Ep. #4, Building Affinity: From College Dropout to SaaS Leader with Ray Zhou

Christine Spang: Welcome to the show, everybody. Today we have with us as our guest, Ray Zhou.

Ray is the co-founder of Affinity and has been building Affinity for the past 10 years.

We're really excited to have Ray on the show to share some lessons from the journey and talk about building.

Welcome to the show, Ray.

Ray Zhou: Thanks for having me, Christine.

Christine: So you've been building this company called Affinity. your website's Affinity.co, and you really started Affinity basically right out of college, yeah?

Ray: Yeah, about a decade ago, I think I finished two years of undergrad studying computer science and then I guess, a combo serendipity and intention happened and then, you know, here we are.

Christine: Nice, did you actually like drop out in the middle of your degree?

Ray: I did, yeah, yeah.

So I did computer science at Stanford where I met my co-founders, I guess, and they have like a very lenient policy generally, I was like personally surprised, but also not really just given how many companies have come out of Stanford, but they may it both very easy and very risk-free to do that.

So I guess technically, I'm a dropout, yeah.

Christine: That's deep entrepreneur cred to be a dropout. Sometimes I'm jealous actually because I finished my degree and I don't have that cred.

Were you attracted to Stanford because of the strong culture of entrepreneurship?

Ray: Yeah, that's a good question. It's interesting. Before I started college, I was choosing between Stanford or Wharton of all places.

And I don't know, there's some points in your life, maybe your career where it's very clear that you're at a fork in the road.

Christine: Oh man, you almost went to B School.

Ray: Exactly, yeah. You know, I was a point where I was like, man, like either of these two choices could lead to a very different version of Ray.

But no, I think like the entrepreneurial gene was like always there. I grew up in a family of founders. Actually, I came to the U.S. in the first place because my dad who used to work in academia, he was like a physics professor in Singapore.

And around the year 2000, back when like the dot-com boom was happening, he was like, "I'm going to go out, strike it and try to build a company out in Silicon Valley."

So there was a bit of that entrepreneurial gene. I think there also was an element of just like being around the area. Like I'm kind of a Bay Area lifer. I spent my whole life within a small perimeter around San Francisco.

The year that I graduated high school, I think Peter Thiel was like teaching his CS-183 startups class on campus and that's when I was like trying to figure out like, hey, where do I choose to go to school? And did some trips out there and like, you know, listened to, you know, that class.

And I was just like, wow, I've been standing on the edge of this like very interesting ecosystem of like founders and you know, builders, entrepreneurs doing things. And I never really took like a formal step in since I was like so laser-focused on just the question of like, how do you get into college in the first place?

I don't know, yeah, there's something that really, you know, stood out to me about it. Before college, I spent a lot of time working in like a wet lab, you know, biology lab, a wet lab. It's very different from building software.

The hard thing about doing bioscience in general is, it's very slow. Typically, you have to always like wait for permission to do things. Like you'd be like, "Okay, I'm going to like run my experiment," and wait for two hours, get my term to the PCR mission, you realize you screwed something up in the middle of it and have to restart. And when I started programming and writing code and everything like that, I was like, "Wow, this is amazing." I don't have to wait for any permission to do something like that.

If you have an idea, you just immediately build it, right? And that coincided with like, you know, Peter Thiel class was happening.

I think I probably watched the Social Network like for the first time like that-

Christine: Peter Thiel actually taught at Stanford?

Ray: He did, yeah. I mean, he has that famous book, "Zero to One."

Christine: "Zero to One." Yeah, yeah, yeah.

Ray: Yeah. So the material for that book actually I think came directly from, there was some like diligent student in the audience who like took verbatim notes on every single word that he had said and then that ended up, you know, and he was like, "Wow, this is really well-documented set of notes."

And then they ended up co-writing the book together.

Christine: That's funny. Are you saying that an AI could have written that book these days, if you just took good enough notes?

Ray: Yeah, yeah, I don't see why not. Back then, people were a little bit less, I guess, bullish on AI than they are now.

And then there was also like an element of like continuing to be in like, you know, the Silicon Valley sphere. Like I just found so much energy, like this eternal optimism that I felt people had here.

And again, it was weird 'cause it was like my backyard, you know, I grew up not that far away from campus.

But those were like the confluence of factors they ended up make making me realize like, wow, this could be interesting.

And yeah, ended up deciding to go down, you know, the path that was very different from like the business school or like, the Wharton path.

Christine: Well, you may have saved your soul with that one.

When you were at Stanford, like what made you make the jump? Did you like just find people that you really gelled with?

Like was it the idea like, you know, it's kind of a big deal to be like, wow, I'm halfway through my college degree and like, wow, I think this is a better idea?

Ray: Yeah.

Isaac Nassimi: And also, did you feel any pressure to find a business to start in the culture?

Ray: No, not at all. Yeah, so it's interesting. So I guess, the founding story of Affinity I think was a combination of both intention and serendipity.

Actually, maybe let me start by very high-level introducing like, you know, what Affinity is. So Affinity, we basically, build relationship management, what we call Relationship Intelligence Software for the financial services industry.

So in particular, when you think about folks like venture capital investors, private equity investors, you know, banks, et cetera, like basically what these people do for a large part is they allocate capital, they make investments in private companies and they manage relationships for a living.

That's essentially, the core of what they do, right? I did not come from a background in financial services, maybe, ironically actually maybe I would've, you know, if I'd gone to university of Pennsylvania instead.

So that's why I say, it was both intent and serendipity. Like the intent came from the fact that one, coming in, you know, the Silicon Valley bug had bit me.

I was like very passionate about building things. I gotten to know my then roommate, now co-founder, Shubham, he's phenomenal.

And we basically bonded in our first year of college together, just building a ton of things. Like it was like every minute that we weren't spending like doing a problem set or studying for a test, we'd just be like hacking on something on the side.

It was kind of weird 'cause like you're trying to finish like writing the code for the assignment so you can hav e more time writing code for the side project. And that was how I got to know him a lot better.

There was a weird inverse correlation with like things that we worked on where the seriousness of the idea was inversely correlated with the success. So the less serious the idea was, the more successful it seemed to be.

Like one thing that we talked about was like back in the day, our most like viral app that we built was this predecessor to TikTok, but it was only good for making one kind of video and that was it.

And it had this like brief moment of fame on the internet where it went viral and we're like, "Yeah, that's cool," but that can never be a company obviously, right?

So we were definitely building things a lot. We also weren't interested, I would say, in like just being founders for the sake of being founders.

Actually one thing that I personally connected a lot with him just on a core values perspective and this guy like, you know, like came from the other side of the world. He like grew up in India and came over here.

But I think one thing that drew us together was realizing we were both like motivated, maybe not just by our craft, but by our cause. It's kind of interesting dichotomy for thinking about these things.

Like you could be driven to say like, "My singular goal is to be the best product leader, the best whatever."

And you know, that's what I'm driven to do. And frankly, I think a lot of like the K-12 education system tries to push you into that.

For us, we ask the question of like, what are problems that are worth working on? You know, they could really spend a lifetime like, you know, just working really hard on that are going to make a really big difference.

And so we had, I think like those elements that kind of maybe led us to have some that like generally founder like Jean where it was likely we're going to build a company.

The serendipity part of it was when we were in college, through total chance we went to this talk that was at the time being done by Joe Lonsdale and Drew Oetting, they were leading a fund, back then, it was called, Formation 8, now it's called, APC.

It's an extraordinary firm. You know, Christine, I mean, they were big investors and Nylas.

Christine: Yeah, Formation 8 also invested in Nylas.

Ray: Yeah. So what happened was, you know, Joe and Drew, they came to campus, they gave a talk and back then, they were talking about this concept called, the Smart Enterprise, which was at the time-

Christine: I remember the Smart Enterprise.

Ray: Yeah, yeah.

Christine: And Joe wrote that article.

Ray: He did, yeah. It was basically talking about vertical SaaS. That was before like vertical SaaS was a thing.

So they coined the term technically, I guess like the name didn't stick, but the concept still stuck. And we thought that was fascinating.

So like we got to know them afterward and you know, one conversation led to another, I think somehow or another that they had heard that, you know, we had built a lot of things, you know, and I don't know, somehow word probably got to them at some point.

And so we got to know them a lot better and they ended up making like the introductions initially that ended up snowballing into Affinity, the idea.

So there was one summer where it was like one intro turned into like 10, I mean I'm talking, you know, in the summer before we found Affinity to probably hundreds of people across financial professional services where all they did was build relationships for a living.

And that led slowly to the realization of, wow, there's like a really interesting company to be built here. There is a way to do things that's probably 10 times better, even though it requires a bit of a counterintuitive leap of faith.

But that's how the company got started, you know, so you could say like, if we didn't maybe go to that talk, you know, like maybe the company would have been started, I don't know.

Who knows? But yeah, definitely a combination of both intent and our luck.

Christine: Yeah, it sounds like there was like, you know, one part intentionality in that like, you know, you and your co-founder were like, you know, had decided that like you love building things and you want to find something that's worth building.

So there was like a little bit of an intentional element of looking for the right thing.

But it also sounds like you kind of like followed your curiosity a little bit, you know, like talking to people, building things, going from one thing to the next and having a lot of conversations and like iterating until you sort of found something that seemed like it was something that would actually be a company.

Ray: Yeah. I actually think that's like really prescient. I think there's like a good lesson buried in there.

We probably made a million mistakes as we were starting off. One of the things we did do right was we were really obsessed with focusing on the problem and not on the solution.

I think like a lot of founders get it wrong or they kind of mix it up the other way around. They're like, "Hey, wouldn't it be amazing if X happened?"

And then they work backwards from there to figure out like what is the thing they could direct that like X towards solving.

And that tends to be a lot less... It can still work, you know, like every company is a hypothesis. There's no mandate as to whether a hypothesis will be right or wrong.

You can follow all the best practices and still get to an incorrect hypothesis or you can break all the rules and still get to the right hypothesis, right?

But one thing that correlates well, I think with like, you know, landing on a good idea is like you obsess with the problem.

And so, you know, again, we were outsiders in the industries that we talked to, we were like, we don't know how like A, B, C makes money or like, what is a limited partner, how do they raise money?

You know, none of those ideas. It was through like just talking to folks and like really obsessively figuring out what are they currently doing. You know, like how is it working for them. If it's not working, why is that not working?

We just kept pulling that thread and that curiosity I think ended up leading us to the insights that inspired the company.

Christine: Totally. Yeah, I think people really underestimate how much just like the foundation for like doing something is actually interest, it's like being like deeply interested in what are the problems out there and how things work. And then that's like the driving motivation in moving forward.

Ray: 100%.

Isaac: Ray, I'd love to know a little bit more about kind of the journey at Affinity and, you know, growing it, growing pains, maybe anything you ran into during scaling and I can't remember who talks about it, but like the threes and tens every time you triple your business, basically everything breaks.

I'd love to hear about some of those breakages and you know, how you're able to get past them.

Ray: Yeah, yeah. Gosh, I feel like there's a million stories that we kind of go through there.

Yeah, I don't think this is anything I guess, super novel, but you know, I think of like the company in a couple of stages.

So in the initial stage, there's this obsession with like just finding product market fit. A lot of that is like, you know, as a founder, you're doing everything hands on yourself.

So I guess, you asked about the scale phase, but like, basically I think the impetus of that phase is while the early team is doing everything personally, trying to figure out how to sell customers, service them, what's the right, like you know, initial feature set, what's the initial veg. It's basically, setting the foundation for the scale phase.

So for me and my co-founder, when we were like involved in every single customer onboarding, we were obsessively spending time with our customers in the very beginning.

The learnings from that ended up becoming the playbook that we then had to figure out like, okay, if you can do it with like two people or let's say two founders, you know, who have a lot of context and like the early business team, like how do you scale that to 10 people, 20 people, 30 people, et cetera.

So a lot of ways in which, you know, we had ended up having to figure that out. It was everything from like creating training programs, hiring the right folks.

I would say, the experience of it all was, you know, as a first time founder, you know, we didn't have the prior experience, like the y-intercepts, I guess, you know, compared to different from slope, but like the linear step of having seen like this happen many times.

And so it personally felt like, you know, every single new thing that we were coming up against, it was a fundamentally new kind of challenge.

So yeah, I guess, I can go into any number of different rabbit holes here, but like how do we scale recruiting, how do we take that and build a machine and engine out of it?

How do we scale, you know, the way that we onboarded customers build a machine out of that.

Probably, the common pattern was for all these things, we had to really go from a roll up your sleeves and do-it-yourself kind of mentality to essentially, how do you like orchestrate a machine and how do you like, make sure you have the right analytics in place to measure like, you know, every step along the way.

Because many things for example, whether you're thinking about, you know, sales or fundraising or recruiting can be thought of as an example as a funnel, right?

And so there's common like, you know, techniques for how do you instrument that and make sure you're measuring all the right inputs and outputs and whatnot.

Isaac: I think it makes a lot of sense. And what you're talking about is over the time span of the last 10 years, and I think people forget how much, even over 10 years things have changed, right?

Even doing simple things, today things that are simple, like AB testing was a little hard then, right? I think Optimizely was like just coming online, right?

Analytics, HR, everything. The way we have a million point solutions that we stitched together into this kind of tech stack of the business operationally and also like product-wise, it just wasn't there.

Ray: Yeah, it's interesting actually. I think on one hand, like, yeah, like some of those tools were super, super critical.

For example, in like kind of the finding product market fit phase for our company.

Like I don't think we would've gotten there, if it weren't for this whole slew of analytics tools that helped us like really hone in on, you know, what were like the points that would hook a customer and like turn them into like very loyal customers like going forward.

That was really critical. I think funnily enough though, one interesting thing about what I've seen shift in how like, you know, buyers have changed their mindsets is, we have too many point solutions, when there's a million ways to like, you know, do a certain thing.

Like I think generally as a shift, this has certainly been true in private capital where we operate, what we've seen is like people want to consolidate more.

And so this increasing expectation that all the things you do need to do to scale your business, whoever, you know, you happen to be serving instead of buying like, you know, a bunch of different tools, like want to buy a single platform tool and the expectation for each tool is that it should have the functionality of all the other tools that are adjacent to it.

That's happened both like vertically, you know, for us, for example, in financial services. But it's also happened I think across a bunch of other horizontal like spaces as well.

Isaac: What do you think the termination point of that is? Like in 10 years, are there just like three tools that do absolutely everything?

Ray: In 10 years, it's tough actually.

Yeah, because I feel you're kind of making the assumption there that all the prior assumptions that we had had are going to continue, which is, you know, SaaS continues to be just SaaS as it was in the last like decade.

But I actually think a lot of this like stuff that's happening currently in AI is going to fundamentally change a lot of that. So we're kind of standing at like the cusp of this next inflection, this like next disruption that will kind of rejigger the way that people think about things altogether.

If you were to extrapolate though, that's what I believe, yeah. You know, so on one hand, you know, you see that in the vertical world, right?

Like we were talking about earlier about Smart Enterprise, Vertical SaaS, I mean, at this point, there's basically become this like Vertical SaaS, not just expectation, but playbook for consolidation of all the tools and technology that, you know, a vertical happens to use.

I think Bessemer did a really good set of articles on this, but they call it kind of like the layer cake strategy where you kind of start with one module.

Ideally, it's a system of record module where you're kind of like the database that stores like the most important data and then over time, you know, all the other things adjacent to it that, you know, the customer that you serve needs, you end up migrating into the space.

So you can consolidate all that into a single tool. So the number of items in the P&L for any of these like companies like just gets smaller and smaller over time.

And very often, like those tools have this like one plus one equals three kind of effect where like, it's better to have like two things living in one tool that are like natively and directly integrated versus like having to log into a bunch of different systems and some of the synergies don't really quite come out.

That's certainly been, you know, the case for Affinity. That's essentially one of the guiding North stars for our current product strategy.

But it's also true for horizontal things. You know, like I don't know, you know, for instance, maybe five years ago, I mean, close to the start of the company for instance, like we used Gong in order to record our customer and sales conversations and analyze them post facto.

We would use a separate tool to do, you know, our product analytics. We used a separate tool, I think, it was called, Fullstory at the time, to like measure and record session replays for our user, see what they're doing in the app.

Now, all of these tools, they're like all consolidating, they're all competing with each other. Like Salesloft does like all the recording stuff, Gong does more of the enablement stuff.

Like it's all like converged into a much more direct fight. So it's very much like a industry wide trend.

Isaac: I absolutely agree. I mean, at Nylas, you know, we're kind of coming to the conclusion that everything is slowly becoming a CRM, right?

Even things that you wouldn't have considered a CRM before, ChartMogul's making a CRM. Gong is building a CRM.

And I mean, even them, because you know, these types of applications are our customers. Like we're seeing them, you know, have these same exact predilections where they want to try and bundle, right?

They want to buy email and calendar from the same provider. And you know, now that we're providing meeting recordings as well, they want to buy those from the same provider.

And I mean, we're able to even offer, you know, these little tiny like product delights as well because they're in the same place because we don't have to talk to someone else.

So yeah, I mean, I really agree. But rewinding back a little bit, you talked about how the thing that we're on the precipice of the whole AI world could really, really change and shake up what we consider the SaaS market today.

There's a really big probability cloud there. I think it could go a lot of different directions. What do you think that actually looks like, if you had to choose one path?

Ray: The first thing I'll say, I want to comment on something you said earlier because I think it'll be relevant in the moment.

You talked about like a lot of these companies are essentially becoming CRMs. It's not surprising to me, like in the world of when you think about like, what is the stickiest flavor of a SaaS product?

They tend to be like these very entrenched systems of record, right? Like when you have all of your data sitting inside an Affinity or sitting inside a NetSuite or something like that, it's just a really large pane to like evict that data and move it onto a new platform.

And so that's what makes these like things so sticky. That's why their net rotation is so high.

Those are probably the products, for instance, in the last few years when some of the venture bubble like had popped and interest rates started rising and everything where that kind of revealed who were like the must-haves versus the nice-haves.

And it was probably like the systems or record that were the must-haves.

That said, one thing I will say generally about like CRMs is no one enters data for the sake of entering data. It's not like there's some mandate of like, oh, you know, you must enter the data or else something terrible happens. It's more you're putting data into, let's say, a CRM because you're going to get value out of it later down the line. Who knows when, right? But it's kind of like an investment you're paying upfront because it's going to help you in some future state when you have to go back and check the data or get some analytics from it or something like that.

So the AI thing is a really interesting question.

Specifically, what's hard, I think today, I think this is both true from a founder builder perspective and from an investor perspective 'cause there are kind of two sides of the same coin, is it's very hard to think about like, where does the future of the market go and like how do you choose good problems?

Like how do you think about problem selection, right?

In a weird way, you know, like starting a company or thinking about what to invest in or build today is like way harder than it was 10 years ago.

'Cause 10 years ago, like the ground underneath our feet wasn't shifting nearly as fast. It was a lot more stable.

Whereas right now, and I don't know, maybe I'm just, you know, overly influenced being in Silicon Valley, but like, what's the timeline for generally intelligent, super intelligent AI?

Like, we don't know, maybe that's like three years away, five years away, 10 years away, 50 years away. Or maybe, it's never going to happen in our lifetimes, like the three of us will be dead by the time it happens, right?

The thing that's hard and you talked about this probability cloud idea, Isaac, the root of it is that we're not sure. The uncertainty is what makes it difficult, right?

Now, how do you think about prom selection when there's all this uncertainty? Here's my theory.

My theory is a rigorous framework is a framework that's agnostic or mostly agnostic, okay, to if, you know, let's say, we get to like very powerful AI or human level AI, whatever that that timeline happens to be.

So put another way, if it happens very soon, you want to have a plan, I think building up to it, right? If it happens, let's say, in seven years, that's fine.

You know, we still had a plan for it, if it was to come, right? But like we're not working on something that's going to be obviated.

The problem itself will be obviated by, you know, AGI getting here or, you know, we have the right strategy where as it gets closer, we know how we're going to pivot and adapt what we're working on in order to better adjust to serving, you know, a world where that ends up being the case.

So then that leads to this interesting question. Like what kinds of problems, whether you're building a company today or you're building, you know, like you know, starting a company just, you know, fresh from scratch.

What kinds of problems tend to be more AI-proof? And my theory there is that, the closer the problem that your company is solving or that your product solving is to the core fundamental need of a business, the more sort of AI-proof it is, the farther away you get, the more levels of abstraction you are moved away, the more risky it is, right?

So I'll give you an example. If you're building a tool that solves a problem for someone who reports to someone within a department, you know, like reports to the CEO of the company, right?

That's a little bit trickier, it's riskier, right? Because you don't know, for example, like, you know, in this entire chain, is that which jobs are going to be disrupted and like what's going to end up, you know, remaining at the end of the day, like after like we get closer and closer to like powerful AI, right?

Whereas on the contrary, if you build, you know, a system or an agent that does the job of the function itself, that's much closer to the core fundamental need of the company, right?

There's a few core fundamental needs that every company needs. You have to build the product, you have to sell the product, you have to market the product, you have to keep the books.

And then there's all these needs that are more specific to different industries. You know, if you're running a venture fund business, for example, like you need to have a fund administration service, like that's just a core need of like running a business like that, right?

So the closer you are to delivering that directly, the more sort of like protected you are because you can build something that will benefit from, you know, AI versus you're always like fearing that it's right around the corner and it's going to remove the job of the person that you're trying to serve.

So going all the way back, how does that apply to, you know, what's next for business SaaS, right?

I predict over time, like, you know, SaaS basically what's incumbent in a lot of it, I know you guys, I don't think charge this way, but like a lot of SaaS like prices per seat and when you price per seat, you're basically assuming that the job of the person that's buying the seat still exists, right?

I think as we get more and more advanced with AI over time, if we're making very quick progression, there's going to be this convergent pressure where every company that builds a tool that is serving a human within like a function or a role is going to start feeling this pressure to pivot toward building, let's say, the agent or you know, the automated system that does the job of that human.

And it's really, you know, relevant for example, in the CRM case for instance, because as I mentioned earlier, like no one maintains database for the sake of maintaining a database, right?

If for example, you know, we had a bunch of, I don't know, like there were a bunch of VC analyst agents and those agents were doing the job of like the analysts end to end.

I'm not saying what timeline that would happen or if it would ever happen, but if that were the case, then the memory of those systems, those agents is the CRM, right?

Like all the data you put inside CRM is essentially, the memory of those agents. So you don't need to have like a separate human CRM necessarily anymore.

And so that's where I think things are really going. We're going to kind of shift basically from, you know, being a very seat-focused to more being like a value-driven, consumption-driven, that's going to be a new model.

I think any company today, if you're kind of building, let's say, a tool for an individual, even if that's like the right thing to be doing now, should be thinking about like what is the path as we get closer and closer as these AI systems get more powerful to eventually be able to mold and adapt and pivot to eventually have the edge in terms of being like the leading player that ends up just doing the work of that department or that person instead of building something for the person themselves.

Anyway, that's a longer term, you know, event horizon. So again, it may happen in 50 years, it may, you know, happen in the next like 10 years, I'm not sure, right?

But I think it's good to kind of have a map of like, what's the path from here until there? So you kind of adapt as you go.

Isaac: I really agree with that. I mean, I think you said a lot of really excellent stuff there.

The way I almost think about this is, okay, let's say, an AI agent, you know, to use kind of a buzzy term, it's ostensibly able to act the way a human would, but under you know, specific set of conditions, it would make about the same decision as a human, do about the same things, take about the same actions.

And the way I see it is just drastically, lowering the cost of not just getting work done, but also enforcing that that work gets done.

Right, if you talk to anyone who's in the management of the sales world, right, biggest complaint is, I can't get my reps to update the dang CRM, right?

Ray: Yep.

Isaac: You know, they just won't do it. And then I'm not able to forecast properly.

You know, if you can make people follow the rules around that a 100% of the time because it's automatically happening and better yet, it's not taking an hour out of their day, it's actually just taking zero seconds out of their day for a 2 cent agent call.

Well, I mean, even if we don't get to the point where these things are happening automatically very soon and you know, these agents are acting proactively based off the CRMs being in the memory, I think that even still just change s, like you said, what you can charge the way the unit economics work of these businesses.

Ray: Totally, yeah, yeah. There's obviously, a lot in the middle to muddle through in order to get to that like longer term thing.

Actually, funnily enough, like one thing I also do believe is, the pure cost value proposition, at least short-term is an interesting one.

It's a bit of a tricky one with agents because the problem is, this is now near-term, so thinking about like building things in the next year or two years or something--

If the only thing that you're saving is cost, the tough part of the trade off is basically for the customer buying that AI tool, they're always going to be wondering, okay, is the cost worth like the trust I have to put, you know, into this new tool, right?

Whereas, let's say, your AI product or AI feature or AI tool is able to do something that a human was doing much better in some way, let's say, they're able to do much more of it at scale because it's all automated or they're able to deliver a much higher quality product, it's much more justifiable, right?

It's not purely than, you know, like a cost versus trust trade off, but you can actually see a lot more of the value centric upside.

So my prediction is that, I think a lot of like those kinds of value propositions feature sets will start to take off first, but it will take some time for us to eventually get to the point where we've kind of closed that gap and you know, now the agents are doing all the work.

Isaac: I remember using the initial OpenAI APIs like before they actually came out with ChatGPT, you know, I think 2022, maybe late 2021.

And I mean, it was a miracle, but at the same time, it was so stupid that there was no way you could hand the keys to it and just be like, "All right, well, we're going to let it make a decision for us."

It's definitely covered a lot of that gap pretty fast, but also I think there's types of decisions in the business, like a 98% accuracy rate is awesome. Right?

Ray: Right.

Isaac: And then there's types of decisions in the business where you need like five nines or six nines.

In regards to it doing things better than a human, do you see any areas where these cannot perform humans today?

Ray: Yeah. I'll use maybe a few, you know, Affinity examples. We've launched a number of like AI-powered things.

Two of the ones that have been most well-received, you know, we have one where we basically built a very sort of private capital oriented bot for note-taking.

So it's a bot, you attach it to a conversation, a Zoom link or anything like that. We've poured all this like context about like who's an investor, you know, an investment type of conversation.

So that vertical context into it and it just does a far better job than a human, you know, having to like scramble and take notes and everything would do themselves.

Especially if like, given the constraint of time, you know, the fact that like, you know, you don't have all the time to like reformat the notes you've taken and so forth.

And so it does everything end to end. You know, like listen to conversations, take the notes, like attach it to the right profiles within Affinity.

That's an example of something that you had just mentioned, like, I don't know, maybe not the five nines, but you know, like the 98% accuracy where by default, you know, like an action like that or a task like that, no one expects to have like perfect accuracy.

Even today, like, you know, when the humans are doing it, it's like, you know, there's not going to be a perfect note-taking something that someone had said is going to be missed, right?

That's what makes that kind of a use case so perfect for something like this.

So I think a lot of those kinds of like use cases initially, that's where sort of the applications will, you know, the kind of as it's finding its way through the crevices of what is sticking, what doesn't, those are the nearer term things that, you know, we're going to see it, you know, sticking.

Isaac: That's fantastic. So overall, you know, I think people seem to have different levels of sentiment, you know, somewhere between doom and gloom and like a Star Trek wonderful bright future.

You seem to be a little bit more on the positive side, it seems.

Ray: Yeah, my philosophy on this actually is, it's kind of like weirdly tautological. It's like, you know, I don't think too much on the gloom and doom side because it's not a useful thing to think about.

I will put another way, if you believe the doom and gloom thing and you could say, "Well, I'm not going to work on anything then because like, you know, whatever we build, like it's going to be wiped away because I don't know, control will shift over to the AI in the future or something like that," right?

But my theory is, if you believe there's any non-zero probability that like a doom and gloom situation won't happen, and let's say, you're not working on AI safety because that's what would actually minimize the doom and gloom situation, right?

Then you should assume that, you know, like as humans are going to be able to keep our sovereignty and build things and your own property and then you can kind of project, you know, into the future and think about like, all right, what does that world look like and what is the path as we get closer and closer to the world look like?

And then what are the actions that I can be doing today or should be preparing in order to get closer to that world to be like a winning company in that world versus I guess, a company that loses or, you know, isn't benefiting from like, you know, the advantages that are created by that kind of technology.

So I think, you know, founders today, when they're thinking about choosing problems, they should think about that seriously.

You know, they should say like, "Hey Christine, you know, this like building companies a five to 10 year endeavor. Like hard problems are hard for a reason. It takes a long time for them to mature," right?

Christine: Sometimes more.

Ray: Yeah, absolutely.

So if I'm going to say like, I'm going to spend 10 years of my life, you know, going and building something, you know, like probably want to be generally thoughtful and think about what are things that are worth working on such that even if, you know, this AI takeoff is very rapid, they can benefit from that transformation happening in the next like decade.

Versus we're always like fearing or like hoping that, you know, the takeoff doesn't happen or something like that.

And it's same with how I think like, you know, leaders should think about building their companies too, right?

Like it's probably worthwhile for, you know, CEOs to think about, okay, like as we get closer to that world, how does that potentially disrupt, you know, the kind of problem that we're trying to solve?

I don't know, maybe I'm building CPA software, even if that doesn't mean that I'm, you know, starting to work on something agentic or whatever today, right?

At the very least, I should be able to thread the needle to say like, from where we are today, how does that evolution, you know, as we co-evolve with these systems happen, so that if you know, now around the corner, I don't know, the VC agent, you know, is now possible that we're well-positioned and we're ready to adapt to go build that product.

So those are like, I think the practical takeaways from it. I think it is only useful to be optimistic and to at least plan for it. And if it doesn't happen, that's fine.

You know, you keep executing your current strategy and still build something very valuable.

Christine: Yeah, that's super interesting. I love that. Do do you read much like stoic philosophy, Ray?

Ray: Actually, no.

Christine: It kind of has the same takeaways as you're talking about in terms of like, if it's not something that is within your control, it's not really worth spending time on or thinking of, you know, it's just like wasted cycles and energy.

Ray: Yeah, that resonates.

Christine: Same with like the AI apocalypse. Like if the robots take over, they're like, "Wow, whoa, should we just give up?"

Something that I definitely like to tell people if they feel very demoralized, you know, "Oh, should we just hang up our shoes and like stop trying," you know?

Ray: Yeah, like what good comes out of that, right? If you think it may not happen, then you should act as if it's not going to happen.

Christine: Yeah, no, I love that. I definitely see a lot of people struggling with this.

Like, you know, if things are changing so fast or like, you know, if this is like the potential end state that we're all sort of imagining like how do I like get from point A to point B or like how do I evolve from where I am today to like make sure that I'm like keeping pace with that.

Like if we're thinking about like the organizational stuff, like, I think one thing that I'm hearing from this is that like, because of this consolidation pressure, like people that are building software, that those software platforms need to do a lot of stuff and they need to do more things over time.

Like how do people keep up with that and get there? Like, is AI going to make us more productive? Should we lean on that?

Ray: Yeah.

Christine: Do companies get bigger? Do they get smaller? Like where are going with this?

Ray: I almost feel like, yeah, you're talking about two independent things, like the AI question is maybe, a bit of a separate question.

Just the first thing you talked about like, you know, just generally, how B2B SaaS buyers, how has the market shifted, right?

Like you need a couple of things like, you know, they expect like more consolidation, a few applications to do more things.

That's not the only one, I would say. Like for instance, there's a very high sort of a bar now, especially after the last few years when we kind of came out of the ZIRP era where people don't want to buy nice tabs.

Like, you know, they're really sort of focused on like what is a must-have, like what is something that has to be on the P&L and even further they try to consolidate there.

There's other, you know, shifts too. I would say, you know, that are probably relevant to you guys as well.

Like one is, once upon a time like buying B2B SaaS, there wasn't an expectation for like great user experience, great design, great onboarding, right?

Like once upon a time, like enterprise software, the impression it gave in people's heads was it was like the clunky old Oracle dashboard or what have you.

And that's okay, that's what, you know, SaaS is supposed to be. It's supposed to be like not very usable.

That's totally shifted, you know, like now the same kind of consumer grade expectations around like design and onboarding and so forth like that's now become inculcated into the way that people think about buying SaaS.

And then the last thing is, like they expect these tools to do all the work or more of the work for you.

For example, the original product insight that we had with Affinity, the idea that you could suck up all this waste to data from like emails and calendar streams and relationships and use that to just populate and automate as much of like the human sort of work of using a CRM as possible.

That used to be kind of a radical thing. It's like, you know, hey, like this is really sensitive data. Now, it's an expectation.

Now, if you talk with like for example, most let's say, like investment firms, who would say, "If you're not doing that, that'd be crazy," right?

And they kind of expect that kind of automation from all the tools that they use.

So I would say, those are kind of like, if I were to think about some of like the product or buyer expectation trends over the last decade, those are some of like the core foundational pieces.

It's almost become like a industry-wide or a sector wide like gold standard.

And then beyond that, in terms of evolving, I don't think there's any substitute just generally for like being maniacally obsessed with and close to the customer and just talking to them constantly.

Because they don't even know, like, you know, what their reactions to a lot of these like changes are, you know, like how does, for instance, an investor that uses Affinity versus an engineer that uses like Nylas think about like, you know, the AI question, how it plays into whatever they're building, whatever they're trying to use, they're probably very different, right?

So there's no substitute for being so close to the customer that you're able to hear their psychology and how they're react and all these things on a constant basis.

Isaac: Great, so I think we're about at time, but first, we have a section we call, picks, which is, everyone just brings something that maybe they got exposed to over the last week or two that you're excited about.

Doesn't have to be some piece of SaaS or something, it can be a movie, it can be, you know, a new vacuum cleaner you got that you're just obsessed with.

Spang, do you have a pick for this week?

Christine: So I actually brought my pick today and it's a magazine called, Summit Journal.

If you guys know me and Ray likes to point out that I do crazy stuff around climbing. I like climbing a lot.

But the cool thing about this magazine is, you know, it's like a large format print magazine with like long format stories and beautiful art and beautiful photographs.

It's just like a nice way to relax on the couch at the end of the day and like soak up some adventure.

So something that I've been enjoying recently as a way to, you know, read for 15 minutes at the end of the day.

Isaac: That's awesome. I'm jealous that you've a physical pick you brought with you.

My pick is, I got a new coffee maker and it's like not a high-end coffee maker. It's like a lower mid-end coffee maker. It's this Ninja and it's like a pour over.

But what I'm so excited about with it is it's got all this little clever functionality that you can make like a normal sized cup on the stand and it still does it through a pour over or do kind of like the large carafe or whatever.

And it's just so clever. Like everything about it, there's a little slot for the measuring spoon for the actual coffee grounds and all of that.

And I feel like a sucker 'cause, you know, I've had like a nice espresso machine for the last few years and all this and I'm just so happy with like a hundred dollar coffee maker now.

That's my pick for this week. Not an AI one this time.

Ray, you got something?

Ray: Yeah. Man, I feel ours are so different.

Maybe one thing I'll give a shout out to, I've always loved this lecture series, but I've recently been watching increasingly more of some of the lectures that Andrej Karpathy posts to YouTube.

And I don't know, I think there's something about the craftsmanship and taste of designing, let's say, like a really good learning experience, seeing kind of like craftsmanship you can put into designing, you know, like a product on your phone or something like that, right?

Like his lecture series is just absolutely phenomenal. So I recommend that to like anyone who is like interested in learning more about this kind of thing for all levels.

But it's been really fantastic.

Christine: That guy is awesome. That stuff is really good.

Cool, well, thanks so much for joining, Ray. Awesome insights to share all across, you know, from building a company to where AI is taking us and really appreciate you taking the time to come and join us.

Isaac: Thanks Ray.