Ep. #8, The Beauty of Vertical SaaS with John Melas-Kyriazi of Standard Metrics

Christine Spang: Hey, everybody, welcome to the show. Today, we have with us John Melas-Kyriazi.

John is the founder and CEO of Standard Metrics, a company that is building a platform for easing the communication and collaboration between investors and their portfolio companies.

Before that, John spent six years in the venture world himself at Spark Capital, which is a really great firm that actually back in the day also invested in Nylas.

So we have a great connection there. And another connection that John and I have is John is from Boston where I actually went to school also.

So great to have you on the show today, John, welcome.

John Melas-Kyriazi: Thanks so much for having me on. I appreciate it.

Christine: To get things kicked off, I'd love for you to just give us like super quick intro about you, where you're from, and also just tell us like what's the vision for Standard Metrics and how did you kind of decide that you needed to build this company?

John: Yeah, so as you mentioned already, grew up in Boston, I ended up going to school out in the West Coast at Stanford and like many other Stanford students, kind of immediately got sucked into the engineering world.

I was an undergrad in engineering, studied more kind of hard sciences in engineering, material science and started a PhD there as well actually after undergrad.

And while I was starting my PhD, I ended up meeting this group of people who were kicking off this new startup accelerator program off of Stanford's campus called StartX and started working there a little bit part-time and then pretty soon, fell in love with the program and with this incredible community of entrepreneurs off of Stanford's campus and ended up dropping out of my PhD and joined there full time.

Spent a couple of years helping to run StartX, which is a ton of fun, really exciting time to be there.

And fast forward a couple of years, my girlfriend at the time, now my wife Christina, decided to go to Boston to go to business school at Harvard.

So I kind of packed up my bags and moved back to Boston where I was from and met the team at Spark and they were, you know, looking to hire a new associate.

Actually, the person who I met first was also the one who led the Sparks investment in Nylas, Andrew Parker, he's also a Stanford grad.

And we kind of hit it off and ended up joining the firm originally in the Boston office. And then after two years there, moved back out to San Francisco and spent about another four years or so with Spark.

So that was kind of my entry into the venture world was from PhD student kind of drop out to startup accelerator helper to you know, VC.

And while I was at Spark, one of the things that really struck me is that because this was this kind of cottage industry that had just recently started really growing and becoming a real asset class, you know now roughly a $3 trillion global asset class.

Very, very few companies had thought to really build great software for Venture. So we were cobbling together all of these tools.

There were some awesome purpose-built tools, like we used this tool called Affinity that I helped to kind of get in the firm a CRM product.

But a lot of the time, we were kind of struggling and ultimately resorting to using spreadsheets and Microsoft's office to Power Core workflows, you know, from that that were, you know, managed and manages today billions of dollars.

So I kind of became obsessed with this problem of like who's going to build, you know, the next, you know, real financial software platform for venture and for other adjacent asset classes?

And that's when I met Joe Lonsdale and Alex Moore and the team over at 8VC. and they had had a similar set of challenges at 8VC, also a Nylas investor.

And they, in typical 8VC fashion said, "Hey, let's go like hack on an internal tool and try to build something ourselves."

'Cause they had tried buying this off-the-shelf kind of PE, old school solution that didn't work well for them.

And long story short, I met them, they had been hacking on this internal solution, they started using it a little bit, it was very clunky, but it felt like there was some sort of product market fit there and I decided to join them and start Standard Metrics and try to build a next generation financial reporting platform, really with the folks to start off how do we automate financial reporting between companies and investors and make that process easy and useful for both sides, which is really what we've been focusing on for the last five years.

Christine: Yeah, I think it's super interesting that it was not that long ago that people had no software that was really built for specific verticals and specific problems.

Like yeah, it's just crazy for me to hear that like, as little as I guess 10 years ago is like a lot in software years, but like in the grand scheme of like the world and the history of technology, like 10 years is like not that long.

And just hearing you sort of tell your story of like, you know, you spent five years building this company and before that, you were in venture for six years.

And that was like the time when you guys like had like no solution at all. The idea that you have to use, you know, just like a basic spreadsheet or Microsoft Word and like all this very general purpose software that has no knowledge of the workflows of this type of business, how people tend to work together, what their specific needs are is something that almost seems like a little wild thinking back based off of the state of the world today where...

I was reading the Stripe annual letter a few weeks ago when it came out. In that letter, Patrick and John, who are the founders of Stripe, they talk about how vertical SaaS is this trend that has really kind of taken off and exploded over the last five plus years, let's call it, where like basically software has gotten easy enough and there's powerful enough primitives that like, now it makes sense to build specific platforms that encode the specific workflows of specific businesses and make it really easy and kind of just like out of the box to support those.

And correct me if I'm off here, but I kind of think of Standard Metrics as being like a part of this like broader explosion and ecosystem where like you guys set out, you know, with this knowledge of like, what are the pain points in venture and what do these people need in terms of like the workflows, the UI, the support to just like make it really easy and intuitive and went out and built like a purpose-built solution, a platform for this specific industry.

John: That's right, yeah. I do think that the interesting change though that's occurring is AI, right? Like that is definitely changing the way that the vertical software.

Christine: And that wasn't something that was really present when you started, right?

John: Not at all. I mean of course, there were AI companies like back at Spark when I was there for example, we invested in cruise in some, you know, really interesting companies leveraging AI.

Christine: The term almost meant like something totally different.

John: Yeah, it's pre-LLM, right? This is pre-OpenAI.

And now I think as a vertical software entrepreneur, you have to imagine that, and I think we're seeing this play out in practice, that the barrier to entry to build software is going down. There's going to be more competition, it is going to be easier for people to build their own software and it's going to be easier for lots of competitors to emerge that can very, very rapidly, quickly build across a larger product surface area than was possible before.

So that kind of opens up an interesting series of questions around, well how do you build competitive barriers in a vertical software business now that the barriers to build software have gone way down? I'd love to talk about that with you.

Christine: Yeah, I think it would be helpful if maybe you could just like kind of like walk us through like the history a little bit of Standard Metrics and like what were like the things that you focused on upfront that were kind of just like the basics to get in place and then like sort of halfway through, how did that change?

And then like today, you know, how are you thinking about like has technology changed that enables you to shift different functionality?

And also how has like the ecosystem changed around you that makes you think differently about just sort of like the strategy for being most useful, driving the most value?

John: Yeah, that's a great question. So I think to start, you know, every company kind of starts out with this question of like, what can we build that people are going to actually find useful and use?

And for us, we had this advantage of being originally incubated within 8VC and had the opportunity to really work hand in hand with the team there and some other firms that were friendly to 8VC and interview users and watch users use our products, you know, effectively in person.

The first thing that we realized is that CFOs and firms and chief financial officers, they have responsibilities that entail oftentimes data collection and kind of data hygiene across the firm because a lot of workflows that use data that a lot of people don't even think about.

For example, valuations and LP reporting in audits and portfolio reviews. These are things that kind of happen in the background to enable a lot of the activities that you then read about, you know, in the Wall Street Journal or TechCrunch.

When people think of a venture capitalist, they think of, you know, the board member at X, Y, and Z firm that led the investment in, you know, Wiz or Airbnb or whatever it is. But there's actually a team of people behind them that's helping to-

Christine: Nobody remembers the like analyst who's pulling 100-hour weeks in the background doing all the research and getting all the data.

John: Yes, but interestingly, I'm not even talking about the analyst, I'm talking about the controller at the firm who's in charge of making sure that data is organized and gathered and clean and audited and available and then used in a bunch of these kind of really important back office, you know, workflows and front office workflows that enable things like investment decision making, fundraising reporting, etc.

So that was one of the first things that we discovered. And that was something that was pretty foreign to me because when I was at Spark, I was on the investment team, so I was that, you know, associate and then, you know, over time, more senior person who was, you know, helping to source investment opportunities, due diligence and then ultimately, you know, start to lead my own investments and kind of take some of my own board seats as I spent more time with the firm.

And I interacted with the finance team and I helped the finance team with some of these workflows. But there's a lot going on under the hood that I actually didn't even realize until after I left and I started working with some of these customers more deeply.

So that was the first thing was really starting to understand the personas within these customers and also starting to see this kind of emerging new area that was growing, which was, you know, a lot of the time people think about a back office in a front office at a investment firm, but there's this almost emerging area called like, some people call a middle office of people who are in charge of data.

And they tend to be more technical. Sometimes they involve data engineers or data scientists. Maybe they're, even now, some of our bigger customers are spinning up a data warehouse and they're thinking a lot about data strategy across the firm.

And so for us, identifying these personas and figuring out their needs is a really critical early step. And what we realized is that they have huge challenges around collection of data from portfolio companies.

It's manual, it's messy, it's error-prone, it creates lots of challenges for the portfolio companies, it creates lots of challenges for them. And so we focus all of our energy on trying to make that process seamless and useful. And that's really, you know, I'd say for the first couple of years by far our largest focus and we did build some downstream applications, which I can, I can talk through around, you know, analytics and reporting and things like that, but solving that data collection problem was actually the biggest one.

And that was the unsolved problem when we kind of started building in this problem space.

Christine: That's super interesting. Did people want that data to live in a platform like Standard Metrics or do they want it also just like in a copy in their own data warehouse?

John: Yeah, good question. Well, so when we first started, we actually wrote a blog post about this over the last, I think six months.

When we first started the company, we're like, "Well, why would anyone want this anywhere else?" Of course they want it, you know, in our platform.

And over time, we realized that customers do want a core system of record for this data and they want the data to live in a easy to use SaaS application that has permissions attached to it, etc. etc. But they also need to access that data in other places.

One of those places is Microsoft Excel. Because a lot of firms are building models of all different types that are really nuanced and really sophisticated and they need to be able to access that data, but they need to be able to do it from a place like Microsoft Excel.

Another example, which I already brought up is the data warehouse. Firms are running a bunch of analysis, they're building applications, some are even building interesting AI applications.

They need all these different types of data that they care about, including data that we're in charge of to live in a place where it's easily accessible.

So as we started building, we realized that, you know, we needed to be a part of an ecosystem and not just an ecosystem of horizontal data tools like BI tools and, you know, spreadsheets and warehouses, but also other tools that are used in our industry that are kind of complementary to what we do.

For example, we just launched a partnership with a valuations provider. They work with venture and private equity firms and help them to value their portfolio.

We now have a partnership where data comes in to Standard Metrics, it can pipe into the valuations provider, they can run their valuations, they can send the data back, and all of that lives kind of seamlessly.

Those types of things matter a lot to our users. They want to save time, they're incredibly busy. And so time savings is a big part of the pitch for us is that we can allow people to collect more data, use that data more effectively in a really, really time efficient way. And so integrations are a really important part of that story.

Not just for investors by the way, but also for portfolio companies.

They're busy, you know, you guys are doing a million things at Nylas. Sending, you know, structure financial data to your investors is probably not the top item on your priority list.

How can we make that really fast and easy for you by also providing integrations in various automation, which is something that we started tackling really from day one with product.

Christine: Yeah, definitely not something I want to spend more than 30 seconds thinking about on a regular basis.

There's a couple things that you mentioned that really kind of stood out to me.

One is that, you know, everybody needs some sort of system of record for all the different sort of relevant pieces of information for their business.

It sounds like in the case of Standard Metrics, that's kind of the financial data, the reporting data from all of these companies.

Another thing that stood out to me was just this idea that data can't just like live in one tool these days.

Like it needs to be usable and sort of connected to a lot of different potential other front ends.

I think there's like the beauty of like vertical SaaS is that, you know, it's very simple and easy to use and you get like, you know, most of what you need in kind of one nice box, but there's always like something that like people want in terms of like extensibility that like, you're not going to predict that particular use case or thing or it's just like not the focus of, you know, what you want your platform to be.

So it makes sense to me that you'd want to make it easy to connect to like a data warehouse.

Other tools just, you know, everyone's used to doing manipulations of data in a spreadsheet who wants to build like a whole entire spreadsheet into, you know, every single app out there.

You know, that's a whole lot of software to have to build.

So I think this sort of tension or maybe not tension, but I think we all have to kind of think strategically about how we design products to fit in like a broad ecosystem or kind of the stack of end users and what you decide to kind of take on yourself versus like having interoperability with other things is a really major sort of strategic process when designing products.

Like for us at Nylas, we also kind of have this same problem in sort of a different way where, you know, we want to provide like these reusable building blocks for people and there are like a bunch of core workflows that are very sort of portable because like people have similar communication needs no matter where they are.

But we expect that people are taking that data and they're saving it in other platforms and we're just sort of like a connector that helps get that data there in a really useful way that's reliable and scalable.

John: Yeah, I think that, as I said before, definitely one of the fundamental strategic questions that we've been grappling with since we started the company, and as time has gone on and as our network has gotten larger and larger, both in terms of portfolio companies and investors, I think we've leaned in more and more to the idea that--

Ultimately users should be in control of their data and they should be in control of how it's used both in our platform and off of our platform.

And our job is solve problems for our users that kind of pertain to their data. And if the best place for a user to solve a problem with this data is not on Standard Metrics, they should be empowered to go and do that effectively and that will ultimately build a stronger customer relationship and be better for us and be better for them.

It sounds simple, but it's actually hard to do well, right? Because you then need to get good at building like, you know, effectively developer tools and it's kind of a whole nother from a product management perspective, a whole nother can of worms.

But yeah, it's a continued focus for us in something that we're very, very committed to and excited to share more, you know, over the coming quarters and so on about partnerships that we want to do there.

But like at the end of the day, we're not going to solve every single problem that our users have and we want to make sure that they can solve other problems with our help and that requires interoperability.

Isaac Nassimi: I think that's wonderful to, you know, try and be a part of their stack or their data stack and not try and not force on the regular the whole thing.

And actually, the question about that, 'cause anytime that I find that you build a platform that kind of generally solves a problem, right?

Where you say, "Hey, we're going to get everything in and we're going to let you solve it inside of this platform," you know, you have an easy time getting in front of customers and you have an easy time kind of getting them mentally, like understanding what the product could do for them and excited.

But there's usually like a large onboarding lift, right? In this case I'm imagining there's a gargantuan amount of data that has to be dumped and then worked upon and you know, you to get cross-departmental buy-in from the customers.

And I imagine this is something that took you guys a while to figure out and really nail, which you have nailed today.

I'd love to hear, you know, how that kind of affected, you know, your business level strategy, your user acquisition strategy, and how you got to that point of nailing it.

John: Yeah, and look, I think there's always room for improvement, to be honest with you. Like, I feel really proud of the way that we onboard our customers.

We tend to onboard our customers much more quickly than some of the other, you know, legacy providers in our space.

We tend to prioritize speed in general at our company, but especially around onboarding. Time to value matters so much.

You know, sometimes you talk to folks who say, "Oh my gosh, it took me 12 months to get this implemented, took me 24 months to get this implemented."

We think that the world like just demands more than that now. Now look, there are some certain situations where things are really complicated.

And sometimes at the outside of your control as a software vendor when someone, you know, needs a lot of time to get organized for one reason or another.

But we try to move as fast as we possibly can. You know, I think it's taken a lot of learnings, a lot of trial and error. I think there are some early customers, you know, candidly that we brought on where a year in we said, "Gosh, we didn't do a great job onboarding them."

You know, like if we had done a better job onboarding them, they would've gotten much more value out of our product.

So there are some, you know, hard lessons learned in the early days, but it has been something that we've invested into a lot that we feel really proud of. We have a great team behind.

I think we've overinvested in having really smart, really mission-driven people who care a lot about this problem and building technology to make that easy.

For example, one of the things we've been investing in a lot over the last six months or so is using LLMs to help us to rapidly and accurately extract data out of documents.

And it turns out that LLMs, in concert with highly trained humans, are amazing at ingesting lots of data out of documents and that the combination of the two is better than either on its own.

And so like that's an area where we can move really, really fast to help customers onboard lots of historical data.

Another thing that's that's interesting for us is that, you know, we are an ecosystem, right? Like we don't just sell to investors.

We also, you know, investors distribute our product to their portfolio companies.

We now have almost 9,000 companies that are on Standard Metrics through their investors.

And as we get more and more investors on, there's more and more portfolio companies and there's more and more overlap where portfolio companies might work with two investors or three investors or four investors that are users of Standard Metrics.

So it actually gets easier for us to onboard over time because when we sell to a net new investor, a decent probably oftentimes double digit percentage of their portfolio companies are already on Standard Metrics and connecting with them on the platform is like kind of a LinkedIn request.

But at the beginning, there was no network. At the beginning, everyone was a new user basically.

And so there's been a lot of learnings there about how to do that well, how to offer the right kind of support to the portfolio companies and to investors who are using our product.

And as I said before, still lots of room to grow.

We're constantly learning, we're constantly iterating and you know, I'm sure that will iterate on our onboarding process many times over the years to come to make it better and faster and more effective for our customers.

Isaac: Yeah, what's the big piece of advice or the single piece of advice you would give to someone who's trying to solve this problem within their company today?

John: So one of our angel investors and advisor, Zac Bookman, who's the CEO of OpenGov, he has an awesome blog post that I highly recommend reading.

I don't remember exactly what it's called, but it's something along the lines of like why professional services are important to enterprise software companies.

And basically, the thesis of the blog post is that a lot of software companies underinvests in professional services because people build software companies 'cause they want to build and sell software, not because they want to sell professional services, but professional services can be an amazing unlock particularly around onboarding.

I think that going the extra mile and in many cases, charging sometimes significant amounts of money to customers to make sure that they have an unbelievable onboarding process where they exit that onboarding process with a system that is ready to meet their needs and solve their problems--It's easy to overlook that as an early stage company, easy to say, "Oh, someone internally is going to figure that out." And I think it's really important to take ownership and take responsibility over that. It's your job to make sure that they got onboarded well.

Obviously, there's people on the inside on the other side who are responsible for that too, but ultimately if it doesn't work out well, it's your fault.

And like that's going to be a problem for you. So that's probably my biggest piece of advice.

Isaac: I really like that kind of like extreme ownership attitude.

And I completely agree, professional services is something that probably feels difficult to pull the trigger on 'cause you feel like you might, you know, reduce, you know, your conversion rate or your close rate, but at the same time, you know, you want people to get to a place where they're using your platform really quickly and you know, getting that value back from it.

I think that's awesome. And you know, also reading between the lines of what you were saying before of like adding an incremental portfolio company for an investor.

You guys are really becoming the industry standard here, so I'm guessing you really have your finger on the pulse of what investors are looking for and what investors care about today.

How do you see that changing in today's world? Because things are moving fast and the macroeconomic landscape is changing, the technology landscape is changing and I'm sure that they're really shifting the priorities.

John: Yeah, it is interesting and one of the things that we started to do over the last two quarters to start to publish some of this data, we have this benchmarking product that we've built.

One of the ways that we try to give back to portfolio companies is we give them access for free to industry benchmarks that are relevant to their company at, you know, their revenue scale and their sector.

That's something that they access just like through Standard Metrics when they sign up and get data into the system.

And we also have a tool that we sell to investors that help them to benchmark their portfolio companies against what's going on more broadly in the market.

But there's this incredibly rich underlying set of data and we started to publish kind of industry trends and pieces around that.

We actually are recently kind of went live with our Q4 report. And it is striking how things have changed.

I mean, there's too many different changes to talk through like in a short podcast, but like, you know, a couple of examples.

One is the kind of relationship between growth and burn is fundamentally changed.

Like late stage companies basically aren't burning cash anymore, which was not true at all a few years ago. They're growing more slowly and they're not burning cash.

But early stage companies, in fact, like some of the, like especially the AI companies, if you take them out and look at them as at their own cohort are growing much faster than what we saw previously a few years ago.

So you have this kind of interesting dynamic where later stage companies are starting to look very fiscally responsible and early stage companies, especially AI companies are growing ludicrously fast.

And so there was this kind of like almost a gulf that exists in many ways between some of these AI native companies. You'll read the, you know, in the news, like, I forget the recent example.

I think Lovable grew from like zero to 17 million in a RR in like three months or four months or something along those lines, which would've been unthinkably fast five years ago, right?

Like no company in history had ever grown that fast. You know, Cursor went to 100 million in ARR in like less than a year, roughly a year. Fastest-growing software of company of all time. Things like that are happening.

And then on the other side, you know, the average company with 100 million in revenue or above is basically cashflow break even.

So we have seen a big kind of change and a big shift there as for what investors are looking for exactly that obviously has to do with taste, right?

And every investor kind of has their own thesis, their own lens, you know, different things that they're looking for. But we do see a dramatic shift in kind of some of the underlying financial dynamics.

And it's fun to have that bird's eye view kind of across this aggregated and anonymized data.

And we want to find ways of making that more and more actionable and useful for people who are using Standard Metrics.

Isaac: That's awesome. And you know, I went to the record, it's staggering how fast some stuff has changed, right?

I think gross margin is just a excellent example. Because that's not taking into account or, I mean that's aside some things like headcount changes and whatnot generally, I mean it looks like the average between all revenue ranges has gone from like 60% to 70% in 2 1/2, three years.

And I mean, that doesn't sound like a lot, but that's a really, really insane average, right? Because no one's hitting 100% or anything like that. I hope not.

You know, what it kind of reminds me of is kind of the olden days of startup world where these early stages companies that really nailed a product exploding onto the scene was very common.

And the idea was we're going to invest and invest and invest in user acquisition and then slowly start to turn on kind of the profit, the profit levers there, which is just, it's kind of cool to return to the roots of a model, assuming that what I said is correct.

Do you have any thoughts on how the VC or just the startup landscape is going to change in the future based off of what you talked about of it being lower barrier to entry to build this software?

John: Yeah, yeah, it's interesting.

So I think what we've seen over the last year or two is the wave of AI native companies that are growing incredibly fast, that have unbelievably strong and exciting early product market fit that can scale to millions, tens of millions of dollars in revenue seemingly overnight. I think what will be interesting to see is how do those companies build competitive barriers to entry around their businesses when they're faced with massive amounts of competition?

How do they deepen the kind of their wedges into the different customers that are paying for their products?

I think Cursor's a great example, right? Cursor's built this unbelievable IDE we have a bunch of engineers at our company that use Cursor.

You know, if Cursor stopped innovating on their product right now and just went like pencils down and waited, there would be like a hundred Cursors in, you know, a certain amount of time all competing on price and functionality and features and stuff.

So how do you continue to stay ahead and how do you kind of broaden and deepen and ultimately build something that is unbelievably sticky.

Because you need something that's an unbelievably sticky ultimately to build a great software company.

Like I think you'd probably struggle to find a single successful public software company that doesn't have strong gross retention and net retention metrics.

So I think that's the interesting challenge. You know, on the way up though, the growth is unbelievable and a lot of these companies start to flow cash very early too.

You know, like some of these companies that have 10 employees and have, you know, $30 million in ARR, these are businesses that are generating lots and lots and lots of cash on the way up.

So it is exciting and I think that there's going to be challenges ahead for people to figure out.

And interestingly, my perspective on this at least is that the way that folks are going to build competitive barriers to entry in the future is not so different from the way that they built competitive barriers to entry in the past.

It's going to be things like network effects, like that's one area that we're really focused on obviously is building a dense, you know, user network.

That's something that AI doesn't change. But I certainly do believe that companies that don't embrace these tools are going to have a lot of trouble because others are.

And so it is a little bit of an arms race and it feels like every SaaS company right now is sort of tooling up embracing these tools both for their organization and also in their products.

And asking these tough long-term questions of like, okay, the barriers to entry have gone down is faster, it's easier, it's cheaper to build software, how do we maintain a competitive advantage at scale with lots more competition and probably pricing pressure?

So I'm really curious to see how it plays out, but I think it's going to be a more fiercely competitive startup landscape than we've ever seen and we'll probably get only more competitive over time.

Like that's just the direction that things are perpetually headed in at this point.

Christine: I definitely agree with that. I forget who it was who said it, but somebody maybe on Twitter said that it's never been easier to make software and it's never been harder to make money with software.

John: Yep, yep, for sure.

Christine: And yeah, I just see that, you know, the sort of cycle of idea to something existing in production is speeding up, getting faster, more people are able to participate in the creation of software because the barriers to entry have decreased.

And then, you know, I think AI has, you know, it has enabled this really fast growth for like products that are kind of ahead of the curve, but there also is kind of on the back end, you know, margin pressure, things people haven't figured out around like charging for the amount of usage that, you know, they're actually using to drive the values of those features.

And then I guess one thing that we've observed from our side is that, you know, we have like a lot of vertical SaaS apps actually that use our platform.

And it seems like there's like a market pressure to consolidate and to include more functionality into each individual platform.

Like you want to have your one solution that gives you like everything, you need to run this kind of business.

And it'll be interesting to see sort of how that evolves over time.

John: Yeah, for sure. I mean it ties back with our previous conversation around interoperability and ecosystem building versus like, do you go truly vertical and try to own something completely end to end?

As with most things in life, the answer is probably kind of in the messy middle.

I do think that this notion of like a compound startup is really interesting and I think there's a lot of great examples of businesses, software businesses that are kind of taking this approach.

Christine: What do you mean by compound startup?

John: You know, I think, I believe it was Parker at Rippling. It might not have been him, but somebody kind of coined this idea of, you know, ultimately, like software, the next generation of software companies that are successful are going to build many, many, many, many, many different applications that are kind of unified by a single platform and need to be able to solve lots of different user problems.

The era of like best in class point solutions that might be right for certain types of customers, but many customers will want to buy like a larger bundle and that there's like significant economies of scale in building, supporting, selling, marketing larger kind of bundles of products that are all sort of natively stitched together than kind of single point solution products.

You know, I'm probably doing a bad job explaining the kind of Rippling thesis, but Parker's written a lot about that. I think that there's definitely, I mean, I buy that for sure.

Isaac: Can you give a concrete example of that?

John: I'll give you an example of that for a product that we use. So we're customers of Brex.

We started out using Brex for card, then when they launched cash, we started managing our cash with Brex.

And then when they launched expense management, we started using, you know, Brex for expense management and we cut out Expensify.

And then when they launched, you know, BillPay, like accounts payable, we cut out, you know, bill.com, which we're using for that. And now everything lives all in one platform.

And it's really nice for us because like, I can search something and it searches across all of these different, you know, payment methods and areas and tools very, very natively.

And I kind of have one place to log in to manage a lot of my finances. But there's a limit to the edge of that, right?

Like for example, like Brex isn't our general ledger. We don't build our kind of FP&A function in Brex.

So like we doesn't need to do everything for us, but it is nice to have a number of different functions all bundled and be able to solve multiple problems from one place.

And by the way, for them, they probably have like one CSM who's in charge of our account across many different products.

And so there are some interesting economies of scale there to think through also. And I think AI will make that easier because AI makes it faster to build products. If you have an incredible point solution, it's faster to go and build some additional product areas on top of that than it was previously.

So yeah, I imagine that there will be kind of continued pressure in that direction as well.

Isaac: Do you see that as a way that larger companies kind of create and maintain their mote? Or do you see that as something that's going to be accessible for even smaller companies?

John: I think historically it has been, I mean, if you look at like a Microsoft for example, right?

It's like why did Azure grow like crazy? Well, like Microsoft has this like, unbelievable distribution into like every business in the world, especially every enterprise and everyone's moving into the cloud.

They already buy lots of Microsoft products like easy to win business against other cloud provider when you have that unfair advantage around distribution.

I think that for startups it's harder, right? Because for startups you have to be like laser focused, at least at the beginning, otherwise you're going to lose, no one wants to use a mediocre kind of point solution.

They want to use an incredible point solution that's way better than what they used before.

But if you wait too long to broaden out into other areas, you risk being kind of isolated and outcompeted.

So again, it's nuance, but I think that startups kind of have to some extent continue to move in that direction.

Otherwise, they run the risk of solving one problem really well, but kind of getting muscled out by another platform that solves multiple problems really well.

Isaac: Yeah. Yeah, I mean, I feel like I've seen this a lot with those, there's a lot of little startups that are really glorified ChatGPT wrappers, right?

And that's not to minimize them like, except for on, you know, a functional level because they really do drive value, but at the same time, for a lot of them, there's not enough to be able to truly get you off of a more robust solution, right?

Because they're offering just a slice of it that is better, admittedly.

John: Yeah. And then back to the notion of competition, right? Like maybe it is enough to get you off, but then there's 15 others that are all like cheaper and faster and better.

So like, and that's where kind of broadening the reach, broadening the set of personas that you're serving within a company or a firm broadening the set of different workflows that you're powering, those things are valuable.

But then figuring out the edges of that is hard. As we discussed before, where do you stop? 'Cause you can't do everything for everyone.

So you kind of have to draw a box around what your area to some extent and figure out where the limits of that are.

Isaac: How much of all of these, you know, productivity improvements do you think that we're actually getting today?

Because, you know, if I look at the average engineer, are they, how much more productive would you say they are using these LLM-based tools?

John: That's a great question.

I feel like it's a hotly debated topic right now because there's like, you literally go on LinkedIn and it's like talking head number one is like, "These products don't work, you know, these products create slop code."

And then talking head number two is like, two engineers can go build like a Fortune 500 size software business and you know, everywhere in between. I don't know.

I mean from what I'm seeing across a lot of these different types of productivity tools, you'll see double digit improvements in productivity. But there's a big difference between 10% and 90%.

Isaac: Absolutely. Especially when you're a smaller company, they're massive.

John: For sure. When you look at the rate of improvements, like we leverage OpenAI's models and Anthropic's models for document parsing, it's kind of like a core application for us.

And the rate at which these models are improving is staggering.

So a lot of the times too, it's like you can look at a efficiency improvement from, you know, a team member and say, "Hey, this person's, you know, 12% more efficient than they were, you know, three months ago or six months ago by adopting this new tool."

But you can also see an arc there that's like, hey, if this type of progress continues, like they can be 50% more efficient in the relatively near future. And that's unbelievably exciting.

So I do still think that we're kind of at the beginning of this wave and there's a lot of efficiency improvements to come. And so it's important that people don't underinvest in these things, but also that they have reasonable expectations for what they can get in the near term.

Isaac: It reminds me of that old quote of like, "Everyone overestimates what they can do in a year and underestimates what they can do in 10."

John: Yeah.

Isaac: And I think like the AI stuff is kind of like that and like, hey, has it changed in the past year?

Like absolutely it has, but you know, you haven't gotten those 40x improvements yet, but I know they'll be here in the next 10 years. I just don't know about the next one.

John: Yeah, for sure. I think it's really exciting. It's a really fun time to be building a company right now 'cause everything is changing so fast. I personally find that really exciting.

Christine: I think the world is kind of split in terms of, you know, there's the super entrepreneurial types who are like, "Everything is speeding up and it's getting easier to build stuff and I can like turn my ideas into reality so quickly and easily and that's awesome."

And then there's like a whole lot of other people out there who are just like, "Oh my God, this is crazy. I can't keep up. Like, it's like a bit threatening."

Isaac: I remember seeing those old cases of people who use like Firebase or whatever to like build, you know, the backend for their startup and there's like this huge changeover when it becomes time to make it real.

And you know, they actually have to start from scratch, but they have, you know, they have a foundation at the same time.

And I think we're going to see that with a lot of these people who are like vibe coding, right?

I've seen someone who they were doing like a build in public thing and you know, building your startup just based off of, and by the way, for those who don't know vibe coding is when you don't do a single line of code.

You just continuously prompt the AI and let it drive the car for you and just kind of hope it doesn't go off a cliff or maybe, you know, steer back off if it starts going towards the cliff.

And you know, they're able to build something and actually come out with it and they run through these problems where like people are abusing the API because it's not properly secured and they just don't know how to fix it.

And I think these are really going to be really, really interesting growing pains. It is just nice being in a period where you don't really know what's going on. Like it's very fresh.

Christine: Yeah, I just think it's like a very entrepreneurial thing to like be okay with the level of uncertainty that people feel right now. And that has really caused split opinions.

John: Yeah, I don't know, like our mission as a company is to accelerate innovation in the private markets and we want to have a large role to play in that we want to help people to like save time.

We want to help people to have better data and make better data-driven decisions, want to help companies build stronger relationships with their investors, want to help investors make better asset allocation decisions and be more helpful to their portfolio companies.

At the same time, like when other people are doing stuff that helps to accelerate innovation in the private markets, that gets me very excited too.

And that's definitely what's happening right now.

Like the private markets are so much more dynamic, even in the private equity world, we're seeing this, you know, very, very clearly where suddenly, all of a sudden, all these private equity firms are saying, "Hey, like we can make all of these companies much, much more fast and adaptable and profitable by leveraging all of this new technology."

The idea of like a tech enabled rollup has exists for a long time, but suddenly now the type of tools that are available are completely different.

So I'm really, really excited about it. And I do think that the, you know, like on the vibe coding side of things, I hear you, I do think that this is one of those areas like kind of with document parsing where the combination of human and machine is unbelievably powerful.

You need to be able to understand what's going on in your code base, at least now, probably forever.

But if you can take an amazing engineer and speed them up by 20%, 30%, even 10%, that could be worth tens of thousands of dollars a year or more in terms of like value that's created for customers. And I think that's just like an unbelievable shift that we're going through right now.

Christine: Hell yeah. I think it's super exciting.

Isaac: I agree. And you know, anytime when you're dealing with engineers or really any people on a project, right?

Every person you add to the project incrementally reduces the efficiency or the productivity of everyone else in the project.

Like very, very, very known phenomena, which means that if you can add kind of a synthetic person, right?

Just by having an operational or efficiency increase for each person on the team existing, like that's worth an exponential amount.

John: Yeah, exactly. What is it is? Is it Mythical Man-Month or something like that? I'm trying to remember.

Isaac: Yeah, big fan of Mythical Man-Month.

John: Yeah, and you don't have to really worry about that in this case.

The issue as I understand it is typically like, if you have two people, there's one kind of communication link, but then you have three people and suddenly there's like, it's one of those problems that gets worse and worse and worse as you scale.

Isaac: Yeah.

John: Whereas if you have a bunch of technology that's helping to automate these things, that's kind of the glue that sits between all of this, it doesn't create additional kind of communication links that need to be established.

Isaac: Yeah, and I think the, what is it? Like N factorial minus or N minus one factorial, something like that?

John: My math is rusty.

Isaac: Well cool. I think this has been a great conversation and you know, I think we're about of time, but we do a little section called picks where we just bring something fun to the table that we're excited about, that we've encountered in our lives. Spang, do you have a pick this time?

Christine: Yeah, so my pick today is, it's funny, we actually talked about this on a different episode a little bit.

There's this guy Bryan Johnson, who does all this crazy health stuff. And he recently launched, or semi-recently launched a brand of olive oil, it's called it Snake Oil.

And I bought some of this, it's a little pricey. It's like a liter bottle and it's cost $35, which feels expensive, but this is seriously the best olive oil I have ever consumed in my entire life.

It tastes amazing. It's like super like, kind of like peppery and like just like really rich and smooth. Put it on bread, I put it on salad.

He says like, you know, it'll like help protect your body against, like, the damage from eating is tough, but I actually just think it just tastes amazing.

And I've been using it a lot actually because I feel like I'm always hungry and it's like a great way to just like add some extra calories to meals.

Anyway, you should check it out. Snake Oil.

John: That sounds great. Yeah, I'm so curious, like what makes it tastes, you know, different and-

Christine: Yeah, I think part of it is just because, well one, normal olive oil a lot of the time is like cut with like, other types of oils.

It's like very diluted at just in general because like people look for olive oil as like a brand and like a lot of people put like canola oil or whatever, you know, other things in it that's not the same.

And then I think another part is the processing. You know, you want to like cold press and then you don't want it to be like stored for really long periods of time.

Like, I bought a couple of these bottles just like a month ago and they expired like in like four months.

So they probably only keep them in storage for like, up to like a year because you want to have them end up, you know, being consumed within a season of them being harvested.

John: I like that.

Christine: Good stuff.

John: I'll have to check that out I'm mostly Greek and Italian, so there's a lot of-

Christine: Ah, you got to try this olive oil.

John: So I got to check this out. We go through a lot of olive oil in our house.

We actually have an olive oil that comes in... One of those like, I forget what the brand is, but it has like a squeeze nozzle on the top so it's like really easy to kind of drizzle on some things.

Christine: Oh, I had to like buy some of the caps just like on Amazon because you know, I just keep a bottle on my table.

John: Okay, I was going to do a boring.. . I was going to do a cool like AI-focused pick, but I'm actually going to do another food pick also 'cause you've inspired me.

So we have one of our big customers, like a big Silicon Valley-based VC firm.

We do sort of an annual like offsite day with them where like, you know, probably 10 or 15 people who get together, our team, their team, we sit down, we talk about how things are going with the partnership and product feedback, you know, where things are going next.

Kind of like open forum, talk about vision. Super fun. We always do a white elephant gift exchange.

And this year in the white elephant gift exchange, I won a smash burger, like cast iron smash burger thing. I don't know if you guys have seen these.

Basically, you like make a hamburger, you put it on like some sort of griddle and then you smash it down and leave this really heavy cast iron thing on top of it and it presses it down and makes it like really, really crispy around the edges.

Christine: I've never really understood what the process was for making smash burgers or how it was anything other than just a really thin burger. And now I understand it actually is smashed.

John: Strong recommend. it's super easy to make at home, tastes amazing.

And it's also very satisfying to have this heavy kind of cast iron thing that you're like leaving on top of the burger.

Something about it just feels great. So I'm a big fan.

Christine: That does sound fun. What about you, Isaac?

Isaac: I don't have a food related this time. I actually found a fun little app. It's called Time Portal. And if you guys know what GeoGuessr is?

John: Yeah, I know GeoGuessr.

Isaac: GeoGuessr is awesome. This is like GeoGuessr but they're historical events so they show you some, you know, they're really cool.

They're like AI-generated quick videos of the historical period or event they're showcasing and then you have to guess the location that happened in the world, but also the year.

And it's just really fun and it's kind of fun, you know, one to have the events that you recognize if you're a history buff.

And then two, also I was surprising some of them are like, you know, different parts of the world and like how industrially far they've gotten along like different periods in like BC and whatnot.

You know, you get some really good surprises.

John: I was going to say it's in a lot of ancient history or like what's the mix is, or is it all time periods?

Isaac: It's all the way from 5,000 BCE to, you know, I think the last one I got was like 1950 or something so far.

John: That sounds cool.

Isaac: It's pretty cool.

Christine: Cool. Well, I think that brings us to the end of our show today.

Thanks again for joining us, John, and sharing all of your great stories and insights. And until next time,

John: Thanks for having me. Appreciate it.