F. I think invested company f is a huge thing as well in venture capital. So you have to know, are you the right person for this company? And then finally, I think making sourcing and selection fun. So show up with market research, show up with customer interviews, show up with competitor intel when you drop with founders, right? When I drop with founders I already tried their product and I've actually used it, so I give feedback, and I was a product manager before this, so I can give them feedback from an operator point of view. I also like found a point of view and market, do all of these things, spend time where in the space and in the part, because for a founder it makes that whole process of sourcing and selection a lot more enjoyable, and gives them value whether you invest them in them or not. It isn't value, and they will want to work with you and they will refer their fans to you, right? And network is such a huge part of Silicon Valley. Whether this network

[49:13] you already have, a network you build by working with founders. So whenever I work with founders I tell them, I'm your friend and your partner in the whole process. Count on me, ask me for things you need. It's not just a one way process. And I would love to see more and more me SE doing that are. Everyone, thanks so much, that was very insightful. Thanks Rick, always great talking to you.