Special Season: Inside Summit 2024 Fireside Chat with Kashish Gupta Co-CEO and Co-Founder of Hightouch

Hi everyone, so excited to sit down today to get started on our INSIDE Summit miniseries.

We're so excited to have had our first ever all-day conference event just last week in New York City with over 120 innovators and founders in In the room where it happens. If you want to hear more about our vision behind Inside Summit and some of our learnings of hosting our largest scaled event yet, listen to our opening episode of this miniseries. But what you can expect next is a recap of the conversations that we had in the room from the actual guests themselves. You can expect our opening fireside with KJ Sidbury, the partner over consumer investments at GV, our next-gen GenAI panel, which was curated by the GenAI Collective, our Future of Platforms panel, which was moderated by the associate editor at Forbes, and two more. Claudia, you want to fill our listeners in?

Yeah, absolutely. The future of direct-to-consumer with some incredible brand founders, operators, and agency partners. And then we ended the day with a recap episode, um, our fireside chat with Kashish Gupta, co-CEO of Hightouch, a previous podcast guest. So without further ado, Let's open the door.

I'm loving the connections I'm seeing, um, but your time is so soon for a final vibes check.

3 hours is kind of dangerous, no?

Oh, it's really dangerous. Um, when they told us we had a certain minimum, we said, okay, well, we'll just be there all night.

It'll be a challenge to hit it. And I think some, some of us are, are down for that challenge.

We have some custom cocktails too, so check out the names. Um, my personal favorite is What's Your AUM Martini. So yeah, give us a shout there on those names. But okay, we're gonna get started here. Kashish, welcome up on stage.

Thanks for having me, guys.

It's been about 2 years since we sat down with Kashish Gupta, who, as I mentioned here, is the CEO and co-founder of Hightouch. Um, this was September 2022, and it's really fun to actually get to sit back down for the first time with a podcast guest 2+ years later and kind of go back in on the story. For those who don't know, Hitech is a leader in the data activation and the composable customer data platform. He spends every day helping companies realize the full value of their data and their data teams. He's been working on the start— on startups in general for the past 4 years. You were a part of Y Combinator and you love working on both product and go-to-market. You're a fan of racket sports and karaoke, and you're passionate about using reverse ETL to make data warehouses actionable. I'm literally not sure how those things go together, but I love that that was the bio you guys gave us. And you attended Penn for undergrad where you had an MSc in robotics and a BS in economics from Wharton. We're super excited to dive back into the story of Hightouch and where they've been since fall of 2022 and the future of the modern data stack. So let's open the door.

Welcome back.

Thanks for having me. Excited.

Okay, so now we're physically in the literal room, which is kind of exciting versus our virtual version. Um, you previously shared on our podcast a little bit about your origin story, being born in India and then moving to the States. I think it was Atlanta, Georgia. Um, and how your dad as an entrepreneur had a huge impact on what you thought about what you wanted to be when you grew up. Maybe just ground our audiences and how that did shape your worldview.

Yeah, happy to share. So I moved to the US when I was about 5 years old. And for about, I think it was 14 or 16 years, my dad actually ran a gas station. So him running this gas station in South Atlanta was my first exposure to entrepreneurship and to what it was to be a very high-agency person. He was doing all of his own tax accounting. He was filing for his own incorporation, his own liquor license. He was doing all of the wholesaling and everything by himself. And it showed me what real business is, right? There's margin. There's top line. There's competition. There's price advantage. There's high margin products, low margin products. Turns out you only make $0.10 a gallon on gas. But it turns out you make 50% margin on the convenience store. So all of these things were things that my dad would talk about at a really young age, like maybe when I was 10 or 15. But it wasn't really entrepreneurship as we know it today. It was small business entrepreneurship, but it really gives you the fundamentals and exposes you to those at an early age. I studied a lot of engineering, math, and science throughout college, high school. And the whole time my dad said, dude, you can learn that stuff in school. You need to learn business. Business is something you can't learn in school. It's something that you have to teach yourself. You have to talk to entrepreneurs. You have to learn on the job. And there's no way that you're going to learn in a book. And so that kind of mentality is actually really what pushed me in that direction. I was always looking for what can I actually learn from other people rather than learn by reading. So that was an early mentality. And really, I feel like a lot of people have parents that lived in Silicon Valley, started tech companies, or taught them about computers growing up. Wasn't entirely the case for me, but it gives you like a lot of that business sense.

Still having that opportunity to learn from your parents and their passions and their drive and grit that brought you into having this desire to become an entrepreneur yourself. You've kind of been a chronic entrepreneur. I think in high school there was something about wireless charging business, Mama's cooking, and then of course the first instance of HiTouch, which is what you were initially admitted to Y Combinator for. Madison, just tell us a little bit more about the, the persistence in your own career journey. And if you could double-click into the pivot that became Hightouch, right? Like you've been a chronic founder. How did you get there? Before you start though, can I just ask for the people in the kitchen area, just if you could be respectful and quiet down? Um, that would be amazing. Thank you so much. We're just wrapping up. Thank you. Okay, back to you on the question around persistence in entrepreneurship?

Yeah, so, um, there's probably a lot of entrepreneurs in the room, and something that's probably very similar between entrepreneurs is they're always thinking, why not me? So you see other people around you that are starting companies, you see other CEOs, and you think to yourself, why can't I do a better job than this person? Or what makes me less qualified to actually go after this market or go after this opportunity or to solve this really big problem? And that's actually like a really important mentality. Like that's kind of what drives people to start a company, start another one, pivot again, keep pivoting and get there. So when we first applied for Y Combinator, we were actually a travel company and we were basically ChatGPT for travel bookings. So you would message us and we would give you like a whole itinerary. We would book your travel for you. We'd expense your travel for you. And it was all for B2B over Slack. So it was a Slack bot. You would ping the Slack bot and say, hey, I got to go travel to Seattle for a business meeting, and it would book your whole trip for you and come back to you with like the expense report. This was before LLMs really existed. We were using small language models. It's a very different story than it is today. And we didn't want to be in the travel space. So every single time we looked at the business, we said, this is going to suck. We're not going to make any margin. Anyone that does travel knows they're going to make 1% to 2% margin on all bookings. Which is terrifying because you have to make $1 billion of bookings to have $1 to $2 million of revenue. Right, so just wrapping your head around that, or like $10 to $20 million of revenue, but that's still, it's not great math. So, um, what kept happening though is people, consumers kept saying, we really want this. If you build it, we're gonna use it tomorrow. So you gotta give this to us. And then we kept thinking to ourselves, well, like, people really want this. We're pretty good at like both in software engineering and AI. So like maybe we can actually go win the space. And what's stopping us from beating people like TripActions, which is a big B2C, uh, B2B like travel company? What's stopping us from beating people like Expedia or Concur or like Kayak? And so we just started the thing and we actually showed up in those kinds of companies' board decks saying, hey, maybe we should be using AI for travel. Hey, this is a big competitor of ours. We should go like have an answer to this question, and that's how it all started. Turns out COVID happened a year later, and so we had to completely turn off the travel business and pivot to a completely different space. But we learned a lot, and it kind of gave us that initial mentality of letting customers lead us to where they want us to go rather than us having just a really strong thesis that we're going to build a company off of.

Let's dig a little bit into what HiTouch actually is from a product perspective. So HiTouch, syncs customer data from your data warehouse into the tools that business teams rely on. A lot probably has happened since 2022 when we first chatted. You guys really coined the term reverse ETL. Tell us a little bit about what the Hightouch product has been up to in the past 2 years.

Yeah, absolutely. So we now run the marketing for some of the largest brands in the world. So we run all of PetSmart's loyalty marketing. If you've ever gotten an email or a push notification from PetSmart, That's us. We do it for Spotify, we do it for SoundCloud, we do it for Warner Music. There's a bunch of really big brands I can't name that we run their marketing for. And so we did start from a really simple abstraction. That was the reverse ETL abstraction. And the idea was you give us a SQL query, we'll pipe that SQL query to any of your marketing tools. That way your marketers can operate on the best possible data that you have. That was actually like something that people really needed. It was not an expensive thing. A lot of people needed it. They could all use it for like cheap and very quickly. So they could just log in online and use our app like same day. So it was a great way to start our business and a great way to get off the ground, great way to create a name for ourselves. That's like pretty much what we did for most of 2020 and really helped us get like our first 130 customers. But then what we found out is that when we give folks access to data, they actually want their marketing teams to be able to use that data, not just their data teams. So data teams write SQL. They were able to use reverse ETL. Marketing teams couldn't access it. It didn't really work for them. So they said, hey, can I have a visual UI that would actually let me pull my audience data, figure out which customers are buying which brands, which customers have high propensity towards which artists that have new music coming out, all those kinds of things? And so now we give them a UI where the marketers actually log in. They're spending day to day running campaigns within our tool. Pretty much what's powering personalization for those kinds of big brands. So just recently, the NBA launched— they have a review of how many games you've watched and what your top scoring players are and all that kind of stuff. We power that campaign. So when you log into the app, the home page is all your metrics. And those metrics are powered by HiTouch.

As your customer base has scaled and the product has evolved, I think the thing that remains consistent is the team from a leadership perspective. Um, in our previous conversation, we spoke about having 3 co-founders and a unique leadership dynamic of co-CEOs, your co-founder and co-CEO being Tejas. Um, I'm also a co-CEO. It's kind of a unique model. Um, could you tell maybe founders in the room that are thinking about their relationship with their co-founder roles, how that might evolve as companies scale? Maybe give us a bit of insight there and some advice for founders that are considering this.

Yeah, so I think I'll start with the co-CEO question because I think it's the more interesting one. There's a lot of advice I have for just being on great terms with your co-founders. It's probably the number one thing you can do for your company. If you don't have a good relationship with your co-founders, you don't really have a company because the company is the co-founders and everyone around them. So co-CEO is interesting. People think a CEO should just be the like end-all be-all solution for anything that ever happens. They should be able to do go-to-market, product direction, product vision, hiring, finance, basically figure anything and everything out. And they want everything to go up the food chain to be solved by the CEO. I think it's a little bit outdated. I think at the end of the day, you want the best person for each task. And so Tejas and I have always done a really good job of splitting go-to-market. So Josh has led product engineering. In the beginning, for the first couple of years, I led marketing and Tejas led sales. And then for the next year or so, I led sales and he led marketing. And the whole idea is that all three of us are actually pretty good at product engineering. So that's something that we know how to hire, we know how to build, but neither of us were good at sales or marketing. So we said, all right, we don't know how to do either of these two things. Let's go learn that. And then we did that. We hired VPs, we scaled it out. We category created, I think, three times now in the marketing side. And that's all stuff that we had to learn. So what I really like about co-CEO is if you can have really strong alignment at the top, and it's not just us both, but it's us three. If you can have strong alignment, then you can get like triple the work done rather than just having everything rely on one person. And so we've actually had like almost all of our revenue sourced inbound. The reason we're able to do that is because the marketing function's so strong. Usually a company invests in either sales or marketing rather than both.

I love this notion of, learning something and then firing yourself from that role as the company evolves and scales. I was recently talking to a founder that is in this YC batch and she was considering the co-CEO model. What's a framework you might have for founders who are considering it to know if that's something that's actually right for them?

I think it really just depends on how much of a fluid relationship you can have with your co-founder. There are some people that are like, this is all I want to do ever, and I'm not going to change that. So if folks want to do just engineering forever, that's really fine, but they have to acknowledge that. So Tejas, Josh, and I are kind of weird because we'll just solve whatever the most important problem is at any given time. So if you ask us, hey, what do you want to do? We'd just be like, I don't know, tell me what's important. And it really stops there. So all of our coaches have always told us, hey, you should have an idea of what you like, what you get energy from, and make your job work out so that you start doing those things, which usually people do know. But for us, it's always been whatever's most important. And that changes as the business grows too. So now we're going to be launching an AI product in a couple of months that requires a ton of product and engineering collaboration. So I'm really glad that Tejas and Josh are working hand in hand on that. And that didn't exist 2 years ago.

I'm excited to dive a little bit into the AI roadmap you have. But you mentioned different folks on the team were focused on sales and marketing at different points in time. Um, let's chat a little bit about Hightouch's go-to-market. How has your go-to-market strategy changed in the past 2 years?

Yes, so, um, we launched in, I'll call it 2020 July, and we originally thought that we could be a self-serve business and that we could be PLG. Um, and PLG was super hot. Everyone was like, let's do product-led growth. Let's have self-serve users that all like pay us with a credit card. It turns out like a lot of businesses are not built for PLG and they should not try to be a PLG company. We found a lot of founders were doing PLG just because it was hot or just because VCs told them that's what they should do. If you're a company that's built for enterprise, which fortunately or unfortunately we are, people will want to do security reviews before using your product. So you need a salesperson on that deal. What we learned is actually that we were never PLG even when we thought we were. PLG was just a free trial or like a demo. You could log in, you could see what the UI looked like, not use it. Talk to sales, do InfoSec, and then actually use it. So what we learned is actually like our funnel was always like marketing-led growth, like MLG if you will. And it was basically like people would read content, they would watch podcasts, they would read like blogs, not written by us but written by other partners. And using all that information, they would learn about reverse ETL or learn about composable CDP and then come talk to us. And so it's like basically the exact same as building a PLG funnel. There are still flywheels, but those flywheels are just very different. And I think one thing that we did very uniquely is we acknowledged the fact that we had no followers. So social media, like our blog, all of these things basically had zero followers. So instead, we worked with partners in our space and we said, hey, can we write for your blog? And so it was the exact same thing that we would have written. We didn't even publish on our blog. We only published on their blog. So we would go to Snowflake. We'd say, hey, can we publish? On snowflake.com/blog. They're like, sure, this sounds cool. And it was like 2020, so they're down. And then we did it. The post got tens of thousands of views. And that's what made Hightouch popular because we were just honest about the fact that we didn't have a following.

Yeah, it's about leveraging what you might have some secret skills in and maybe not doing everything under the sun that doesn't maybe necessarily resonate with your team. Well, certainly the markets have responded to your continued growth. No surprise going from a $450 million valuation in 2022 to a $650 million valuation last year is a huge feat. So first of all, congratulations. Thanks. Yeah, a round of applause is warranted.

For sure.

Um, we have a lot of founders in the room that are either fundraising or thinking about fundraising. What advice would you have around when is the right time to take on more capital?

Okay, so I think this is gonna be a relatively controversial answer. Um, the right time is when you have the most momentum. So like something weird happens in venture capital where as soon as you start having metrics, people scrutinize your business way more. Than when you don't have metrics. So a lot of people that have raised like pre-seeds and seed rounds know this, that generally you want to raise when you have the most momentum or leverage. Leverage or momentum can come because you have great user growth, you have maybe great revenue growth, maybe you're just in a space that's really blowing up and no one else is in that space. So whatever it is, or maybe you have a really good team. But you kind of want to find the things that give you leverage and give you momentum. There's this true saying for acquisitions basically, which is that great companies get sold, bad companies— or great companies get bought, bad companies get sold. You never want to be someone that's selling and trying. You'd rather be someone who's executing well. Because you're executing well, the folks that you're meeting through warm intros are saying, wow, this is exciting. And then they're kind of encouraging you to fundraise rather than the other way around. So yeah, the TL;DR is like, we, for example, fundraised in November of 2020. We didn't really want to. We were like, hey, we have plenty of runway. This doesn't make any sense. We have sub-$100K of ARR. We're not a big company at all. And the market really wanted to put money in. And our advisors were like, look, if people want to invest, this is a good time to take money. And so we had less than $100K of ARR, and we raised that. $75 million post. Different market back then. The same market does not exist today. But the reason we were able to take that plunge is because folks around us said, hey, look, this kind of momentum and leverage may not exist later on when you actually have more revenue. So just project out how well you think you can do next year. And as long as you can justify this based on next year's revenue growth, do it.

It sounds like— sorry to cut you off. I was just going to say, it sounds like you had some great advisors around the table. And especially as we're all a little bit on the younger side, who knew to us quite that 2020 was such an anomaly at the time. And it sounds like those people have spoken into the experience of building and scaling Hightouch. Usually with those capital commitments comes a board creation at a certain point. When did you guys find was the right time to create a board?

So in some ways, we've not really created one yet. So it depends how you define creating a board, right? So for us, it's us three as co-founders. And then we have 2 board members that come from our Series A investor and then our Series B investor. Everyone else is a board advisor. And so really there's just 5 board seats. Usually, like when people think about creating a board, they think about bringing in independent board members, folks that have been founders in your space before or like can give you really good advice. And that would be like a bigger board that would like collectively make decisions together. At our stage, what we do instead is really like the 5 board members we feel like are enough. And we want to operate in a very lean way. And we invite our whole leadership team. So our head of sales, head of marketing, head of finance, head of people, they all come to the board meeting, and they're all active participants and contributors because the board really wants to hear from them what do we think the next strategy is. And that's been working well. I think maybe in the next couple of years we'll probably create a board. But in many ways— so this is the part that gets controversial. A lot of founders think of the early stages as let me have as much flexibility to operate and learn and iterate. So I wouldn't want to scale too quickly in the wrong way. I'd want to like learn, iterate, learn, iterate, then scale. And a lot of companies make the mistake of like they have a product that's sort of working, they scale the shit out of it, and then it doesn't really work. So then they have to go pivot like later stage. You want to pivot as much time as possible with like less staff and like less capital so that you know what to do and then scale that thing.

We had KJ Sidbury of GV on earlier today, and we also asked him this on the converse side as an investor, and he was saying, I don't like surprises, right? Like, I want to be brought along for the journey, even if it's a 2 AM text. And so what I'm hearing for you a little bit is sometimes, you know, you go to the moon and then when you have to pivot, if you haven't brought those people who were on your cap table along for the journey, things can get hard from a communication standpoint. So what has worked well for you or not worked well when managing your board today?

Yeah, so because it's so small, like imagine it's like us 3, 2 people, and then like, and a, what's it called, a board, like the board advisor like from BainCap. Like we basically just text them updates all the time. We have a sync with them every 6 weeks. So board meetings every quarter, but the sync is every 6 weeks. So I give them updates. And pretty much any risk we see in the business, we tell them first. If people are pretty smart, they're going to see the same risks as you. So here's one thing that you learn. If you have smart board members, they're going to figure out what the risks are. You should just proactively talk to them about it. And they're going to be like, wow, great, you're smart. You thought about this. I either agree with you or I don't, but at least we're talking about it. The same thing goes for your employees. So a lot of founders and a lot of CEOs make the mistake of trying to hide all the business problems from their staff. They're like, oh, things are going amazing. We're just crushing. We're just going to succeed. And then they lose trust with their employee base because people know what's broken. They know that certain things are scary. If you have a big competitor that copied your product a few weeks ago, people want to talk about that. So we pretty much did the exact opposite. During all hands, we would say, here's the top 3 things we're worried about right now. And then we would talk about celebrations, but honestly, we often missed the celebrations. We would talk about the 3 things we're worried about all the time. And we would invite help. If you have ideas for how to solve these 3 things, talk to us. And what we learned is that because we like to hire good people that are high agency, that are really smart, they already knew those were problems. They were like, hey, thanks for being honest. I agree that's a problem. And actually, I have ideas to solve it. And then we'd be like, great, can you lead that? Maybe you can lead building the next thing that will actually help us differentiate. But I think we've gone through the cycles of being super candid, selling, being super candid. And now we're back to just learning that super candid is best. Even as like a 150-person company, we have competitors that copy us all the time. Every single time they do, we just tell the entire company right away. And we're like, we're either worried or we're not worried. Here's what we're going to do about it. And that's what like all the best leadership coaches will tell you. Like ambiguity actually hurts the most. But if you can just say, here's what we're doing about it, or we don't know, ideas are welcome, that's like the best place to be.

Well, thank you for letting us in a little bit to that window of stakeholder management between the, the board, the senior leadership stakeholders, and then of course your employees who you're ultimately responsible to, and then of course your customers. I feel like we've gotten to have a broad range of all who you're working with. And when we initially had you on in September 2022, it was actually just 2 months before ChatGPT launched. So it's been a crazy couple years, a bit of a watershed moment in how people think about using and querying data. So peel back this for us, how is the AI revolution impacted high-touch and your strategy?

So I guess I have a bone to pick with the idea of GenAI being the transformation because there's so many formats of AI that are not generative. And I think GenAI is exciting and people are really excited about using this to make data more accessible, to make it easier to write marketing copy, to make it easier to write marketing creative. But a lot of marketers we talk to are not ready for that. So if you asked a marketer, should we personalize your emails using GenAI, they would say, hell no. They want to write the copy themselves. They want to use GenAI internally to see what the options are, but they want to write the copy themselves and approve it. So what you learn as you talk to more and more CMOs is GenAI is super exciting from an accessibility perspective, an iteration perspective, brainstorming perspective. In fact, You can go ask ChatGPT for ideas, things like, if I sell glasses, what kinds of campaigns should I think about? And it'll tell you, actually, at the end of the year, people are going to be done with their HSA benefits. So you should sell people like, hey, you haven't used your HSA yet. You should use your HSA on our, like, ibuydirect.com. So those are pretty cool campaign ideas. So that's what people want GenAI for. That's how we're integrating with it. And we're pretty excited about that. But where we think the real innovation is actually is on the more traditional machine learning and deep learning side. And that's actually basically helping marketers decide which campaigns to send to its customer. So I, as a marketer, I love writing creative. I love writing copy. But I want help choosing, should I sell you food for your dog, food for your cat, or actually doggy daycare? And that is exactly what PetSmart is working on right now. They're like, so actually in their CRM, it's pretty cool. They have you as a person, they have all your pets by name and by breed and by birth date. That's a novel thing that basically only PetSmart does. A few other retailers are adopting this for parents and their kids or pets and other things. That's novel. They want for people that have a dog, as soon as you start shopping for another dog, a separate dog or a cat, they want to onboard that next pet. So they call it pet onboarding. If as soon as you start shopping for services, things like doggy daycare or doggy spa, They want to make sure you come at least twice to Doggie Daycare or Doggie Spa. That way they get you as a recurring Doggie Spa user. And so those are the kinds of tasks that we think AI is excellent at because it can help decide which of these 3 campaigns is the most important on a per-customer basis. And then it can run the automated campaigns using copy that the marketer likes and approves. So copy's always safe. Copy's always like CMO approved, but data's being used for the stuff data should be used for, which is like personalization.

Man, I didn't know there was doggy spas out there. I'm kind of jealous.

I think a lot of PetSmart locations have this.

Yeah, this is pretty sick.

Okay, good to know. But yeah, so we're— this is what we call like an AI decision engine. Uh, the idea is that like machine learning should decide what we send to what customer. Um, if you wanted to, you could say, I'm going to basically stick an LLM in the decision-making process. So every time the customer does something, the LLM decides what to do next. And then the customer does this, and now the LLM decides what to do next. But it turns out LLMs are the wrong architecture for that. You actually use reinforcement learning instead. And so that was how we got to where we are today. It's not generative, but I think it's actually much more exciting.

No, thank you for drawing out that distinction. And I'm actually going to take us in just one layer deeper, pun intended, on the data. So it has changed the way data is being stored, and it's there's been a bit of a rise in data lakes over actually the stricter databases and data warehouses. And I know Snowflake is a really big partner of yours. What have been some of those key changes to the data stack that you're working with in your day-to-day? And what are we looking towards next year in 2025?

Yeah, so best way to articulate this is typical data, just based on a strict data warehouse like Snowflake with tables, can do a really good job of deciding things you already know. So, like, person shop for product, didn't buy product. Obviously, I want to sell you that product. It's, like, really clear. When you want to make decisions off of, like, infinite things and you don't actually know what those decisions are going to be, that's where you need access to all the raw data. So you'd want to have basically, like, let me just look at all e-commerce shopping patterns that you have ever seen on your website. And then based on that, I know what people buy this product. And then often buy this next product. So I can draw conclusions in a very data science-y way and then apply that back to the marketing. So to do completely unstructured analysis, you need two things. One is you need access to unstructured data. And two, the data layer or the AI layer should be sitting on top of that unstructured data in order to have access to basically an infinite amount of data. With regular data engineering, it was impossible to parse all that information. But with AI, it's quite easy actually. So that being the core, change in need exposes need for now infrastructure like MLOps and just data lakes in general to be able to expose that data to LLMs. I think actually like, so like RAG is pretty popular. I think RAG is actually quite insufficient for what people need to do, especially when they're doing it internally on their own datasets. And so there's like really interesting ways where you could, I guess, backtracking, like right now what's happening is we're using LLMs to write SQL queries. Against our data, the LLM does not know what the right structure is to write those SQL queries against. So it might be querying the wrong table for users, the wrong table for transactions. It has no idea what the right tables are. So if you had a semantic layer sitting between the LLM and the data, you could actually tell the LLM where all the good data is and where all the canonical definitions of like revenue, LTV, et cetera are.

I'm glad you brought up the semantic layer, um, because that's what underpinned your product High Touch Audiences, which you would really recently launched in 2022 when we last spoke. And at that time you gave a really helpful description about how teams are externalizing these data points to be able to be used by non-technical stakeholders, me. Um, and I'm just gonna like re, re-quote you from 2 years ago and say, um, your words, as semantic layers get more mature, we would hope to interact with them and power high-touch audiences based on existing semantic layers. Are we there yet?

So we never got there. We were hopeful. So there were a lot of companies that were like, we're going to build the semantic layer. And Hayat was like, great, we're just going to integrate with you. Everything will work out really well. We haven't seen adoption yet. So what ultimately ended up happening is we ended up building our own semantic layer. It's different from others. We're not trying to expose ours to the rest of the universe. Just broadly, what is a semantic layer? It helps you define what the relationship is between different pieces of your data. So how does a user relate to orders? How does orders relate to product SKUs? What pictures should be in those product SKUs? Or how does a list of transactions per customer relate to LTV? Like, those are the kinds of things a semantic layer would define. We basically had to say, like, you know what, like, we'll have to build one internally. We'll expose it to our customer base and it'll power Hightouch specifically. So we built a very specific one that powers Hightouch. And that's actually why, like, for really large enterprise, we're able to query completely different datasets between enterprises, but with the exact same UI. Because our UI is talking to the semantic layer, which is talking to the database, and that is actually what's teaching us how to query your data. So now we can query things like, show me people that have bought dog food recently. But not bought cat food. So I need to be able to know if you have both a dog and a cat and whether you bought food for each rather than just food in general, right? So that's a very different task. No other marketing tool in the world literally at all does this task that I just described. And almost every single retailer in the world has this because they have the concept of you buying for your spouse, you buying for your kids, and you buying for your household. So Instacart now has Instacart Plus. So Spotify has a family plan. None of those concepts work for any other marketing tool other than HiTouch. That was like the core like problem that we found. The semantic layer was the actual solution to it. And when it comes to our audience builder or the next AI thing, both of them should hopefully like keep working really well because of that layer.

We've gotten pretty nitty gritty into the data stack component here. So I'm going to bring us back out up to a layer that maybe I'm going to say everyone can understand, including myself, better. On the GenAI app side of things, how can we think about your product and some of these learnings you've had about empowering and surfacing data for, again, like I said, the non-technical user. Like, how are you guys thinking about how you're interacting with the apps? Yeah, so, uh, great, great question.

And I think a great way to think of it is imagine if we were deciding which campaign to send each person, right? So we could tell you, here's everything we've ever learned about this exact consumer, um, you being the LLM. So at the very edge, right when we're ready to send the email, we could say, all right, we've learned so much about our customer. Here's everything they've ever done, what they haven't responded to, what they have responded to. What should we send next? And then we could actually have the LLM propose the best copy or the best creative. So PetSmart could, based on your breed, send you the picture of the dog that very much looks like yours rather than a random dog. And so that's where it gets interesting. That's what we're integrating with. And our belief is— so also controversial— our belief is that actual LLM providers, Anthropic, Mistral, and OpenAI are going to get better and better and better. So we'd rather just keep integrating with them directly, keep benefiting from those integrations getting better. And then ultimately, the channel tools, all the email tools, are still the email tools that we still integrate with.

I know everyone likes GPT, but I'm kind of a Claude fan.

What do you think, Claudia? Well, I mean, I think I'm biased in that a little bit.

What's the bias coming from? Claude?

Claudia? Oh, oh. Because she, she. Oh, OK.

It's OK. It's almost happy hour.

That was a tough one.

Well, we chatted a bit about the future of personalization, the modern data stack. We've started Insight Summit to bring together tomorrow's IPO founders and really create meaningful connections for the folks that are building for the next decades. We all graduated in 2018, and so I think we've been in the workforce for 5, 6 years now. What's your take on this generation of technology founders and how are we doing things differently than the founders that maybe came in 2008?

Yeah, I mean, I would love to get your take on this as well since y'all are both founders as well. I have, I guess, positives and negatives. I'll start with the positives, hopefully. There's just so much more agency than ever. Like, it used to be the case that you would think, I need a job experience in order to be a founder, I need like wealth, I need like network. Now folks are like, fuck it, like, I can be a social media creator, I can be a startup founder, like, I can do anything I want, I can teach myself how to code. So the amount of agency has just gone up like crazy. The amount of access, the amount of like access to VC dollars, like all of these things have just tremendously improved, which is great. Like, that's exactly what I want to see. Because anyone that has the hunger and the grit to become a startup founder ideally should be able to try and should be able to succeed. So I think that's the really positive thing. I think founders support each other more than ever. I feel like— and I would, again, really like to get your take on this. But I hope that people lift each other up more than ever. What I love about San Francisco specifically is that people just always play positive-sum games. What I notice about a lot of the rest of the world is they're zero-sum. It's like if I win, someone else has to lose. But the SF mentality has always been if I win, everyone else is going to win around me. Or if I help you, it's going to come back around. It's going to help me later on. And so people are always introducing folks to people they can hire, folks that can invest in them, folks that can become startup founders with each other. There's always introductions being made. And I think that amount of flow has just increased like crazy. In a really good way? Would love your take on that.

Yeah, no, that really resonates. I feel like by making it easier than ever to start a company, you do get people who are super passionate about it and should be founders becoming founders. But in that same vein, I think it's become a popular path to pursue. And so you have founders becoming founders for the sake of being a founder. And I think, you know, with the ease of, you know, social media and all these different tools, you do find founders maybe building without really solving deep problems. And that creates a lot of noise in not only just founder communities, but the market and distrust from customers that might have been burned by another startup that maybe did not have the best product or it wasn't all there. Um, so I think that's kind of an interesting part of this current environment, but at the same time, there is so much of a founder community now that I think that's really beautiful. Like, you know, you can get a room together like this and people are all building alongside each other. And I think that's a unique benefit that founders that were building, you know, Web 2.0 just simply didn't have.

And I'll just add too, I think even our story of how we met, I, um, can admit to Twitter DMing you and being like, can we have you on the podcast? And you got right back and were like, vibes, I'd love to be on. Thank you so much. And you know, that couldn't have happened 10, 15 years ago, really, right? And I think this URL to IRL friendship moment and community building, you were telling us just before that a part of your go-to-market is throwing like a party at these big conferences for— Yeah. And so I just think the way we do business is a little different. It's a little more friend-forward, um, and it's also digital native. And so oftentimes these friendships, especially because of the pandemic, happened first online, then, um, in person, and then back online. So I don't know, that's kind of kind of a shared mindset of how we've built our companies and you've built yours.

Yeah. I mean, only in this day and age are you potentially misrecognized as a club promoter. You know, that, that wasn't happening 10 years ago.

We can't tell people about that. Yeah. Uh, no, but you really took the words outta my mouth. Um, what I see from a lot of founders these days is too much of a desire to make money rather than, as you said, solve really hard problems. So when we think about where did entrepreneurship or mostly tech startup founder ecosystem start, a lot of it started from Bell Labs way back when. And people were like, I'm going to go solve the thing that is going to change human society forever. And they all felt like they were working on something that important. The issue is that each of these gets ticked off, and then there's less and less stuff that feels like it could change human society forever. So you get left with B2B SaaS, right? And so that is like, then there's the amount of ever-flowing VC dollars. VCs want to see you build B2B SaaS. And so that is kind of like my bone to pick with this huge inflow of founders. I think more founders is great. I think the topics that they choose to work on, if they were more driven by solving hard problems or solving important problems ideally, even if they're not hard, that would be where I'd like to see it go. Rather than trying really hard to make money. If you put money first and then you try to solve a problem later, you often end up in the wrong place. Whereas if you try to solve an important problem, usually, and I hope this is the case, the money comes later in a really natural way.

On the other side of the table, with more founders, there are more investors. And there are a lot of funders in the room today. In your opinion, how can tomorrow's funders better support the startup ecosystem?

Yeah, I think a lot of patience. So our investors did a fantastic job at this. They've never once pushed us to grow faster than we were comfortable. Every single quarter, Tejas and I have a higher revenue number that we want to go for than our board wants us to go for, which is novel. I'm sure everyone in here thinks that your board is usually pushing you to grow faster. You're like, no, I can't grow that fast. No, it's the opposite. They have very high patience. They're like, look. We want to get to the best possible end outcome. End outcome's in 10 years. Let's optimize that thing. Let's not optimize this year or next year. Tell me the big go home, go big or go home story. And so they have a lot of patience in us executing. They were like super down for self-serve. They were super down for marketing-led. We told them we're going to go enterprise, but bottoms up through marketing. That was the most controversial thing to do. Everyone was like, yeah, that doesn't work. No one's ever succeeded. We still did it and it did work, right? That's the kind of stuff that our board really trusted us on. So I would say patience is the number one thing. And basically the same mindset, figure out what the big long-term problem is to solve, go for that rather than look for a mill of ARR that's going to grow to 7, that's going to grow to 14. That's not that exciting. Figure out the big long-term thing.

Thank you for sharing those words of wisdom. As we come up on time today, let's chat a little bit about the future for Hightouch specifically and for you personally. What are you excited for? This year?

Yeah, I'm excited for a lot of things. I mean, I just love team building. So I think the thing that gives me the most fulfillment is bringing great people together, empowering them to just do amazing things. And as a leader, I'm learning so much more about how to just unlock people and let them run. One thing that a mentor of mine shared a long time ago that was just hugely eye-opening and it was specifically about introverts, but I think it applies to everyone. If someone you work with is like, hey, like, I think that this is going to be a problem, first thing you ask them is, all right, do you have any ideas about what we should do about it? Because very likely, if someone is bringing a problem to you, they also probably have solutions, and they also probably want to lead those solutions. So usually that's like where it's coming from. And if you ask them that simple question, they'll be like, oh yeah, here's all my ideas, like, I can do this for you, we can get this done. Like, that's like a great way to unlock people, right? It's Just continuing to become a better leader and hire great people and just unlock all of their potential, that really excites me. I know it's kind of a soft answer. Maybe on the more literal side, our landscape, especially for marketing, is going to change completely. People are going to depend on AI more than ever to help them understand what marketing is going to work versus what marketing is not going to work. And we're really excited to be the pinnacle of that. We feel like right now, we're the most data-integrated marketing solution, period. And so we have the best chance of actually being able to solve that AI task for them. And I feel like we're the only folks that are thinking about it in a non-LLM-driven way. A lot of people are trying to just ping the API for ChatGPT and see what works. And that's their entire tech roadmap. It's not like, let's go build a model from scratch and really understand. So I feel very excited about going in that direction. We're working with some of the top CMOs to help deliver this for them. And so if it works for 5 of the top CMOs, we know it's going to work for another 50. And that for us is where marketers can really beat their competition. So another thing I learned really recently is marketers want to beat their competition. You go talk to a big clothing brand, they're competing with all the other big clothing brands and they think that AI is going to be their competitive advantage. And if the competitor gets AI first, those competitors are going to win. So digital is the big first competitive thing, then e-commerce, and now AI. So all three of those things are things that I'm really excited about and Hightouch is really prioritizing.

Well, I'm excited for us to sit down again in 2026, and we'll see how things have changed. Maybe there's another bombshell drop like ChatGPT launching after our last conversation. That'd be cool. We have a little bit of time left, so if you're down for it, I'd love to open it to the audience for maybe one question. Yeah, let's do one question.

If you want to raise your hand. Yeah, I think I saw Henry's first. Oh, kashut.

Hey, before you pivoted to data software, did anyone on the team have a data background? And if not, what gave you the confidence to go into that space? Yeah, so a lot of people have asked me this question, which closely approximates to, if I'm starting a company, should I start a company in a space that I know well, or should I just do the thing I'm passionate about, or things that are just a really big problem? My answer is always, do the thing that you would keep wanting to do even if life sucks, even if it's really hard, even if it doesn't go well. We had some people that had done CDP, like customer data platform, which is right in the space that we're in. But no, we learned about the data space basically from scratch. A ton of big enterprises told us, we don't want to use the thing that was called CDP before. We're going to switch everything over to the data warehouse to Snowflake and Databricks, and here's why. And they taught us why. They taught us how. They taught us how we should architect our entire product, basically. And then we architected it and built it for them. So we were able to capture their insight and productize it rather than having our own novel insight right away. And it turns out they're probably smarter than us. They're lead data people at Fortune 50 companies. And so I'd rather listen to their insight than my own. And I feel like everyone should have that kind of humble perspective that I probably know less about every topic in the world than other people.

Well, thank you so much, Kashish, for both, uh, being back with us in the room, joining us the first time, taking a, a chance on our podcast. We're so grateful to have had the chance now to twice share your incredible story of building HiTouch. So can we just get a round of applause for Kashish?