Earnest Sweat 00:06
Did the SaaS era do more damage to human creativity than in our natural capabilities of being creative and builders? Because it’s essentially got us all in. I’ll speak to our industry like set it and forget it. There’s these hard fast rules, and when you say, “Hey, you know now building these loops and you know automations and AI, no matter what you do, there’s this frustration that something can change and break. But where my mind goes is like, that’s like 99% of life.

Santosh Sankar 00:46
Yeah.

Earnest Sweat 00:46
And so, did SaaS really get us set up for failure?

Santosh Sankar 00:52
I think SaaS. If you’re unwilling to change, then yes. And I think part of what I like is just thinking about software. What is software very good at? It’s when there’s a known workflow or pattern that is repeatable and doesn’t change. A business will find success using software. And there’s a lot of industries, including the one I invest in, where the world’s built off exceptions and abnormal events happening, and they’ve historically also been very poor adopters of SaaS,

Earnest Sweat 01:22
yeah, and

Santosh Sankar 01:22
so kind of bring that back to people. I think venture, yes, perhaps, especially if there’s an unwillingness to change because venture or even like stepping back, investing is something where the incentives are not built around you changing something that your backers or your LPs believes has worked in a prior fund. Yeah, you almost keep your strategy the same. The types of companies that you invest in are the same. You might maybe on the margin innovate around how you do your work, and that kind of limits creativity and the bounds of creativity. Think about venture firms. Can you think of one where they say creativity is one of our values? We have

Earnest Sweat 02:11
it. If it is on it, I you know it’s probably people that I don’t believe because their website looks exactly like everyone else, like

Santosh Sankar 02:17
someone else’s. Yeah. So yeah, there is a bit of that, and and and I think it’s its incentives, SaaS, being an investor in SaaS, and and also probably just the personalities that VC draws may not historically have been creatively maximized. Does that make sense? Like software engineering, probably not historically an occupation you would say creative people go to.

Earnest Sweat 02:46
That’s true.

Santosh Sankar 02:47
But now with AI, like AI allows you to be more creative and imaginative. And in order to actually find something like a breakthrough, sometimes you need to help. You need to co-opt an agent to unlock your creativity. Yeah, but you can’t do that if you’re not default having a creative bone in your body.

Earnest Sweat 03:05
Yeah, I was being a little

Santosh Sankar 03:07
facetious about it.

Earnest Sweat 03:08
Facetious, but you know, I wanted to push on it on how SaaS impacted us. But I think if you look at the SaaS era, the goal was efficiency, and now AI, I think, comes in with this idea of productivity, but also I think pushing boundaries, right? And it’s causing a lot of tension. The reason this is so important right now is because I was speaking with an LP for my other podcast, and they were saying that every pitch that they were getting, or even some kind of quarterly update from a GP, everybody mentioned their AI strategy. Everybody, and this LP spoke about their desire to kind of really uncover why people what was the so what of having an AI strategy, and if it actually aligned with what they felt that GP’s advantage was, or how they fit within their book, and so it had me thinking about how novel tools started out our novel approaches in our industry have started out as really something as a key advantage, but then quickly becomes table stakes, and then so like how does that impact the once advantage or someone who’s coming in maybe a little bit later? How do they continue to enhance their advantage, and we’ve seen it from, you know, some firms focusing on data, to some firms focusing on providing platform services, to some firms, you know, just specializing in specific industries. What do you like? How do you see this kind of thing? Like movement with AI,

Santosh Sankar 05:02
So it’s interesting you say that because I found data somewhere that 85% of private market investors now report that they’re using AI. So I mean it’s table stakes. Yeah, you cannot not do it, which I think is a net positive thing for productivity. Yeah, but kind of reflecting on what you just said is interesting, right? Because you said data platform as the next kind of driver of advantaged or perceived differentiation, vertical specialization such as Dynamo or you know John Gleason at Success Venture partners that could be defined a few ways, but I think what’s interesting is there’s always been in those prior cycles roughly a one to two fund cycle commoditization of that novel tee, right? Whether it’s data, and then probably within five six years, most people kind of like, oh, we have access to data. We can help you with that, and then similarly to the platform. Although platform does require, and it might have been more a Series A and later thing that you needed an AUM base in order to kind of pay for all of that vertical specialization. Probably something that sat in the earlier rungs of venture because AUM plays a role in helping you have a defensible portfolio strategy against a vertical. It’s hard to do that as you have more AM. But I think what AI is doing, and I think there’s a bit of irony here, right? Because on one side of the equation, we’re investing in AI, and we see it commoditize and remove defensibility or the notion of defensibility in the industries it is being adopted in, and against the competitive base or the legacy competitive base of software services. So I think actually it’s going to accelerate potentially this one to two fund cycle commoditization of a novel concept, and let’s kind of go back. Right, we’re sitting here, 2026 three years ago, 2023 I mean that that has arguably happened already in one fund cycle.

Earnest Sweat 07:14
Yeah. So

Santosh Sankar 07:15
so it’s already here, and I guess the question I now think about, and I, I rumble on about from time to time inside a Dynamo is so what makes us unique? If we’re all using the same AI, we’re using one of like three vendors. What makes us unique? And maybe some of that is if you don’t have anything unique and advantaged without AI in the equation, AI isn’t going to actually bring that to you

Earnest Sweat 07:44
at all, at all. And I think this is something that LPs are discovering, right? When they’re seeing the same things or the same approaches, yeah, you’re using it for email. You’re using it for you know writing your VC’s book reports, which are essentially our investment memos, for you know sourcing for kind of having multiple touch points with different founders at all times, but I think something that needs to change is like that everyone can do that right now, and everybody has to do that. You really can’t rely on just inbound anymore for your deal flow. But my approach is, and that my opinion has been that there are a couple things that you need to have change management within your organization to really assess out what your advantages should be, and so I think one is you have to be extremely self-aware as an individual investor and your firm of like what are your strengths. I think tools need to be utilized, whether it’s AI or anything else, to enhance your strengths. So, if you’re really strong at something, the tools that you’re using, especially with AI, should be able to magnify them by 10, 100x The other piece that’s important is you have to have a mentality of tinkering, and that’s very different from when we were starting out in this industry. For example, here are the hard and fast rules of seed investing, Santosh. Here are the hard and fast rules of Series A enterprise software investing, Ernest. And so, how can you figure out one? What are the most important workflows for you? Two, have some type of process where you’re constantly in a balanced way without bullying the ocean, seeing what’s good out there as a tool and what’s not, but also even if it’s good and works for you, Santosh, doesn’t mean it works for me. That’s right. It means it goes back to like what. You are really great at, and you have to have people who are willing to tinker with that in a way that again pushes to the so what, right? Yeah, and so I think those those two things are important, and it made me think of like an analogy of, and I always like to bring in sports, but when all the nerds from Sloan Conference came into basketball and let us know, us idiots know that three points is better than two points, right? And so if you shoot 35% it’s better than 50% twos, 35% threes, and so everyone’s just shooting a lot of threes and layups and free throws, and so that quickly becomes table six. So what, what, what differentiates champions from people who are mediocre and barely make the playoffs? It’s the talent and the expertise, and so for Golden State Warriors, this really puts you in a good place if you have legendary three-point scorers like Klay Thompson and Steph Curry, and also have an interchangeable defensive player like Draymond Green who can deal with the pace in space, right? And so taking that analogy to venture is like, all right, here’s the tool, but what is your skill set that’s going to differentiate you and that’s going to leverage that tool in this current situation? Yeah.

Santosh Sankar 11:31
The one thing that I stop reflecting on, and we’re a small team. We have one principal, and the mountains have been awesome, but I think about a year ago I stopped and we were on the road and we’re just kind of thinking about industry. How does one develop in the industry? And I stopped and asked her. I said, like, what do you think your superpower is? And if you look at a firm, you want to idealistically bring people who have different superpowers together, and your team is formidable. That’s a lot easier said than done. I would actually argue that most funds probably have people who have duplicative superpowers, and they’ve done okay, and we could kind of visit that at some point. But I do agree. Like, what is one superpower? A superpower could be as simple as, hey, they’re just technically very astute, or they’re an amazing product person. To where I like design, there’s I don’t know if you remember the early 2010s. There were a few amazing people who had design backgrounds or product backgrounds, like GV, which was originally seeded with these types of people out of Google itself. What is it that is your superpower? And it could equally be rather than being functional, it might be someone who has just an amazing EQ who can have hard conversations and get tagged in where there might be a challenging situation or a challenging set of personalities, and if you constitute that in the right way, that could be viewed as differentiating and hard to replicate, durable as well in the eye of an LP. But I think going back to the earlier statement, it’s not going to be differentiating or durable through a cycle. If like you’re showing, I’m saying we do AI, we have a platform. Like, depending on where you sit in the venture pecking order, that’s going to be table stakes. Yeah, for you. But also, kind of going to your point on strengths. Like, don’t use AI to help you improve a weakness that might be better done by actually adding to your team.

Earnest Sweat 13:46
Absolutely, in

Santosh Sankar 13:47
a certain way, is the one thing I also think about when we think about AI. I think one of the more important artifacts inside of a firm that’ll emerge is the skills file that you’re creating and all the accompanying references that that skill has to go tap into, and that’s what like we spend time once a year revisiting, because you’re right. Like our book reports, our LP letters, our book reports. Let’s take that. We have an agent that’ll go through all of the notes, all the context that we know about a company, and we basically have documented like we’re training a new employee a skill file that says this is how we think about teams, and there are these five traits. It’s not that we need every team we invest in to be maxed out. It’s an awareness of where they spike, and based on what we believe about this company and the market they’re in, is this the trait they should be spiking on to be successful?

Earnest Sweat 14:50
Yeah, yeah.

Santosh Sankar 14:51
But if you don’t think like that, if you don’t believe in the kind of skills MD file actually represents, kind of in the AI world, you’re. It’s going to be very hard, I think, because everything else, like access, is probably the other thing LPs talk about. I think access is one of those things. It’s very hard to show in data, perhaps that you’re better than the next fund over. But it feels like the one few things, one of the few elements of venture that still continues to be the differentiator.

Earnest Sweat 15:25
Yeah,

Santosh Sankar 15:25
fair, maybe unfair, but like that’s the thing. I don’t think AI is going to crack its access. That’s a relationship thing, right?

Earnest Sweat 15:31
Yeah, yeah, yeah. Yeah. I don’t think access is really going to ever. We’ve seen we’re starting to see it be challenged with these AI only venture agents or or groups like a like a boardie or something like that that’s trying to connect a lot of different people who have never been connected, and you can talk to different people about that type of solution and and they’ve had mixed results and typically not great results at all, because I think what is missed, even if you provide somebody in one conversation of like, “Hey, this is what I’m looking for, it’s taste in someone’s specific type of taste and what they’re looking for. And so,

Santosh Sankar 16:17
Could you teach me about AI taste? You think I’ve experimented with this notion.

Earnest Sweat 16:23
I mean, that’s a really good question. I never like to say with technology no, never, because you probably could be surprised. And but today, no, and I think that’s that’s where you can’t outsource your taste right now in a market because I think if you if you balance bringing in these tools of being able to take on more context, more of your insights, give you more room to meet with more founders, more people. Theoretically, you should be able to like it’s it’s like the I was reading some book about a famous chess master, and when people were bringing in computers, there was a lot of controversy of these cinnatars, I believe centaurs, and that’s what you should try to be as an investor. Like, how do I bring this machine to help me so we can be like the strongest entity ever? So for the human, it needs to be the taste, it needs to be the feel, it should be the hard asking questions, because I think you shared this in our kind of pre read the McKinsey stat. If you could share that with the audience,

Santosh Sankar 17:43
yeah. So McKinsey did some analysis around deep research against expert interviews across 10 different industries. So there’s a good sample here. Gen AI showed a more optimistic happy talk bias. I think this is what you’re talking about, right? In seven of the 10, roughly 40% of the important data points uncovered in expert interviews were just absent, which is scary, right? Because think about how it’s just common practice to go say, “Oh, here’s this dense reporter. This has sent me. I’m gonna throw it at Claude, and Claude, like, give me a one-page summary,

Earnest Sweat 18:24
yeah.

Santosh Sankar 18:24
And you may not be getting everything that you really need unless you’ve taken the time to make sure Claude has the context you care about,

Earnest Sweat 18:34
yeah.

Santosh Sankar 18:34
And you revisit it to your earlier point on a frequent basis to make sure it’s like it’s reflective of how you’re growing, how you think about the world, and your brain, and the frameworks you’re adopting and retiring. Yeah, ultimately,

Earnest Sweat 18:47
It’s a weird time because we’re just in this transitional period, right? Where an example could be like reaching out to a contact and it feeling a little bit like somebody’s like, did you have AI write this? I don’t have that as much, given I’ve always written pretty formal emails. Yes, and so as people are just now, if they see an M dash, they’d be like, but we’re in this transitionary period where it’s like, all right, there’s going to be a little bit of Rocky of like how much we use AI, how much is us, and I I think of like when I was in college and people said when I started college, people said don’t use Wikipedia, and by the time I finished college they were like it’s source, don’t worry about it. It’s credible. It’s credible, and so you just kind of don’t want to be able to build and take advantage of the transitionary period, and not wait until it’s already credible because you could have missed out on opportunities of, of reaching out. Because like, what is the unit that’s most important for me as a venture capitalist? Is it that somebody says, “Hey, you fully wrote this email, or is it, “Hey, Ernest.” Let’s meet and have this call.

Santosh Sankar 20:02
Yeah, right. That’s

Earnest Sweat 20:02
more important to be able to consider the context. But I think again, bringing back this builder mindset and being able to tinker is like, all right, which workflows do we use it for? How much do we bring in? How much do I need to check? And where should I check? And so we, I’m seeing it even in kind of the AI note takers of like making sure, hey, if I have that’s why they have the scratch pad there on Fathom or Granola whatever you use, or if there’s another out there, yeah, let me and Santosh try for free. But essentially giving you the say, hey, this is really important. Make sure that this is in the notes because a lot of times it can be missed.

Santosh Sankar 20:42
Well, I think your point is around constantly tinkering, having the curiosity to grow yourself. Right? There’s more than ever. I think as an investor, you have to be willing to constantly learn, evolve, change how you think about the world, but also how you actually execute in the world. And for the last couple years, there’s been this obsession around how you prompt and engineer prompts. And what I just pulled up, and we could even include in the show notes, is there’s now kind of the concept of the loop because people have seen what software engineers have been doing over the last nine to 12 months, and how do we adopt that for work? Which is set the goal, discover what you need in order to achieve the goal, plan how you might actually accomplish it, execute, have a check, and then iterate on it till you get to your goal.

Earnest Sweat 21:36
Yeah, yeah, yeah.

Santosh Sankar 21:37
And that ultimately, I think is important because it demonstrates how quickly you need to change as a venture firm, but two, it kind of gets to the point of what you’re saying is that you need to keep your eye on the prize and what is ultimately the prize. Like our friend Eric Sippel, I remember the first time I met him, he goes, “Listen, I distill venture as a business down to sourcing, selection, stewardship,

Earnest Sweat 22:03
yeah,

Santosh Sankar 22:04
and that you can unbundle into a few other work streams. But that’s why we are all here, yeah. And some of us are more advantaged in certain vectors in order to push on sourcing. We may not be great at stewardship, so we need to be cognizant. Do we bring a syndicate partner in to help with stewardship, selection. I mean, if you’re not good at selection, at some point you’re not going to stay in business. But one would argue that that is second to sourcing wealth and having access.

Earnest Sweat 22:33
Yeah, yeah. Have you felt for about like I think one narrative when AI was really starting to roll in 2020, you know, when Chat GPT was officially out, public launch. There was this thought for both large business or you know tech companies as well as thinking about our industry and venture that you could do more with way less people. So there was this idea of like, oh, we’re going to have a billion-dollar company, which now in today’s times seems like you’re not dreaming high enough. But anyway, you have a billion-dollar company with only one person there or two people there, and how the you know for venture, you know, we’ve had organizations like you know VC Lab, Decile Group share like, hey, now with all these tools, you can be a solo GP and run like a 10 person fund as as a you know GP yourself, not a solo GP, but a kind of a small but mighty team. How do you feel about that narrative?

Santosh Sankar 23:41
Part of the way I think about AI and where I think it is table stakes is all of the work that is repetitive. Not to say it’s not important, but it’s repetitive. You need to do it’s like you cannot go without it. LP reporting, for example, you have to figure out how you can set up a system where you spend less time actually doing the assembly and the lower value parts of that. So, for example, whenever we do a quarterly report, we do an LP letter. We usually do some type of very deep industry thesis that has emerged that quarter, or a survey of something that’s specific to our domain, and then we do like the traditional, you know, slides. Here’s the portfolio update, headline numbers, news. Yeah, that by and large gets written by AI because I’ve been able to create an environment where our agents have all the context it needs, and it gives me a draft, and we go through multiple people on my team go through and comb through that in order to make sure it is accurate and exhaustive and inclusive of everything that’s important. But that allows us to spend time on that kind of thinking piece that we take most of the quarter actually bringing together, and at any given point before we decide to. Write it. There’s probably three or four things that we actually invalidate and say, “Hey, we don’t know if we want to spend our time as a group to do that. Why don’t you go take the core and do that on your own? You can come back and maybe next quarter bubbles up where we share that with our LPs because it’s more baked. So when I think about the solo GP, I think there’s something similar like table stakes. If it’s overhead, you need to figure out how to get yourself out of it. It is not differentiating. It’s not going to make you a better firm or investor. For those things that you can lean into your advantage on, it’s either a time function. How can you lean into your advantage because you avail so much more time in your Google Calendar, or two, is there a thoughtful way to use AI to make your advantage even better, and maybe it’s better, faster, or deeper?

Earnest Sweat 25:54
Yeah,

Santosh Sankar 25:54
depending on what your advantage is, and yeah, like is there a benefit to solo GPs, sure, but you still, I think, are going to have to answer the question. Maybe you don’t ever want a GP, a fellow GP, or a partner, but you need to determine and still answer what I need on my team to be able to win on a you know durable, enduring basis. And I think that comes back to some of what we’ve said before, like how do you bring someone who has a different type of superpower to you, but is still relevant for you to win? I think the the one thing I tell all my solo GP friends is I’m not sure how you’d do it because it would feel lonely, and sometimes you can get in your own head,

Earnest Sweat 26:38
yeah,

Santosh Sankar 26:38
positively or negatively, and it’s nice to have that outlet, and I think different solo GPs over the years have found different ways to manage that. But I personally love at least having one other person I need to speak to before we commit to doing an investment or even working with a portfolio company that might be in a dire situation and needs more than one head at the table.

Earnest Sweat 27:03
Counter. What if that other person is

Santosh Sankar 27:06
an AI? Is an AI. I. I don’t think so. I don’t think the person should be. The question is: Are they AI or are they not AI?

Earnest Sweat 27:16
I’m saying I should like the counter. What if like oh I can get that other entity or

Speaker 1 27:22
as AI as

Earnest Sweat 27:23
AI to challenge me.

Santosh Sankar 27:25
I think I think you raise the point about taste, and there’s also like a lot of VC is EQ stage I’m at. It becomes more quantitative as you go late stage, but I’d argue that actually probably the best investors have been late stage and company builders, what do they say? Culture eat strategy for lunch. Yeah, it’s still about kind of the humanistic qualities, and maybe going back to quote you, never say never. But for the foreseeable future, I don’t think AI demonstrates those soft skills that one needs. If it’s analytical, totally. I think a lot of people now have access to information, or sorry, access to knowledge and information and intelligence, that is kind of going up. Access is going up, and therefore the value of it is slowly creeping down, especially intelligence, because you can offload that. Yeah,

Earnest Sweat 28:16
yeah, yeah.

Santosh Sankar 28:17
It’s all the other humanistic things that you write a lot about as well.

Earnest Sweat 28:20
Expertise, relationship. relationships, people actually returning your calls, like all those things. Is it

Santosh Sankar 28:28
weird? We’re kind of devolved back to what our parents were like teaching us when we were like eight, 910, years old. Just be a good, be a good dude.

Earnest Sweat 28:35
Being a good person is an advantage. I think we’d be remiss if we didn’t actually share some of our experiences with all of this. Like, how do we determine our workflows? What are some of the mistakes we’ve made? What tools are we using today? Into kind of where we see it going. If you want to go first,

Santosh Sankar 28:59
So we’re a very small team, so in total in our firm we have five investors plus an EIR, and then we have Madison who does all operations and non-investment for us. But I think part of it is like I always think about what I tell our portfolio companies. Your only advantage at your size is your ability to be agile, flexible, move fast, and I believe the same as a firm that is investing out of a $55 million fund.

Earnest Sweat 29:32
Yeah,

Santosh Sankar 29:32
and so one of our values. We didn’t document our values till about eight years in because we’re such a small team. It was largely the founders operating, and so when we made them explicit, the one thing my partner John pushed was he goes. We’re actually a lot more creative than we think, and that’s given us a lot of freedom and ability to win. So creativity is something that we really cherish and we promote and we encourage, but. Statement that anyone at Dynamo is allowed to bring a tool, put it on the table that they’re playing with, and try to convince someone else to also adopt. I love that. And there’s a general open mind in this as to why I’m going to try it. I’m not committing to it, and if I don’t think it’s going to work, I’m like no harm, no foul. I’m just not going to use it. And so right now we are in a world where our core AI harness is actually a startup out of Europe called Victor, V I K T O R, and prior to that, we’re actually building a homegrown agent off the back of Lang Chain. But we actually found it much easier to get to the productive bits and the interesting bits by using a pre-built harness with all the integrations. My partner John also uses AMVC on the side for some more venture-specific items, and also he likes some of the UX there. So right now, I think by and large we’re in an era of experimentation, learning quickly, changing quickly, fast I/O loops. Now, about three weeks ago, Claude came out and basically has said we’re introducing Claude Tag. It’ll do the work for you. You can speak to it like a coworker. And actually, in the next couple weeks, as it goes to general release, we’re going to re-platform Dynamo to Claude Tag. And the big question is like, as we do this, yet we’re looking at investments that might be specialty harnesses or agentic frameworks for different industries. How do we reconcile this? That’s like a really challenging thing that I don’t have the answer to, frankly. But where I’m trending is if it has to do with general white collar work, writing documents, doing analysis of that elk, building presentations. I think Claude OpenAI is gonna. You’re always gonna have that base, and over time, you’re gonna have like maybe the Amy or something homegrown, and it’s gonna probably reflect this growing belief that we’re going to be a multi-model world,

Earnest Sweat 32:03
absolutely,

Santosh Sankar 32:04
and so equally, I think you’ll have different harnesses effectively for different parts of your workflow as you really dig in and embed into your industry. Atio is our CRM, and I think it’s been great because it’s super extensible and integratable, similar to granola.

Earnest Sweat 32:22
I love it. A couple things from how you’re approaching it, I think, is important for any firm builder to embrace a culture of being curious and bringing in new things. And you know, there’s been so many questions like where does labor go? Like, there’s this huge tension of like AI is going to take away a lot of jobs right and then you know people counter to that of like no if you look at all these different technological shifts it’s always created way more jobs than it than it did maybe but it doesn’t take away the tension that people had with like hey they’ve invested their careers invested their skill sets, and it might just be wiped away. And so, people ask, like, what should someone middle-aged do? What should people out of college do? And I think, you know, it comes down to self awareness of their things we’ve already been talking about, but self awareness of their skill set and expertise and their superpowers and willingness to bring in new things, and so younger people should bring in, and people who are not so young should be able to bring in new tools and be able to have that ability to show, hey, this is why it’s worked for me. Here’s the ROI, and this is why I think we should implement it as an organization. So that’s a huge skill set, and I think something that all founders, all firm builders should be looking for in talent. The other piece is like you all being able to rapidly prioritize what is important. And so the example of like, hey, this is what we’re going to report to our LPs and work on. This is more important. It’s fully baked out, but keep working on it. You have to have a mind for rapidly prioritizing things, yeah, and having an overall like you’re adaptive and flexible. But I think the best investors, a friend of mine and I were talking about this, the best investors in our industry have a worldview and they believe their investments are pushing towards that worldview, and so anything having a worldview of how your fund should work, what the culture should be, what tools should be utilized is all a part of that.

Santosh Sankar 34:32
So we have an adage of Dynamo’s: as investors, we’re not here to not have opinions, and it’s not enough to have an opinion; you have to share it. Now, sometimes you like to run into some gnarly personalities in the world adventure, almost like the sound of their own noise. But but but but there’s someone new that, like having a worldview, like having an opinion. And one thing you likely know is like I hate someone who doesn’t have an opinion. Like if you don’t have a spine, you’re not. On Instagram, something similar to not having ambition, like nah, like it’s gonna, it’s probably gonna wear on me. I’m not gonna want to spend too much time with you. Yeah, but at the end of the day, things like this are moving quickly.

Earnest Sweat 35:14
Yeah,

Santosh Sankar 35:14
you have to be willing to move quickly with it, and if you don’t, there’s gonna be a finite life for the organization. You’re either charged with building or you’re part of building.

Earnest Sweat 35:24
Yeah.

Santosh Sankar 35:25
At the end of the day,

Earnest Sweat 35:26
My IRO does real quick on kind of my journey like this. And so, as you know, doing SPVs, working with clients who are either emerging managers who are becoming more institutional, or LPs or large corporations. I’m getting to test out a lot of different things in a lot of different scenarios for different stakeholders, and so I have like my worldview is that you should be like that example for the kind of like grandmaster in chess. You should be able to use a machine and become a centaur, right? And this idea of literally blowing out the Dunbar’s number should be, if you’re an intentional, very connected person, you should be able to blow past that 140 number. And so I’ve always been focused. This is why I’ve, in my prior life, wanted to create essentially an agentic CRM. I’ve always been looking for the best CRMs. That and now with like where agentic AI is, Gen AI is. There’s a belief. Okay, I should be able to be way more intentional and utilize my great memory, but also have a backup brain, and so I’ve always just been focusing on that because that’s my strength and that’s what gives me energy. And so, first journey about two years ago, for some reason, was just using a lot of fucking money on Red Lit and just trying to create this like cerebro been there, just like yo. This is cool, and my family’s like, my wife’s like, what? What are you doing?

Santosh Sankar 37:05
What is this thing that you just put five grand on over a weekend

Earnest Sweat 37:09
that you’re not even using? It’s just like, oh, that was cool. You know, vibe coding, just vibes. Then moving on to, all right, using. You know a premium version of chat, but still kind of using it as a chat bot and asking questions. It’s like how do I automate more? And then as you know, co-working and codecs came out, realizing oh I can actually build these workflows, but I still had a mentality of like, oh, I’m going to really be prescriptive. Assume you know for a workflow, this agent does this, then it moves over to this agent that does this. So essentially, I would have like this to do bot that I called it associate, and anything it couldn’t do on its own, then it moved over to partner bot, and then we would have like, yeah, discussions and and help it like, okay, this is what you need to do for this project. Here’s my expertise, and I was getting some pretty good, you know, productivity from far better

Santosh Sankar 38:17
than 95% of people out there are. Yeah, yeah,

Earnest Sweat 38:19
but it can get very weird as you have your MCPs and all your data, and it can know I had to get a second computer or I had an old computer that was like I’m just going to have this on doing that. It could go a lot slower, right? And you can make some mistakes. And so I think one thing that’s also key is not being afraid to ask questions. So there’s a founder based in San Francisco, Gen Z founder, really focused on how do we not think about this kind of human to agent AI loop, machine loop, but actually have it separated where they’re actually their own employees. You do your work, they do theirs and actually even blowing it out to like they run whole departments, and he was so he was telling me this story about how he helped his girlfriend with her law law school exam and helped her do it in like 30 minutes instead of her doing it in like like three hours, like exactly, like putting it in and then asking for the answer and all that, and so I was like, “Hey, man, I was like, “Hey, I need to redo my my workflows, and so he goes to his bath and I show him like, you know, partner bot and re engagement bot and associate bot, and he was like, “I just want to before I get into my feedback. He’s like, “How technical are you? And I’m like, “I’m not that technical at all. He’s like, “Well, let me gage. He’s like, “Do you know what a system prompt is? I was like, “I know what a prompt is, but I don’t know if I know what a system prompt is. He’s like, “Do you know what a context window is? And I was like, “No, I’m sorry, I don’t. And then he was. Like, do you know what tokens are? And I was like, Yeah, know those. And he’s like, Okay. So if it’s not in headlines, you don’t know what is. All right, cool. He called a VC a VC. And I was like, Fair. All right. But dude, so nice. Walk me through, and I’ll probably write about this at some point. But like, how to actually leverage these models where you should be using each now, and this is what I do now. Like, what is the entire workflow for one thing? So it’s like, if it’s like you know having the first pass of draft of emails, it takes the whole thing of that up to

Santosh Sankar 40:39
the idea to the draft of the email, exactly,

Earnest Sweat 40:42
or if it’s some workflow around, hey, prep for you’re going to have a guest on a podcast. It does the entire thing. It doesn’t give off to other thing, other people, or all the to dos that come from your from your AI notes with with on granola or fathom, and then set it up as those agents never do the work because we’ve all been through this example of like, damn, this it’s getting slower as I tell it more because it can’t remember a lot of things. Context window learning, the more you know, but saying that those should now be managers essentially, and they’re designating to all these sub agents every time, so now they can do a ton of sub agents at the same time, and now your work can be thorough and complex. And so even that process is like, all right, I need to be able to be fine with redoing everything that my approach, which I thought made sense in my head, but because of the technology, actually didn’t fit.

Santosh Sankar 41:49
And you might need to redo it two or three times till you’re like this good for a couple years.

Earnest Sweat 41:54
Exactly.

Santosh Sankar 41:56
The one thing I remind people is a lot of how you get AI performant is not different than when you have new hires coming in. You need to document how work is done. You don’t kind of compartmentalize it. You basically go, hey, like we’re doing diligence, or you’re drafting cold outbound to companies. This is how you do it from like zero to actually hitting the send button.

Earnest Sweat 42:20
Yeah,

Santosh Sankar 42:21
and that’s not dissimilar. Where I think a lot of work is for those who have actually documented key processes for training, a lot of that work actually is relevant. If you’re trying to just get your AI going to like a version zero, where like you NASA do something, comes back and it’s remotely what you want, and what we actually did was we took some of those documents and artifacts we had, but I used it as an excuse to go revisit them and improve them because some of them had been around for 12, some 20-four months without being, you know, touched, and then with AI, I was like, oh, we don’t need to do this. This is a function of a human being there,

Earnest Sweat 43:00
yeah,

Santosh Sankar 43:00
or we can actually do that better because AI can take all the information in the world. We only focus on these things because a human only has so much headspace.

Speaker 2 43:08
Yeah.

Santosh Sankar 43:08
But I mean, it’s fascinating, and I’m excited to see what people discover.

Earnest Sweat 43:17
Yeah.

Santosh Sankar 43:18
And we kind of talked about there’s going to be discovery. It’s going to feel like a novelty for the first two quarters, and then next thing you know, as soon as it’s shared, every other firm is going to start, provided they agree that it’s valuable. Yeah, I am going to adopt it.

Earnest Sweat 43:32
It’s why I think I’ll leave us on this note. It’s why I think the IC individual contributor is dead forever, because even if you are working just with, you’re going to be managing agents.