Howie AI: Austin Petersmith on Building an AI Secretary

[00:00:00]We think that professionals spend a ton of time on meta work which is just work about work. So anything a great assistant can do involving the calendar.

[00:00:09]Howie can do welcome to the peel. I'm your host Jordan Novak, founder of Banana Capital. Today's guest is Austin Petersmith, co-founder and CEO of Howie, the AI secretary.

[00:00:20]>> Every AI product is called assistant or co-pilot and secretary is a word that more accurately describes what Howie does. Howie officially launched two weeks ago with a viral video and much customer love. This conversation goes inside the early days of the company, including Austin and the team's very opinionated approach, making Howie one of the most narrow use cases of any AI product I've ever seen. It has turned out to be a lot harder than we thought to solve it. We've built the flywheel and the architecture and all these things to solve the one thing and then we can do more and more over time. But all this has made it one of the few AI products that actually works, growing double digits every month for the past year.

[00:01:01]>> When you build for text message or Slack, people expected to respond at the speed of chat GBT and email has latency built into the system.

[00:01:08]>> We talked about their decision to completely rebuild the product last year, adding a human in the loop element, similar to how Cruz and Whimo train self-driving cars.

[00:01:16]>> We can always have a product experience that is far ahead of what the models at a given time are capable of. We also go inside the making of their viral launch video.

[00:01:24]>> We were planning for September 22nd and I think it was like September 5th that I texted Matt >> why more software should be augmenting humans with AI.

[00:01:33]>> Everyone is kind of trying to build the fully AI accounting firm instead of the AI accounting firm that has phenomenal accountants.

[00:01:41]>> Lessons on building a brand from scratch and the unique pitch deck strategy Austin used to raise $6 million. you should just stretch it out farther because they're going to give you 3 seconds for every slide and then they're going to keep paging through. And what he learned working for Jason Calcanis, >> he was like literally like studying the tapes of these interviews.

[00:01:57]>> A quick thank you to my co-investors in Howie, Adam Doelli, Sophia Amaruso, and Alex Goen for helping brainstorm topics for Austin. A reminder, I publish episodes of the Peele every week exploring the world's greatest startup stories just like this one. Check out the back to lock of over a 100 episodes and tune in next week for a conversation with Eric Burnhardson, co-founder and CEO of Modal, building AI infrastructure that developers love. Now, let's talk to Austin after a quick word from Handover Park. Hanover Park vertically integrates fund admin, portfolio management, and the LP experience for finance and investment teams. Most of you have probably interfaced with a fund admin provider in some way. They're a necessary evil for every type of asset manager across not just venture but also private equity and private credit. They provide bookkeeping and accounting so investment firms can report to their investors on a quarterly basis. What's crazy is they charge hundreds of thousands or millions of dollars per year to basically not screw up your accounting. They sit together thirdparty software like QuickBooks, Bill.com, Salesforce, and Excel and then throw a bunch of bodies at you. And that's where Handover Park comes in. They built their own accounting system from scratch, which ingests all your firm's data and documents. And their AI native solution automates all the manual work that drives private market investors crazy.

[00:03:11]Head to hanoverpark.com/turner and try the AI native ERP for private market funds. That's h an ovar.com/turner and 10x your fund admin.

[00:03:26]Austin, welcome to the show.

[00:03:28]>> Thank you so much for having me. Hey, I think this will be fun. We were talking right before we started recording all the stuff we're gonna we're going to hit on. Really quick though, for people who are not familiar with Howie, can you just give us a quick high level on what it is?

[00:03:40]>> Yeah, so Howie is the modern secretary.

[00:03:42]He's a mix of AI and then also we use humans in the loop to correct some mistakes before they happen and totally focus on scheduling. So he's email based. you email Howe@ howy.com or can even like white label him and give him a different name and an email at your domain and anything that you do with your calendar that is tedious how he can do it. So you CC him on a thread. He can offer time slots. He can reschedule meetings, set up Zoom links. He knows where you like to take your in-person meetings, can pick up subtle signals like maybe a secret phrase that means a meeting is low priority, and then he'll schedule it far out in the future and will notify you if there's like conflicts coming up on your calendar or if you have a meeting coming up that you haven't rv to to just like help you with your general calendar hygiene and he'll click calendarly links for you so you don't have to click on those and can look at other people's calendars within your organization. So if you're scheduling a call externally and want to offer the overlap of of your availability plus two to other people then we can do that. And so anything a great assistant can do involving the calendar how he can do.

[00:04:46]>> Yeah. I think my favorite feature is the giving it a different name. So I had a tweet couple months ago. I called mine Chamoth and that got I don't know. I think you said it was one of like the best days of for the company for like new revenue when I tweeted that. I called it Chumat and then >> yeah it was like right when we had rolled out that feature and people want that feature because they it's like super cool to be able to name it. So most people have not done like kind of most people have done more serious names but we've seen a huge mix of different things. Some people call it like bot at whatever because they really want it to feel like AI and other people create like a whole persona of a of an actual kind of like human and and name it based on that. And so we've seen a really really wide mix but but we have tons of them now and and it's really fun to see and we've seen also times where there's like two howies on the same email thread but both it's like Victoria and and George and they're like interacting with each other and they're both actually Howie under the hood but they're but they're these two different identities.

[00:05:44]>> It's like one of the meme with like the interconnected brain like going like the AI like the AI is going back and forth.

[00:05:51]>> Exactly.

[00:05:53]And so you mentioned one thing that I want to talk about later, but you said there's like a human in the loop element. I know you that was like a big decision on doing that. We can hit on that a little bit, but but you kind of had to do that because it's so hard to build this thing. Like why is it so hard to make an AI assistant that specifically lives in your email? Like it kind of seems like the easiest possible thing you could build really with AI.

[00:06:19]>> Yeah. So I mean the email architecture is not is not the hard part but what is hard is there's a lot of context in email threads. By the time howi gets CCed on an email thread there might be 10 messages that went back and forth already that involve lots of mentions of dates and times. People's email signatures mention times or time zones or locations and the actual metadata in the emails has lots of times and dates in it and the time zones often are different from one to the next. And so there's a lot of kind of red herrings in an email thread. And then there's also a lot of human intuition in deciding what to do. When do you actually respond versus not respond? And and subtle signals that people include. And so and all these things, you look at them, none of them are hard 100% of the time, but they are very hard for an AI model to get right 100% of the time. So it's like the the models can easily 50% like two days into building this product we had 50% accuracy that was just like it just half the time it does exactly the thing you want and then it's like this climb where each 1% past that is harder and so the main thing that makes this really hard is that mistake tolerance is very low. So if chat GPT makes a mistake then you just kind of move on. You're like oh chat GPT made a mistake. GPT 3.5 was a phenomenal product, the fastest growing product in the history of the world at the time and it was pretty bad overall.

[00:07:42]It would make tons of mistakes and hallucinate and couldn't get a lot of like really basic stuff right. But scheduling there's a very low tolerance for mistakes. Like we one time sent a customer to a coffee shop and we had forgotten to include the other person on the calendar invite. So he was sitting at a coffee shop waiting for a meeting that was never going to happen because the other person hadn't been invited.

[00:08:00]And you can't make that type of mistake very many times. And and so we when people our customers schedule with their investors with with candidates they're trying to hire with with their customers and those are just not interactions that you can afford to screw up. And so so the low tolerance for mistakes is a big part of what makes it really high. And so you've got to be able to create a lot of human intuition in the product where it understands what people are trying to say and it doesn't get above it ahead of its skis in terms of like trying to engage in the conversation too much versus just like focusing on the scheduling aspect. And then there's just a lot of ways that mistakes can be made when you're scheduling across time zones and um scheduling with a whole bunch of different people who have all shared different things about their availability. And so there's just like a lot of ways that that these things can fail on the margins.

[00:08:54]>> And I know this is kind of a long time in the making, like the product, like where did it first kind of start?

[00:09:01]>> Okay, so my co-founder Dave and I worked together a long time ago, 2013, 2014, and and then in 2015, we Slack introduced the Slack platform functionality where you could build a Slackbot. And we built a Slackbot that was like you put it in your company Slack and you ask for anything. And we quickly got a bunch of people using it who really liked it. So they would they would ask for whatever they wanted.

[00:09:23]They'd be like, "Review this stack of resumes for me or schedule an offsite for my team to go to Cancun tomorrow or next month or something or like get sushi delivered to our office on the Lower East Side right now." And whatever the tasks were, we would just like figure out how to do them. And >> wait, so like was it was it AI or you like literally like doing this stuff on the back end?

[00:09:44]>> Yeah. So it was totally humans and but our idea our idea was that it would be humans but we thought we can build a bunch of reax tech like we can build some technology that will streamline the requests so that we can route them and by the time that a human sees it it's like kind of been through some filters that that simplify the task and make it repeatable and we thought we could build technology that would make it scalable.

[00:10:06]We quickly saw with the the kind of variety from one request to the next that there was no way there was no technology to build. Like if we were going to build this business, we were literally building a call center and we decided not to do it because we're like this is crazy.

[00:10:20]And that was 2015. Then years later, we both were like playing with with LLMs a bunch. We we've been closed ever since and and have like talked about building something together over the years. And so a year and a half ago or whatever, we were like, "Hey, maybe actually now we could build that thing we tried to build but have it be technology this time."

[00:10:40]And we started prototyping a little bit and quickly came to the conclusion like it still is not possible like to build the generalist assistant that you can ask to do anything. We did see at the time a bunch of companies that had raised money kind of on that promise and still now like none of them have really caught on. Even Microsoft Copilot is like kind of an example of that. It's like it it purports to do a thousand different things, but it doesn't do any of them particularly well. And so we felt like the answer here is pick one task and a task that you can do really really well and blow that out of the water and then over time layer in more and more. And so we chose scheduling because we thought it's it's a frequent tedious use case for for billions of professionals around the world. like no matter what industry you're in, scheduling is a part of your is a time suck on you that happens every single day and it has distribution kind of built in because there's usually multiple people involved in a a scheduling interaction. So a scheduling product that does well will get visibility to other people and and can kind of grow through PLG in that way.

[00:11:46]And then it it felt solvable like it's narrow enough that we can actually solve it. And we we're naive on that third point which we'll we'll talk about kind of some of the challenges like we it's it has turned out to be a lot harder than we thought to solve it. Um and but we're proud that we're like we have delivered a product that really stands out in the market on this this narrow task. And so if we had picked anything less narrow than this, we would have really fallen on our faces. And instead what we picked was like what we thought was like not too difficult to solve. and instead it turned out to be difficult to solve. But it's still a really great kind of starting point where we've built the kind of flywheel and the architecture and all these things to solve the one thing and then we can do more and more over time. And so that's kind of the origin story. And we really quickly validated that people want the product and are willing to pay for it and more slowly kind of were able to get to the product that that we have today that actually works.

[00:12:42]>> Yeah. Yeah, cuz it's interesting that it's it's so narrow in like like niche or just specifically like the one action of like scheduling a call. Like I remember when I first met you like couldn't even do like in-person meetings to your point about messing it up. Like it was so simple but also was super broad. Like everyone takes meetings. So it's like narrow and niche but also extremely broad at the same time. So it's kind of like this interesting, you know, Goldilocks just the perfect amount of, you know, finding where you kind of enter this personal assistant AI agent entry point into like, you know, expanding the use cases over time.

[00:13:21]>> Yeah. I mean, I think for other people building AI agents, there is a lesson there because there's there's a tendency. It's scary to put a product out into the world and there's a tendency to then because it's scary, you want to try to have it be able to do more. So, it's less likely that a person is like, "Oh, that's not that's not for me." Because there's just like more things that it claims to do. And LLMs have made this even trickier because you can throw anything at an LLM and get something back. So if you're making an AI agent, it's like why not let it go research the web for our customers or why not let it like I don't know do whatever other thing that's like just an API call to GPT but but it the narrower you go even though that's scary it actually like allows customers to self- select like it you don't sign up for how if you don't schedule lots of external meetings but if you do you're going to have a phenomenal experience and you're going to absolutely love it and it's going to be indispensable for you and after they self- select it also lets customers know once they start using it what to do with it. And so there's lots of generalist AI products out there that like I I try all these tools and it's like there's lots of them where I'm like okay it said it can do like all these different things. What do I even start with and which one is actually going to work for me? And with how we don't have that because it's just like you use it to schedule your meetings and that's what it does. And if you're not interested in that, it's probably not going to be useful to you. But if you are, it's going to do a great job at that and you're going to know how to use it right away. And we can stand behind the experience you will have within that because we've designed it very carefully and so it's going to work for you and it will become indispensable for you and you'll build a lot of trust and then when we start doing the next thing you'll already have that trust in place.

[00:14:58]>> Yeah. Even like when you talk about like how hard some of these things are like even with chat PT for a while I just didn't know what to do with it. Like you you sometimes with these products you like need someone to tell you like I think there's a reason that sort of like tutorial videos are so important just like how do you use insert X products? Like I've been using Atio the the kind of I know it's like a modern CRM and it's like a little the product is like pretty complex. you just watch YouTube videos like oh that's how you do that thing or like another one would be like notion like I'm sure when you first started using notion you're like what do I even do with this thing but you like watch a video like oh you can use it to like plan out a product road map or like internal wiki or content calendar like I use it to plan a podcast use it for pipeline for investing like things that I'm looking at and it's like you kind of know maybe you could do those things maybe but just like seeing it and having somebody explain to you how to do it it does it does help But again, it's like super simple, granular. People know exactly what you use it for. How did you decide email? Because I think it's probably one of the very few AI agents that I've seen that like lives in your email. It's kind of like a you know interesting entry point and that was just it was just a function of of email like scheduling meetings. Most of it's done in email.

[00:16:18]>> Yeah. So I mean types of scheduling happen in a in a variety of different places. So like a lot of in internal scheduling mostly happens in like Slack and Microsoft Teams and I hear a lot from like crypto people who are like I I want to use your product but I need it to support Telegram because that's where I do all my scheduling or or signal or whatever the different different apps are. And so and then like tons of customers from the beginning have been like we really want I really want text messages because I like do scheduling on text or I want to be able to text Howie to say like put this soccer schedule on my calendar and it's just not like conducive to me to switch to email to for that type of thing. And so it's just like lower friction to fire off a text message. But for the coordination of external meetings, email just is where that that happens. So, it's it's like where booking links are sent more than anywhere else and it's where the back and forth of like how about this day, how about that day or like any of that that it just happens on email more than anywhere else. So made a ton of sense as a place to start. And then it also is a fun place to build an AI product specifically because when you build with for text message or Slack or any of these other things, people expect it to respond at the speed of chat GPT. And email has latency built into the system and then social latency built into it too where it's like you don't have the expectation of the fastest responses in the world on emails. And so we like between the humans in the loop and then also the approach that we have to inference which is we use lots of fine-tuned models, we use lots of different models and we and we use heavy reasoning on a lot of different steps.

[00:17:54]And so there are some tasks that we do where even without a human in the loop, it takes us 10 12 minutes to complete the inference. And that's not an acceptable response time on on text message, but on email that is that is totally an acceptable response time. And so it gave us the ability to to like really focus on how do we get the result correct as opposed to how do we get a good and fast response.

[00:18:22]>> Yeah. What do just generally maybe maybe how we or more broadly too but like what do what are you seeing some of these agents being specifically really good at today like different types of use cases or on the other side like they still just can't do like what are the limitations? Generally speaking, >> it's really hard to get them to be perfect at anything. And that's hard for us to kind of comprehend sometimes because it's like you play with it and it works. And so then you expect it to always work. But but it's just the like the 5% mistakes are just just don't work because the in any interaction there are a whole bunch of options of opportunities for it to have that 5% error. And so it's it's like the it just you it's really hard to get perfection in pretty much anything.

[00:19:15]I think where the models are really good today. satisfied because I'm like for for our task even like the model I'm very dissatisfied with the I mean it's like I'm obviously we are living in an unbelievable era and it's magic and create mind-blowing what we can do that we couldn't do years ago but like we I do not think that the models are capable of being an executive assistant even within the scope of scheduling today I don't think anybody could build it even if you had a thousand AI researchers and and because there's so much intuition and subtle nuance to the right step that needs to be taken, but at the same time, they're very good at speeding things up.

[00:19:54]So, what we have for our humans in the loop is this kind of like interface that allows them to do executive assistant tasks faster than anybody ever has before. So, the tasks actually run the models make predictions in a simulated environment and the human by the time they see it, they can see start to end what the model is trying to do and they can chat with it to interact with it. So, they can say, "You got the time zones wrong here." And then they'll see it reimulate right in front of their eyes. And that's a pretty magical experience where in January when I was our human in the loop and the only human in the loop that we had and and then and then Nicole on our team started doing it and it was like just the two of us but it like we were spending like 30 minutes on these tasks that now we have people doing in like 90 seconds. And so, so where where it's worked really well for us is like massive efficiency in still a human doing a lot of the kind of heavy lifting of the task or at least like spot checking to catch the mistakes.

[00:20:49]>> Yeah, maybe that's worth talking about because I I remember it was it was kind of a big decision to kind of tear it all down, rebuild it, incorporate the human in the loop element. What kind of went into that and how did you decide to do it? We were doing the same thing everyone's doing which is like prompt hacking trying to get the models to be better trying different fine-tuning approaches RL we were trying all the different things and we we could get the numbers to move but we could not see have any line of sight to them to the product being good enough that it never makes dumb mistakes and dumb mistakes were just costing us customers. It's like people love the promise of it. they would start using it and absolutely love it and then and then a couple of dumb mistakes and they're just like I would rather just do this myself or pay for a full-time assistant. And and so we could see that the that like chipping away at at like one basis point at a time to try to get to 100% accuracy was was just not going to work. And like we don't it's not that we are the greatest machine learning engineers in the world, but we're also like pretty good at what we do. And we think we thought that like if We weren't just like one or we weren't it wasn't engineering resources that were the thing that was stopping us from actually like climbing that hill all the way to the top. It was like these are constraints that are built into the system that we need to work with. And so acknowledging that and then acknowledging that we have chosen space where the mistake tolerance is really low made us start to realize, okay, what we actually need to do is just pull forward the best possible product experience and and then chip away as we go at getting closer and closer to the models being able to do that. And then as we do that, we can also like give the humans a more complex task. And so we can always have a product experience that is far ahead of what the models at a given time are capable of. and and that will allow us to like get better at training the models, get better at doing the actual actual product experience and then move faster in terms of like taking on new new types of tasks that are beyond what we've originally kind of set out to do. So once we had that clarity, it still was scary. Like nobody software people don't like operationalizing a human team. We have 65 people now, full-time contractors that work around the clock that are the humans in the loop behind Howy. And that has been hard like in terms of recruiting the people, onboarding them, and managing them. with three three people on our team right now are at an off-site just planning like the actual onboarding for like the next cohort of of people that we're that we're bringing on and the documentation and like they've created simulated environments for people to like practice and and all these different things that just it's like we've ended up building all of our engineering most of our engineering time goes to building this product that the humans use so that we can create make them incredibly efficient and incredibly accurate at these tasks and that was like daunting and we knew that if we went down this path, that's what we'd have to do. And so, it was a tough decision. It's capital, it's time, it's energy, it's like we have to deal with human challenges that are totally different from the software challenges. And and so we have two different layers of like the challenges in terms of getting the thing right. But it was so clear that it would allow us to pull forward a better product experience than was possible today. And that was really important because we wanted to serve the customers at the top of the market who are people that really actually need a full-time executive assistant and we want to be as good at at managing their calendar as a full-time executive assistant. And to do that, we just had to find ways to fill in the gaps where the models weren't capable. And and we had the capital to do it. And that's like why you raise venture capitals to be able to take risk. And we started to see like no one else is doing this.

[00:24:37]We inevitably are going to have competitors like might be two two kids in YC that that are probably smarter than us. Definitely don't have kids so they can work around the clock and so they can work more than us and and so what do we have that they don't have?

[00:24:51]And one thing is like we actually can operationalize this. We have the capital to do it. We've we have managed big teams. we know how to do it and we know how to build the software tools at scale for them to use and and so suddenly it felt like okay if that if it's right that that is the the path to do this then us doing it now puts us makes us much more defensible actually and and that has like really played out well for us so far where we've just been growing pretty pretty quickly since the moment we made that decision back in January >> Oh really? I didn't realize it changed the growth trajectory. Yeah, we last year I mean we were weight list only so we were kind of constraining the growth but we the churn was a big challenge and and usage retention like we had people who weren't turnurning but also weren't using the product that much and from January we it was like the last two weeks of December over Christmas that Dave and I mostly Dave really actually rebuilt this the entire kind of system to support the humans in the loop and in January we onboarded one customer and it was a customer who was happy with the old product, but it was not working well enough for him. His name is Diego Oppenheimer. He's awesome and and has since become a friend. I didn't know him at the time. And we said, "We're not going to add any more customers to this until Diego is happy." And we saw his usage just skyrocket. So, it's like he was telling us that he liked the product in December. And but it wasn't until he had a better experience that he realized how much more he could get out of it where it's like he was just subconsciously narrowing what he would trust Howie with and suddenly was like I can it it can fully manage my calendar and and so he leaned into it so hard and then yeah and then since then like it's really just been I mean yeah been a steady clip of like 30 to 50% month overmonth growth and that's been totally just the word of mouth from customers talking about how great the experience is And then similarly like with the retention like the DAMO ratio has been 30 to 40% all year which is just for an early stage product like ours is just really good to see in terms of how often people are actually using the thing and and so that's really exciting because sometimes the the MR and the new customers can can be harder. It's like lot people will come in and try anything basically and so like seeing that they actually end up using it every single day is just really really cool to see.

[00:27:13]Yeah, when you said Dow mouse you said 30 40%. So what is that like two or three days a week?

[00:27:18]>> Basically what that tells you is that is that on a given day of all the people who of all the monthly active users all the people that used it in the last 30 days 30 to 40% of them used it that day and every day that's true. So today 30 30 to 40% of the people who used Tawi in the last 30 days have used it today and that will be true tomorrow and the next day and and so for a lot of products where like that can be just a red a red flag where you see it's like lots of people are interested in it. Lots of people might throw in their credit card, but the dowo is 10% which just means most people, the vast majority of people are just not using it every single day, which maybe means you have a useful thing with a less frequent use case or maybe means they find it kind of interesting but not actually like that useful or maybe they're all just turning out really fast.

[00:28:07]>> And then I think doesn't Whimo kind of classically had this like human in the loop model forever and I they still they still might. I don't I don't actually know. Yeah. So they do a lot of different things. So actually Cruz was really the one where and I mean we yeah because Cruz was doing this kind of for a long time where we got to know lots of people there in SF back in the day where Whimo was a little bit more quiet about it but but both of them and yeah I mean they the the model like anyone who lived in SF over the past decade you saw self-driving self-driving cars driving around on the streets with sensors on the roof and but there was a person behind the the behind the wheel and then another person on a laptop in the front seat and the two people in the car and >> in the self-driving car two of them.

[00:28:52]Yeah.

[00:28:53]>> Yeah. So, it's like Yeah. So, it's not a self-driving car. They were just they were just teaching the cars how to drive and and then they do that at a much larger scale in simulated environments where they have people in the Philippines who are in these environments and what they're doing is creating really structured training data finding where the where the models because when the when the human is driving the car the model and the sensors are still trying to predict what to do and they can cross reference that with what the human actually did and then they figure out okay here's a scenario where our models are predicting to do this and we actually need to do that and it helps them helps kind of guide them toward the right answer. And so that has been kind of a inspiration for how we do this. And same thing, it's like today the vast majority of interactions we want the human to spot check because we just don't want to make a mistake. Even though most of those the thing is right. So it's like they don't have to correct anything but it's worth the quick once over.

[00:29:50]Over time we we will give more to autopilot. So, like the the very first self-driving cars that I ever saw were in San Francisco in the Sunset District and it was like a fourb block radius in the sunset where it's a perfect grid and they had mapped every inch of the of the thing and they and they had no human in the car and it could drive but it was a fourb block radius and then it was like the whole sunset district and then it was the whole city of San Francisco and now it's like all the way down to down the peninsula and all over the place.

[00:30:18]>> Yeah, you can get the airport now.

[00:30:20]>> Yeah, exactly. at the airport and but if you see a Whimo on an icy road in Lake Tahoe, then there's going to be somebody sitting there holding the steering wheel because they're still training that. And so it's kind of a similar thing for us where it's like we can always push out to something that is harder than than the technology can do at a given time and have a human holding the steering wheel and we can give more and more to the models to to kind of keep taking those things on and that gives us a really nice kind of like innovation pipeline. Yeah, it's kind of interesting when you just think about balancing that. Like there's probably other cases where I don't I'm just making this up like logistics, you know, emailing co-pilot type thing or like like different industries where you kind of see there's a lot of these agents, but really should it just be like superpowered humans? Like I feel like there's like this weird almost trade-off of like people trying to make it pure software where in reality you should just be like software superpowered or supercharged human. I I don't know. I don't know where the trade-off is on that. But yeah, I mean it's like yeah the like you see it with employees in companies that are the ones who are using AI effectively are able to just be unbelievably productive to an unprecedented degree. And so it makes sense that it's like, can you build a fully autonomous AI accounting firm today? Probably not. But could you build like a suite of tools that allow good accountants to be insanely highly leveraged so they can so they can be a hundred times more productive than they would have been before because like a ton of work is happening for them, but they're able to have the like you build the tools so they can have oversight of the actions that are taking place. And so they don't have to spend any of their time on the tedious stuff, but they can but then but they still are actually like accountable for the work because they're they're fully in the loop and seeing it. And I I to me that's like a pretty exciting area in general and I wish there was like I wish there were more people doing it and again software people don't love to staff up a human workforce and so that so everyone is kind of trying to build the fully AI accounting firm instead of like the highly leveraged tech first AI accounting firm that has phenomenal accountants that that work really efficiently. Well, I feel like we almost have this graveyard of people that have kind of tried to do these in the past, like prelims that haven't worked. So, there's almost like a negative connotation, but I don't know. I mean, I feel like we're just going to keep we're going to keep trying and eventually someone will crack it.

[00:32:51]>> Yeah. My last company I used Atrium, which was Justin Khan's law firm that was and that was kind of their pitch was like we're building we're getting tier one lawyers and we're building technology to streamline them and and the company didn't work out. And I have wondered like if they had just stuck with it like they suddenly would have been able to actually streamline these lawyers to like a crazy degree and and like they didn't miss it by that much.

[00:33:16]It's like it was I think 2019 that I was using them or something like that. and a few more years maybe could have could have totally changed it because legal is definitely like one of the areas where AI has become really really valuable kind of across the stack really quickly because it's just words and the models are pretty good at words and so yeah I I do think that there's probably going to be a lot of like I mean and even how it's like there were previous attempts to build our same product 10 years ago I mean the Slack one that we talked about but even specifically like the email based scheduling X.AI AI was doing this and and so I think there are a lot of ideas that have been attempted in the past that suddenly are like super valuable today. So you think the thing you've kind of cracked just to to make sure I get this right is with the way that the kind of the how a product works. You're making a prediction on what you think needs to happen. Then the humans come in and at the end those edge cases they say they do actually follow the the prediction or they do something else. And then you look at where the deviation is and you say okay we need to figure out how do we make the models to actually match what the human actually did. And then that's kind of how you're tweaking and fine-tuning these things at the edges to keep solving these edge cases that don't quite fully automate yet.

[00:34:31]>> Yeah. But you can think about it more like a a log of work because the predictions are are there's a bunch of them. So it might be like we're querying this person's calendar and that person's calendar for these dates on this time zone and then we're creating this calendar event and this calendar event and we're updating the user's preferences with this because they said I no longer want to take meetings on Tuesday and we're replying replying all with this message to say that we've scheduled the meeting but we're also replying to the customer to say hey you told me to schedule it at this time but also you have a flight an hour later and so we just want like so just let me know if like we need to actually move it or something. So there's like all of those could be predictions that then by the time a human is in the loop looking at it they're able to kind of like evaluate was this the correct set of steps or if there's a mistake was it at the very top or toward the bottom and and is it something I can just quickly tweak or do we need to like rerun this simulation with some some new feedback and yeah >> makes sense. Yeah, it's I'm excited to see now that you guys are kind of We can talk about the launch in the sec in a second, but yeah, now that you're like you're out there, people know you exist.

[00:35:31]I'm interested to see who can borrow in in different categories.

[00:35:36]>> Yeah, totally. And we're kind of laying it out there. I mean there's tons that makes this really really hard to build and which is why we don't really mind kind of talking about it but but and yeah and I mean that is like an oversimpl simplification of it but I do really think that this architecture makes a lot of sense and and I think that I I won't be surprised if competitors of ours start to move in this direction but it's also hard. Like if you are open AI and you're trying to roll a product out to a million people on day one, then can you actually like staff up a human workforce or do you need to just focus on the use cases that you think the models can do to as like to a level of accuracy that that solve the problem on day one?

[00:36:18]>> Yeah. And so really quick just before we forget about this, I wanted to ask you the name Howie bold name. I mean easy to remember like how did you come up with the name initially when you started everything? Yeah. So, we talked about we Dave and I kind of we knew we wanted a human name because we had talked to lots of of potential customers and we knew that that like that the idea of how we kind of presenting as a human made a ton of sense and we wanted the company name to match that so that it's not like we're building two brands at once of like this company name and then and then the assistant name. And so that was very clear which is interesting because because most of our competitors have gone totally different directions with with kind of without giving these things a human names but then there's like Siri and Alexa and plenty of precedent there.

[00:37:06]We knew we wanted a name that felt friendly which is kind of ambiguous but but kind of you know it when you see it.

[00:37:13]And we wanted a name where if somebody mentions it in a conversation, then the the other person who heard it can actually spell it correctly when they go type it into Google. And we wanted two syllables. We wanted to make sure the.ai domain was available.

[00:37:29]And we also were checking for the.com to at least like we weren't going to buy the com on day one, but we wanted to make sure it wasn't like some like a domain that had been actively in use for for many years and was like would never be getable if we did decide we want it.

[00:37:44]and couple of others. I guess we were like we weren't totally sure on gender, but we did feel like a male name kind of would be cool because just because like so many of these of these virtual assistants with names have had female names. And it's like why not kind of switch it up? And as we've gone with like secretary as our branding, that also kind of helps because because some people did find the secretary thing to be kind of a call back to an era that that we don't necessarily want to go back to, but but it's like, well, how is also a male name? And also, we're building this so that you don't have to have the 1950s secretaries. And so, so a male name works really well in that in that way. And yeah, it's just memorable.

[00:38:28]We had a bunch of parameters. We were have building this list of names that kind of fit those. And then Dave sent me this meme that is about a guy named How We Do It. And it's like a screenshot of his of his Facebook profile with the with the Montel Jordan song, This Is How We Do It, playing in the background. And immediately that was at the top of our list. We're just like, "That makes so much sense." And and so we yeah got the got Howe.ai. Now we have Howie.com. And now Howie has been seen by hundreds of thousands of people and scheduled tons and tons of meetings. And so everyone in Silicon Valley knows Howie now.

[00:39:04]>> Yeah. I'm looking at is it this guy?

[00:39:06]He's almost like It's like a selfie in a truck. He's got sunglasses and like a white beater on like a black.

[00:39:11]>> How do it tattoos? Okay. I'm looking I'll throw it I'll throw it up on the screen so people can see it. We'll throw a link in the in the show notes. There's a lot of videos here. People made a lot of videos about this one.

[00:39:23]>> Yeah.

[00:39:24]>> Didn't see either. That's awesome.

[00:39:27]And then so we mentioned so you launched you guys kind of like I mean you weren't really in stealth because people were using it. You were talking about it but you did like the you know the canonical launch video which is actually pretty cool. So maybe maybe talk about what all went into that.

[00:39:42]>> We had not like announced our seed round so we wanted to kind of do that. We also had for the longest time we had we we had a website that we built a homepage that we built in like three days really early on and and we had not updated it at all and it it had like maybe 10 words on it total. It was like how email based scheduling assistant I think and a couple of other words beyond that but it was like didn't have much in terms of like social proof. We have all these customers who want to talk about how much they love Howie. It didn't have much branding that was firm in any way.

[00:40:16]It was just kind of I don't know it was it was nice enough and it worked but but there was that and then we had had a wait list for the longest time and we wanted to actually go to general availability so anyone can sign up and use Howie. And so all of those ended up being kind of the things that we launched was we announced the funding, we rolled out a new brand that we worked really hard on that that we think will kind of carry us into the next era of the company and and we rolled out general availability. We still don't have like a free trial option, so you have to pay to use the product, but because that filtering has just like always worked really well for us so far.

[00:40:51]Everyone who's ever used Howie has paid upfront to do it. And it's obviously like we could optimize more for growth over time with free trials and premium and all those things, but it's been great to filter to the people that really need this product. And and so so we've been working on the website and the brand for a while and getting ready to roll that out. And then as we were getting closer to the launch, I was like, "Okay, we probably have to do a launch video if we're going to do this."

[00:41:17]we put so much energy into the product and and the brand and the customer stories that that like a video is going to be the way that we actually make it pop. And so the launch we we were planning for September 22nd and I think it was like September 5th. We so we worked with an agency called New Kid that's in Toronto.

[00:41:36]They're awesome. They did our brand and our website and everything and they're and they're the best. And I think it was like September 5th that I texted Matt at Newkid and I was like I think we should do a video. Can you work on this with me? and he turned around like in a day this this like presentation that they did for us of like 10 different ideas for videos. Literally 10 different completely different ideas all across the board. One was like something with the we get Montel Jordan to do the this is how we do it and and a bunch of different ideas and but one of them was the return of the secretary because we had already gone with we the homepage was already saying how he's the people's secretary and so we already were leaning into that and and the reason for that is there are 10,000 assistants in the world and every AI product is called assistant or co-pilot and secretary is a word that more accurately describes what how he does because it's like administrative tasks like scheduling not not everything that an executive assistant does which could be more kind of chief of staff type level things and and it's a word that everybody in the world knows and no other company was using and so and so we could like just lean into it and own it. So, we knew we wanted to do that. And so, then the return of the secretary video made sense to just lean in harder to it. And so, by I think September 12th or 11th was the day that we signed the contract. So, like basically we were like, we want to do this concept. And so then Matt was like, okay, I know a production house and we'll talk to them and we'll see. And so they had like we had very little time. I think that normally these things take a bunch more time than this, but anyway. So they it was like they had a director in mind.

[00:43:13]the director already had an idea for an actress to to play the secretary, which was super helpful because like the casting call part of this would be like one of the most time-consuming pieces and they so they had her really quickly send like a self tape of her herself reading the lines and we thought she was fantastic and so that was the really key piece. But then and so then ultimately it was like 10 days before launch that we signed the contract to do it. And so then like three days later they shot it.

[00:43:39]They had to lock down the location and everything. They found a a great location in Toronto and then we got delivery of the final video I think like the night before the launch. So, it was all like a kind of frantic race to make it happen, but they totally delivered.

[00:43:53]Everyone delivered and and yeah, and it went really great. So, we were very happy with it.

[00:43:58]>> What was um what was the name of the the firm you said you worked for to do that though? We'll just give them a quick shout out.

[00:44:02]>> Yes, New Kid. Newkid.services.

[00:44:05]>> Newkid.services.

[00:44:07]Okay, cool. I'll throw a link in the show notes if people want to check that out. I want to maybe what we'll do, I'll do a screen share and we'll we'll watch it.

[00:44:20]>> It's time to bring back the secretary.

[00:44:23]>> What?

[00:44:24]>> Hold on. Hear me out.

[00:44:29]>> We're bringing back the secretary. Not the job title, not the dress code. None of that 1950s bull.

[00:44:35]>> Hello. But the poise, precision, >> the prestige.

[00:44:42]>> Someone else is managing my calendar.

[00:44:44]>> You get a secretary.

[00:44:45]>> Secretary.

[00:44:46]>> No, I don't have a pen. I have a pencil.

[00:44:48]No more triple bookings or battling for bio breaks. We're not scheduling meetings to schedule meetings about meetings that are soon to be scheduled.

[00:44:55]Hi, I'm going about the meeting for the meeting. Thanks for setting the meeting.

[00:44:59]>> It's a secretary for all of us. Yeah, even secretaries.

[00:45:05]Meet Howie. The return of the secretary.

[00:45:08]Things got a little crazy back there, didn't they?

[00:45:13]>> I I loved it cuz it's just so I feel like a lot of the launch videos you kind of see right now, they're like I don't know how how would you describe the current like meta of launch videos?

[00:45:26]>> I don't know. I mean, there's been a lot of different ones and like I think that we wanted something that felt tied to the brand and to the product. We didn't we didn't feel like it needed to be like a whole kind of like product rundown because that is like one category of product launch videos is just like you talk about the product and we felt like we want people to come check out the homepage and so it's okay if they're left with questions about okay I just that video was funny but what is this and and so we wanted it to be connected to the brand and to the product experience but we didn't feel like we needed to sell the features in the video and and I think we did a good job of that. I mean it in terms of like we got we got about 1.2 million views on the video across the different posts that we had and and it it converted to a ton of customers and a ton of revenue. So, we were like really happy with that conversion. Like it would have been great if it got 10 million views, but in terms of like how did it like it blew our expectations out of the water in terms of customer growth and so that was great. And so I think it did a good job of teasing it and then and it was just interesting enough that you go check out the website and then you get to see like tons of customer testimonials and and we talk about the product more specifically in lots of ways there and have a brand that we're really proud of there. And so yeah, so I I think that I don't know the other videos. I'm trying to think of other ones. I mean, Clearly obviously is like completely crushing it and I love their their videos. I I don't know how well they do in terms of converting, but I think that I'm guessing that they probably do pretty well. And then there's like the friend launch, which like is mega viral, like on a scale beyond anything that like Yeah. at least an order of magnitude bigger than than ours. Does it sell the product? I'm not totally sure. And so that was really important to us is that we actually like focus it on the business and not like just getting the views for the sake of the views. And what else? I don't know. I mean, we we introduced a character. Some people do these like high production videos, but it's like kind of just the founder talking about their thing. But we felt like it was cool to introduce a character who we now might end up using as kind of a recurring character in some other videos potentially. And so that idea was fun. And I don't know. Yeah, I think that like for what what I was most happy with is that it was a very short turnaround time and we produced something that that like feels really really polished. that one of the kind of they had a few different inspirations, but one of them was like the the like I don't know what you call them, but like the the cutaway videos in the big short where they would do like the like a story a short like anecdote and yeah and I can't remember what they are specifically, but they're like these quick quick cuts kind of and and cool camera movements and and the idea was just like super fast pace and really high velocity. And the actress that we used was was just so good. She like her energy when you what I was envisioning from the script and all the ideas just was way less energy than what she ended up ended up like bringing to the table.

[00:48:35]So it's like I just the energy was just phenomenal. And so anyway, I'm really happy with it and I'm happy that it converted to customers because like the obviously if it just flopped altogether that would be the worst thing but but also if it like went viral but didn't convert to customers then it's like was it worth the energy we put into it? And so we're just thrilled that it like led to a huge increase in our customer base and tons of new people who now in the couple weeks since have been giving us tons of feedback and loving the product and and the the signups are still rolling in. So that's been really cool.

[00:49:07]>> Yeah, it's kind of interesting in the sense of like everyone kind of knows what a secretary is or like an executive assistant. So you don't necessarily like to your point about how some people like demo the product. Like if you just say like, "Hey, it's a secretary. Like it's an executive assistant." You're just like, "Oh yeah, I know what that is."

[00:49:24]And then I feel like the biggest question is, "Okay, but it's AI. Like does it actually work?" Right? Like I've used all these tools. They're all kind of sus. So the so it's like go to the website and it's like, "Oh, there's just a bunch of customer testimonials about how how well it works." So it's like, "All right, cool. I'll try it out." So, it's like an interesting like I think it's almost like you need to position your marketing depending on how mature a category is or like how well the buyer understands and then like what their questions probably are based on like you've given them some general awareness like all right what's their next step in terms of discovery is it like do more research is it just like convert immediately is it like they need some social proof do they need like know the price etc so >> yep totally yeah I I think that like the Social proof is just I mean for anything I think social proof is really key and we are excited that we have so many awesome people that I mean we had like way more than we even could include customers who just were like thrilled to provide testimonials. So that's been really awesome and I think that yeah just it's like ultimately the our job is to just get people interested enough that they have an idea what it is and then the customer testimonials do the heavy lifting of like okay if it works for them then then it's going to work for me.

[00:50:39]Yeah, I feel like social proof is very it's like underrated but also like fairly rated. Like I think a lot of people like you kind of realize how important it is just to like have a couple other people tell you just to like get you across the finish line really with like anything whether it's like buying a product or even like listening to a podcast if like you see three people say like oh that episode with Austin Turner talking about how it was so good you got to listen. someone will be like, "All right, fine. I don't listen to podcasts, but I'll give this one a try."

[00:51:08]>> Yeah, totally. I mean, I think the reason that you don't see a lot of social proof in in many launches is is not that they like don't think it's useful. It's that it's that it's really hard to like have customers that are that like love your product so much that they're out there amplifying. And that's like what made us so thrilled about the launch was that the amplification more than our investors, it was our customers who like came out and were in the comments. like the LinkedIn especially the comments in the LinkedIn posts that we had are just it's like just comment after comment of people talking about how much they love the product and and so that was just like super exciting to us that it felt like there was real substance to the launch of of like we are actually making these customers very happy. We're making people's lives easier because they don't have to deal with something they had to deal with before and and they are like proudly shouting from the rooftops about that.

[00:51:54]So we were really thrilled about that part. And so when you when you do a launch like that, like I remember I maybe I'm in like a a special group chat with you where you're like you're showing us the like I remember actually I think I was at Disney World throughout the when you're talking about filming like how fast it went. Like I think you sent me a draft and like watched it like two days later but it was like finished or something like that because I was at at Disney World. But I was like a little bit disconnected.

[00:52:19]And then like do you so when you're launching a product like that like do you involve the customers like hey we just launched like tell us what you think of it or do you just put out there in the ether and see what happens like is there a good way to guide those and make sure they go a little bit better.

[00:52:33]>> There was some so well customers who had testimonials obviously I was like sending those to like to say hey like would we we'd love to do a testimonial about with you with you. can you give us some content for it or are you interested whatever? And so those folks definitely got like a a peek at some of the some of the stuff and then some other customers who are like founders that I think are great at marketing. I was just sending them the kind of like work in progress website. The video, very few people saw the video. You saw it on the group text and and a handful of other people, but not many people saw the video partly because it's like the we were pretty locked in on the direction. So when we were like riffing on the script and the the kind of concept I did I was sending it to a bunch of different friends and and getting their thoughts and but then like once we were kind of locked in on it was like the types of feedback you don't want someone to be like well what if she says this instead of that and it's like oh we already shot it so we can we're just like we're at the editing phase like do you have feedback on the color correction or something? So, so like that was I mean we did send it to a few people though like the one of the early cuts was like a little bit long and we got that feedback from some people and so we we kind of cut it and so there was some there but but the brand and the web page was definitely like a ton and like Sophia Amaroso I have she like she provided a ton of feedback as we were building this brand and um so just like an ongoing kind of text thread where I was sending her some of the early web designs and things. And cuz with the with the brand, we also kind of went in a pretty like novel direction, which is that it's very people first. So, we did a photo shoot with actual models that are like kind of laced across the entire homepage. And we felt like everybody in SAS and productivity just has these like dark mode purple gradient homepages that are all they're doing is showing you like screenshots or or like kind of illustrations of features and it's just like all just like features and we felt like that the experience we hear from customers is that they love Howie because it is like it engages like a human. They like that there are humans in the loop because they can trust it more because there's humans and they like that it is sol it is helping them with connecting with humans and whether it's like customers or investors and there's just like so much kind of humanness to all of it and we felt like why don't we then make the brand feel human and so the like the some of the folks that we sent the really brand idea to people they were like this looks like a makeup brand or something.

[00:55:07]>> It it kind of does. Yeah.

[00:55:08]>> Yeah. It to it totally does. But we're like, "Yeah, that's fine." Like, who does like we think our customers are awesome and we want to we want to showcase like what we the vibe that we feel like our customers have, which is just these these great people who are are real people and who are awesome. And we and we felt like photography was a much better way of doing that than than like some illustrated an like figures or or characters or whatever or or just like screenshots of of like how you're writing an email. And >> yeah, like if if if you did show screenshots of the product, it's like, oh cool, a screenshot of superhuman, a screenshot of Gmail, like it's like the meeting has been scheduled. It's like it's a little bit harder to show much. I guess the original version of the website though I do remember you had like a calendar and like lots of blocks and you like kind of played around with like animations like as you scrolled of like how like it was kind of like scheduling but again it's like you know I feel like the to your point it's like a makeup brand like it does feel a little sexier in a way.

[00:56:08]>> Yeah it just feels elevated and that's like the thing is that people feel like a baller when they use Howie because it's like they used to have to do this thing themselves and someone's doing it for them now and so it's accessible.

[00:56:19]It's not like it's like a high-end luxury good that nobody that is inaccessible, but it's also kind of baller. And actually, one of the analogies that that I used with with UKID, the agency early on was Uber, like early days Uber, because it was like at the time you had yellow taxis, which were like dirty. They probably wouldn't show up if you called them, but they might. And they and like they're just like all these things that are that are like not great about that experience.

[00:56:45]And then there was like the a chauffeur, a private chauffeur, which was like this baller experience that almost nobody has, but the people who do just it's amazing. And then Uber came in as like they they positioned the brand pretty really like kind of luxury and high-end, but it was also accessible. And it was called the everyone's private driver and was their like slogan early on and and so it was like taking this luxury out of reach thing, making it accessible but still kind of baller. And that was like the kind one of the analogies that I had early on of like what I wanted Howie to kind of stand for is like the executive assistant in this luxury thing that's that's inaccessible for most of us. Um, but we all know it's pretty baller. And then the alternative is Calendarly, which which like is a meme basically.

[00:57:27]How how like kind of inefficient and and like it it is the taxi and and we are pulling that baller experience forward in a way where where it's accessible to everybody. And that doesn't mean we need to like cheapen it and make it feel more like a taxi. You can still be baller and elite and and have really polished branding and and feel really kind of luxury, but it's also accessible.

[00:57:56]>> Yeah. Two two things that the way you said it reminds me of is like when you get introduced to someone and there's like, you know, they add their EA, some people, you know, you might be get like annoyed like, oh, this person doesn't have time for me or whatever. wi with Howie like that jumps on pretty quickly and it's like responds fast but also like there's almost like the power element of like let me add my EA like I'm so like you there's almost this like you made it in life if you have an EA or like a secretary. So it's like you you're like I I kind of hate I kind of hate the word democratize but you're like democratizing and just allowing more people to have an EA. It's kind of like in between of like sending someone a link and then actually having like a human assistant that you know you're paying, you know, pro possibly six figures all in of comp of taxes, social security, health insurance, salary, etc. Um, and it's it's interesting. I remember when I first met you when was investing, I was like, I don't know, man. Calendarly, like a better Calendarly, like is this is this really an opportunity? But I I think in 2023 County did like 270 million in ARR or something like that. It's probably a lot higher now. And this is almost like a little bit of a more fully baked version. So like the opportunity is actually a lot bigger maybe than someone would think just from initial initial glance.

[00:59:14]>> Yeah. I mean to me it's like even just that like talent is a great business and and it's awesome and it solves one thing really well which is like find a time for a meeting but when you talk to executive assistants about the calendar tasks that they do find a time for a meeting is like one of a litany of other things and and so the scale of a business that just solves the calendar entirely for millions of professionals I think can be at least an order of magnitude bigger than Calendarly. And then and then it's like have we built the framework and the brand and the trust and the all these things that allow us to go from just like we are just a scheduling assistant to an assistant that does many many things for professionals. And we aren't in a hurry to go from one thing to the next because because again it's like nailing any problem that we we purport to solve we want to truly solve to to a fantastic degree. And so we're not in a rush to layer on like features for the sake of features, but but over time there's no reason for us to be constrained to scheduling and calendar over the long term.

[01:00:24]>> Yeah. Well, to the point of time, it's like time is money, right?

[01:00:28]Like that's the ramp slogan. Save >> time is money, save both. I love that.

[01:00:32]>> There's probably like some some play on words with how like time is money.

[01:00:37]Sa save I you could say save both with Howie. I don't know. Obviously, you can't you can't you can't steal the steal the slogan. Exactly.

[01:00:45]>> That's like one of my favorite lines of copyrightiting I've encountered in many years. It's like it's like time is money is just is colloquialism. So, it's not they're not like they're not breaking new ground there, but it's like and just say both adding two words to It's like just so beautifully simple and resonant and and like every marketing slo or many marketing slogans over many decades have tried to sell save time or save money or save both of them and it's amazing that they that they found a tagline with like with that that that just in hindsight feels like it was just sitting there to be grabbed by anybody.

[01:01:22]>> Yeah. So you kind of mentioned you there's probably opportunities to expand over time. Like what does this look like? like what does how we look like over the long term? Like I'm talking 10 years, 20 years in the future. I don't know. I don't I don't know how far out you think about this stuff, but what do you think the big kind of opportunity is and what the product might look like?

[01:01:41]>> We try to focus most of our energy on what is what is right in front of us and we and so it's like we think a lot about like if we achieve what is right in front of us, which doors does that open and where might those lead? But we try not to obsess too much because I like think it's really common in startups to be way too focused on like what your company is going to look like when it's doing a hundred billion dollars in revenue and you haven't actually built something that people want to use for $10. And and so we do really try to focus on like what is right in front of us and how can we better serve the customers we have and and have end the day with more customers than we started with every day. And like that that's really like where most of our energy is.

[01:02:19]But we do think about like where does this go? And I think there are different paths like you could see it being a kind of single player use case where it's like all of our customers are individuals using Howie and it spans industries and and like consumer use cases then professional use cases but it's just this like super assistant for for the individual. Or you could see it being this like calendaring scheduling product that's sold into 10,000 person organizations where it's like the assistant is a part of it. And then there's also all this architecture around like helping a company know where time is being spent and which meetings are happening and which aren't happening and how many people are in the different meetings and and lots of different things around just like making creating efficiency in a a large team with the calendar. I don't know really at this point like we I really think the technology we're building can can work really well in both of those paths and there probably are 10 other paths that I that I haven't thought about. I think we're excited about both. We definitely are excited about like the even without having built many team features, we're still seeing a lot of adoption among teams like ramp since we were just talking about them. Ramps recruiting team uses Howie and and that has been really awesome because that's like one of our first like kind of larger teams that's that's using it. But we're seeing organizations where we grow from one person to three people to seven people to 10 people and and we haven't built much functionality that like kind of incentivizes that. But we're but we are starting to and so so over time I think the main thing that is like a clear northstar is we think that professionals of all sorts spend a ton of time on meta work which which is just like work about work. So it could be scheduling it could be like setting reminders to follow up on something. It could be it could be like reviewing notes from your calls or anything like that. There's just like any a ton of things that all of us spend time on that is meta work and and there's also a lot of meta work that like lives in our brain at all times. So you we're constantly thinking like I have to follow up on that with that prospect and do I have enough time after this call to get to the next to to get to my like coffee meeting this afternoon and do I like I don't know did I I got that customer support email over the weekend and I think I've like totally dropped the ball in responding to that and there's like all these things that just live in our head and and if how we can just be plucking those things out of people's brains so that they can have the confidence that the thing's not going to get forgotten and is going to get taken care of then that's creating.

[01:04:49]So we're more interested in solving the meta work than like building an assistant that does the actual like your actual job for you if that makes sense.

[01:04:59]And anything that you could classify as metawwork is probably something that we are intending to solve for professionals over time.

[01:05:07]>> Okay. Yeah, that's something to look forward to. I guess there's a lot a lot you can do as a customer. There's like a lot of things that how we could probably do for you.

[01:05:15]>> Yes, totally. And like it's it'll start with like just breaking out of email. So we're building Howie for SMS right now, which I'm really excited about. And and we'll be like just a much quicker way for people to interact with with the product. And we eventually will have Microsoft Outlook support hopefully by the end of this year. But but we'll see.

[01:05:34]We've kind of like there's there's enough demand in the Google ecosystem that we haven't been in a huge rush, but we also get emails like literally 10 times a day asking us for Microsoft Outlook support. So, it will unlock a lot of potential new customers and and so that's really valuable. But >> I mean Outlook is enterprise like once when you get Outlook, you're going to start signing those like thousand seat 10,000 seat deals. So, that might be valuable.

[01:06:00]>> Yeah, totally. I mean, there's tons of market opportunity there. We don't really have like a good offering for 10,000 people to use Howie today, but but as we go like yeah there I mean we get interest from any any number of like company size industry like people are reaching out and we just have had to kind of like keep the blinders on of like we serve people who need to book a lot of external meetings and have have a lot of pain in doing that and that's like just limiting because we we want to serve many many more people than that over time but but it's allowed to have like a ton of focus on just better serving those people and understanding that that kind of need and just nailing it. So that's where we are now. But yeah, but we'll add Microsoft for sure soon.

[01:06:46]>> Yeah. And actually I meant to ask this earlier, but I mean you mentioned we're growing 30% 40% a month. Like what's the current state of the business? Is there like a public or even like a private?

[01:06:58]like you're only saying it on this podcast, but like what's like currently kind of like the size of things, kind of like the the state of everything?

[01:07:04]>> We went into the launch with over a thousand paying customers and we have grown our customer base a lot since that launch. So, that's really exciting.

[01:07:12]And what else can I talk about? Oh, how he's booking well over 5,000 meetings a week for our customers. So, that's a pretty exciting metric. And and he's interacting with tens of thousands of people every month on on email. you see it. Any question updates? We just have been growing.

[01:07:29]>> Yeah, that's fair. Well, I mean, I don't I don't want to be the one to say. I want you to be able to say what >> I know. Yeah, I think I think those are the main like kind of public numbers right now.

[01:07:39]>> Okay. So, actually going back to like the early days, you kind of like you talked about 20 2015 the Slack bot or like the Slack channel. You kind of skipped ahead like you know eight nine years later, but there's a period in the middle. I think one of the most interesting things you did was you worked for Jason Calccanis and you said people are always curious what was that like and I am also curious what was that like working for Jason.

[01:08:04]>> Yeah. Everyone everyone always wants to know. I very regularly meet people who immediately when they learn that I worked for Jason they want to know.

[01:08:12]>> He's good at what he does obviously.

[01:08:14]Like it's not like an accident.

[01:08:16]>> Oh yeah. So he's a really close friend.

[01:08:17]I was gonna be like he's a No. Um he Jason so Jason and I definitely like would have pretty intense arguments and things and I think what I really respect about Jason is that there are a lot of people like so our entire industry lives on Twitter and I meet so I meet a lot of people who I've gotten to know their Twitter personality over time and then I meet them in person and it's pretty common to feel like a disconnect then where you realize like there was this fully kind of manicured persona that somebody created around themselves And what I respect about Jason is that that he he literally like lays himself out there and and is actually like he is actually like the person that you see online which and so you might love him, you might hate him, but like that it is it is I respect the kind of ability to to actually be yourself online because there's a lot of the algorithms and the kind of social norms and all these things pull us to try to like present a manicured version of ourselves. and and so it's actually pretty hard to just genuinely like be yourself out there and especially if being yourself is a polarizing person that's going to piss a bunch of people off and so and so I really respect that and I definitely learned a ton from Jason about about investing and and like a lot about him kind of so he when back back then this is like 2016 he was like studying constantly like he would be he I would get in his car and he and he would like sit there and make me watch like 30 minutes of Charlie Rose interviewing someone or Oprah interviewing someone or Howard Stern interviewing someone and he was like literally like studying the tapes of these interviews. He'd watch the same interviews over and over and over again and I and and at the time I was like do you really learn anything from this? And but and now I can see like once they once the All-In podcast came out and he had had to like serve as the this pretty challenging role as moderator for a bunch of really powerful personalities that are that have like all want to hear themselves talk and he and including himself also and and so like I can see that there that that work that he was putting in for like years and years and years and studying the art of moderating conversations and interviewing people is like actually working and and it helps that he doesn't mind being the kind of punching bag in in a lot of the conversations. So So that's all cool. And yeah, I mean it was it was interesting. He definitely like as you would expect if you if you've followed him online like and be brash and so we definitely had like some some pretty heated arguments where we disagree about things because both of us are pretty passionate people. And so there's some funny experience or memories there. But like overall, I think that he it was fun to work for him and and like he's been the first investor in two of my companies now. And so the like he's definitely a person who has deep loyalty and and I've felt that and and and so like it's cool that from the day first day that I ever met him I have felt like he truly like has my back and and and so like as much as he could be a dick or he's polarizing says all sorts of things on mine that I that I like may or may not agree with or that I'm just like why are you saying this? And so there's all of those things, but at the end of the day, it's like that it's you don't often find somebody who who you feel like actual real loyalty from and to and and so yeah. So he he is a good friend >> and he did I mean did you learn a lot about media from him because he really runs like pretty interesting really like a media company?

[01:12:06]>> Yeah, totally. It really is more of a media company than anything and I Yeah, a ton and like he and he's been doing it forever. I mean, he was like like talking print magazines in New York in Silicon Valley something. I can't remember what it was called, right? Like late 90s. He's been like doing this forever and he he knows everybody and he's seen so much. And so that is like one of the things is that I started to learn is I would often think he didn't know what he was talking about and then and sometimes realize that I was right but more times realize like oh actually he's like really done this and he knows and he knows what he's doing and yeah so like the with media specifically like I really think he has a good grasp on how to do it and and I mean even just he was investing in video content for his podcast at a time when nobody everyone was just doing audio only and like he would have cameras for every do every single podcast interview in person, never remote and never audio only and and that was just like ahead of the curve. and now he has a back catalog going back 16 years or something of these video podcasts and and so yeah, all sorts of things like that where it felt like he had a pretty good grasp of like what where where the media world was going and what was kind of resonating and then also had a willingness to just kind of keep going forward because like there's you see a lot of people kind of try to start a podcast or start some sort of form of media and then just like when it doesn't become the next big thing in three weeks, then they just kind of like fizzle out and abandon it. And there's a lot of value in just like continuing to do the next episode, the next episode, two episodes a week. You just keep going and and that over time like really does compound. And so the determination is definitely a big one, too.

[01:13:53]>> Yeah, it's hard. I mean, I've been doing my podcast for like two and a half years at this point. It's It's almost like unpredictable. Like I'll be like, "Oh, this one's going to do so good." and then I'll post and it's like the worst one of the year or it'll be like one like R. I mean honestly when I go back and look at the data like I did I could not have predicted what the better episodes would have been. It's like with all content like even with Twitter like some of my most viral tweets that got like 10 million views there's like a typo because I like posted on my phone really quick and it did like auto correct. I didn't realize it. And then there's some where like, you know, you spend like an hour on it and it gets it gets like 1% of, you know, whatever what you thought it was going to get. So it's like a lot of it's just like a volume.

[01:14:36]Like it's just shots on goal. Like it's just staying in the game, keep putting stuff out there. I think that's just generally the thing with like really like media or like the internet or content on the distribution side is like a lot of these algorithms the way they work like you can there's like almost like unbounded upside like right like there's like a 100 million people that might potentially see the content and it's just like figuring out like how do you get in front of all those people but you never know what's going to work and so if you're only posting once say you post once a month right versus if you post once a week once a Hey, once an hour just like the more you're putting things, the more shots on goal you have, the more surface area like chances that you'll people see your stuff. I guess you got to balance like not being annoying also.

[01:15:23]>> Yeah. I mean, that is tough.

[01:15:25]>> Yeah. It's like just post about what you're doing. Tweet about what you're doing. Put content out there that's like natural to you, that's easy, that like kind of fits in with like what you're good at and like what your unique advantage is. And I don't know, I feel like not enough people think about it that way.

[01:15:38]>> Yeah, totally. And it's I mean it applies to like anything building a product and building a business. It's like that you can get so distracted in where you want to be or in like how difficult the challenges in front of you are. And for us like every day of building this company has been just like can we get one rung higher on this ladder? I mean we chose a problem that's like definable of like scheduling. And so it's like can we be early on it was the only thing was like can we be better at scheduling today than we were yesterday. Now we have a business, we have customers and all these other things but and so it's still it's like the same thing I said before. Can how can we end today with more customers than yesterday and and end today with better solving the problems that our current customers have today than yesterday. And and it's like just doing that then the compounding kind of starts to happen. And so it can really apply to anything. I think that the idea of just like you you just need to move forward a little bit every day.

[01:16:36]>> Yeah. So, speaking about compounding, I know you've had just like pretty steady continuing to grow product kind of it's interesting when you it's like you think about how someone maybe discovers Howie.

[01:16:46]It's probably in their email inbox. Like that's maybe the place they discover it.

[01:16:51]Do you remember how did you like first get people to start using it? Was it just you you know a lot of people and you're like, "Hey, I'm building this thing. Check it out." Kind of took off from there. What was kind of the process?

[01:17:01]Yeah, the first so we like put up a weight list early on and people started to sign up and like I had it in my Twitter bio and there were a few people early on who started using it and then like tweeted about it and and that would lead to like because at the time the product was not very good but it was good as if you like there was at least like a 50% chance that your first three interactions would be fantastic and so then that people would lead to people tweeting about it and then and then eventually they would like run into the issues and and so but it would like so that was enough to validate the demand and then the weight list kind of started to grow and so then we would like kind of look at who's on the wait list and and go reach out to people and we have a couple of the very first customers we ever onboarded are still using the product today which is remarkable because they really like rode through the I mean some of them who are friends that like that that's still remarkable. Um like Alex Cohen for example used it all the way through and was not like dissatisfied with the old product, but then once we rebuilt it was just like this is mind-blowingly better and like I can use it for everything now, but there's some folks that I don't even know who were in the early cohorts who have continued using it and but yeah, the word of mouth definitely is a huge aspect of it. So it's like people see Howie in the inbox. We don't makey's emails an ad for Howie. We try to be really subtle in that regard. So it's not like there's a link to the website or anything in those emails. And so it is really subtle, but then also just people that are having a great experience. The reason the word of mouth growth really kicked in in January of this year is that the product experience became something that people just really want to talk about. And so that's a huge part of it. And then yeah, there's been some like viral moments and things along along the way, but again, it's like it's just this steady process of like we like we we went into September with a thousand customers and we want to end September with more customers than that and now we're in October with with the number of customers that we have and we want we want to end that this month with even more. And so it's like we can try to manufacture viral moments or we could do like a marketing campaign or whatever else we want to do, but ultimately it's just it's like not trying to catch lightning in a bottle. It's just like how can we reach a few more people today?

[01:19:22]>> But you you were telling me the other day about you have this really interesting strategy of doing your pitch deck that kind of unique kind of like simplifies the process, but you think it works pretty well. What's what's your strategy for making a deck? Yeah. So, I've looked at a lot of like docs send metrics. Is there doc send when people send a pitch deck? I I suggest sending a pitch deck via PDF, not via docsend link. But I've looked at a lot of docend metrics and gotten to glean an a really interesting fact that I haven't heard a lot of people talk about, which is that the vast majority of people you send your pitch deck to are going to click through all the way to the end. They're going to make it to the end. But the vast majority of people who you send your pitch deck to are never going to spend more than 3 seconds on a single slide. So if you have a 100 slides or if you have 10 slides, they're going to go three seconds per slide all the way through to the end. And so what you can do with that information is not optimize for how do I get less slides, but optimize for how do I get less information on a given slide because you know you get three seconds whether it's your your 20th slide in the deck or your third slide in the deck.

[01:20:27]you're going to get 3 seconds to of someone's attention. And so more slides with less information is just by far the best thing to do. And so basically the strategy that I have is write it like a thread of tweets. So you you create bullet points. You should have not even 140 characters of words per slide, like eight words per slide or something. And but the shortest possible thing have a narrative arc that pulls people through.

[01:20:50]So, I don't love the kind of like standard templates of problem, solution, whatever. Like, you need to get you need to hit all of those things, but you should have a story that gets somebody like to to want to go to the next one.

[01:21:03]And then and then again, put an amount of information that can actually be digested in 3 seconds because even your friends that you send it to for feedback are going to scroll through 3 seconds per slide. and someone who's serious is going to later maybe go spend more time and look at like and really like read through each slide slowly. But but to even get the chance for the slow one, you have to have passed the 3 second test. And so so don't worry about like some I've seen some people recommend like make sure you have less than 10 slides because you don't have much of someone's time. And they're correct that you don't have much of their time, but they're incorrect that you should jam like a ton of content into 10 slides.

[01:21:39]you should just stretch it out further because they're going to give you 3 seconds for every slide and then they're going to keep paging through. And so, so that's my advice is create a slide deck that has very very very short bits of content, has some narrative arc that pulls me through to the next to the next slide and and get all the points that you want across, but but never try to try to make like more than one single nugget of information fit onto a single slide.

[01:22:06]So I was just thinking I was like I need a problem slide, I need a solution slide, I need a market slide. Like the problem slide might be five slides and it's just one sentence with like a graphic or whatever for each almost like phase or setup of the problem to get you to by the fifth site it's like holy I get the problem.

[01:22:25]>> Yep. 100%. And the next set is like, yeah, it might be like a couple couple screenshots of the the way you solve the problem or like sentences on how you solve the problem and like however you need to define the market opportunity.

[01:22:37]Like maybe it's a couple slides and again it's like just really because to your point people don't read like a lot of the times people are like there'll be like tons of text everywhere and you just read like the header or like what's the most bolded image or like thing that stands out. Nobody is reading a paragraph of text on the deck that you send. And people spend a lot of time on those paragraphs and every word of them and then nobody reads it. So instead, figure out what you can do in three words and in four words and and get that whole paragraph, but just stretch it across more slides in punchy short sentences that that actually tell a story and then people will actually read it and otherwise like they're just going to keep moving. So even three seconds might be generous. It's you're not going to spend a lot of time per slide. So, so you want to get the least amount of information possible and then obviously you want to only include the most important information and you want it you want it to feel like a teaser. You want someone to leave afterwards wanting more and so you got to kind of cut it before they're fully satisfied and you want them to have questions because then they have to take the call to answer the questions. There's a lot of things beyond that but it's like just never try to fit fit more than like more than 0.2 ideas per slide or something like that.

[01:23:48]>> Yeah, that makes sense. Oh yeah, it's a good it's a good strategy to think about. And you you've kind of mentioned before in the past when just we've been talking that like people tend to optimize for growth too soon.

[01:24:03]I mean I guess there's like different things you can mean by that, but but I guess how like why do you think that's the case? Because don't you want growth?

[01:24:10]You want ARR to go up super fast and like you get a higher valuation, raise money, right? Like isn't that good?

[01:24:17]>> Yeah. So, I do think it's good, but I think if if what you're trying to build is durable value, there there are many proven ways to kind of growth hack a product and to optimize your push notifications to get people back to it and to do all the things that make the metrics look the way that you want them to look. and we and I've done all those things in the past and and I've experienced a product building a product that has all the growth levers kind of optimized and tuned and it's still not growing and that really sucks because then it's like okay we tried all the things and and still nobody cares with this product.

[01:24:54]I don't know if it's the nature of it being in the inbox. I think that's part of it. And then maybe the clarity that we've had since day one about the problem we're solving to where we never had time to work on growth hacks because instead we just have had to work on like fixing the problems with the product.

[01:25:10]And that continues to kind of be the case today. And so we've never had time to build out like you we literally don't even send emails. You sign up as a new customer. We don't send you emails to remind you to use the product or teach you how to use it. you get one email when you first sign up and then and we're going to add drip campaigns that that go out and we and we're going to add more like kind of user education like here's how VCs use howie here's how founders use howie and all that type of documentation and stuff which like admittedly it's it's not ideal that we don't have those things but the thing that is ideal as a result of that is like we've just lived in the ground truth of is this product useful so you sign up for howi you don't ever need to come back to our website again. So, you use the product by going into your email client and remembering to type in Howie's email address and we don't send you a push notification to to encourage you to do that. We don't like there's no bookmark that you save in your browser.

[01:26:04]There's, you know, you don't get how he doesn't just like email you. Actually, now he does do email you proactively for like conflicts and things. So, we've we've I guess add some of that proactive stuff, but that even six months ago didn't exist. And and so there is like I guess some proactive things but for for a while there was none. And so the retention, the word of mouth growth, the usage numbers and all of those things looking so great was really kind of validating for us because we're like we we are in the ground truth. There's no we're not tricking people into coming back and and we haven't optimized. We are not pulling any of the leveries that we could pull over time. Over time, you want to for sure, but if you genuinely want to be intellectually honest with yourself about is my product valuable to customers, then peeling away all of the growth hacks is a good way to actually do that if you're willing to. And but that it it's hard because it's like you you want the numbers to go up as fast as possible and you want every you want it to look like you've solved the problem, not that you are solving the problem.

[01:27:04]But we have treated it as an intellectual experiment from day one of can we make this product useful to people and that for us really really helped us get to a product that people actually want to use because because there was no smoke and mirrors in the metrics.

[01:27:21]>> So it's almost like just don't even focus on growth at all and if the product is growing that's probably a good sign.

[01:27:28]>> Yeah. I mean the best products in the world grow themselves. Yeah. So, it's like why why should you have to like spend a bunch of time on like the sharing functionality or the referrals or why should like why should you even have a referral program for a product that's like in the early stages pre-product market fit and why why should you spend any time on that if the product is not so good that it's grow that it's growing by word of mouth and so if you can just every day try to get closer to a product that people can't stop talking about then your time is much better spent I think.

[01:28:00]>> Yeah. And I guess to the way that the product works, it's like a it's sort of like a multiplayer product because in order to use it, there's somebody else there that sees it. So that probably helps a little bit on the grow side is just generally speaking. I'm sure you get people that sign up is like, "Oh, what's this howy thing like how.ai or howie.com I don't know what URL or someone might obviously use the custom URL." So maybe maybe they don't even see it.

[01:28:24]>> Yeah. Yeah. one I mean like sometimes people don't know one that I saw the other day that was hilarious is a white label howie so the person thought it was a human and then but the person went through the whole scheduling experience and I think they had had to reschedule the meeting so there was like some tricky element of it and and the person was really impressed and they sent an email to the howy customer and to the white little Howie saying do you have any friends like I really need someone like this to do my scheduling so they like thought it was a human and were trying to like hire the friend and then so Then the customer was able to just be like, "Oh, just go to howi.com. It's 35 bucks a month and you and you can have like literally the same assistant. You don't need one of his friends." And >> no way.

[01:29:04]>> Yeah. So like that it does happen for sure sometimes through like direct word of mouth or everyone's just kind of always talking lately about like what AI tools are you using and so if people like I hear these stories of like we're we were I was at this coffee meeting and my friend asked me what like what AI tools I'm using and I said howie and they said wait like the thing they used to schedule this coffee meeting and and I said yeah and so like it happens that way where less than like again we don't put like a link to Howie in the emails that Howie sends and so there's a lot of ways We could be more aggressive in doing that, but we care about the product experience over everything. And so, so we just don't want to dilute the product experience for the sake of a growth hack.

[01:29:46]>> Yeah, that's always good. And when I first met you, you were living in Bend, Oregon. I think you you're you're in Seattle now. What's the significance of living there? Like, shouldn't you be in San Francisco? You're building an AI company?

[01:30:01]>> Yeah. So, I had not spent a ton of time in Seattle before we started the company. Dave was living here and had moved here in CO. Both of us lived in SF for like 10 years. And it does definitely help that we have like a deep networks down there.

[01:30:16]And we didn't know at the start for sure what we were going to do like build the team remote. Like we lived in different states at the time, so we the two of us were remote. But over time, we started to feel like it made sense to set up an office in Seattle. And so we did that and we started to kind of grow the team exclusively in Seattle. I was traveling here every other week and and so then eventually it made sense for me and the family to move here which we did this summer. Um the reason that I think Seattle is awesome for us is SF is ma amazing obviously there and there's a more vibrant startup community in SF than anywhere else in the world and and you just like can't compete with that and there's more tech talent especially tech talent that's interested in startups. There's more in SF than anywhere else. But the challenge that friends that that of mine are having is there are the coolest companies in the world are all in SF. So every every three months YC has a new batch of 200 companies that all raise three to five million dollars and they're all competing to hire engineers within a seven square mile block of California.

[01:31:17]And there's only so many engineers for those people to hire. And three months later there's another batch of 200 companies. And and then in addition to that, there are all the coolest AI startups that you can imagine from seed stage to series A to series B to series C to series D to public companies and everything in between. And and so the competition for great talent is just insane in SF. So friends that I have who are in SF and are excited to build their team there and be fully in the office have then ended up building a remote team because they literally can't find anybody. and and so Seattle is interesting because there's a ton of talent. There's a lot of AI talent because AI has been happening here at Microsoft and Amazon for a long time um before it was it was kind of cool and because of those companies also there's a ton of tech talent there's less engineers that are excited about startups. There's a lot of like we got an applicant from an engineer a few weeks ago who has is at Microsoft and has been at Microsoft for 31 years and we were like okay you're not going to be a great fit for our culture here if you were at Microsoft for 31 years >> who knows they might be hey they they'll stick around for 31 years of Howie that could be good >> that's true that's loyalty yeah so may I don't know maybe and but anyway I think that like there's there there just aren't nearly as many cool AI startups and so we we were able to kind of stand out in in that regard and and there's still plenty of tech talent and then like most of our investors are are down in SF but but we travel down there and and know lots of those folks already and and so for us it's like it's working really well and and it's fun to be kind of building in a place where we feel like we can really like be kind of on the map especially as we grow. Yeah, makes sense. One other question I wanted to ask you, but I know you do I I don't think this is not like your full-time thing, but you do some angel investing.

[01:33:16]You actually work with your brother. You guys have a small fund. It's kind of like an angel fund. What your you have a pretty interesting pretty good rap sheet. What's kind of like the the greatest hits on the Austin and Stew angel portfolio? a long time ago in like 2014 maybe we we said we wanted to like do investing together and we found there was there's some hacker news thread somewhere I can find the actual post but there was a thread on hacker news about about angel investing and Sam Alman wrote this comment that was like here's how I did it I like I made a couple of small angel investments with my own money because I didn't have much money to actually do it but because those investments were good I had track record so then people wanted to give me their money to invest and and then that allowed me to do a lot more investing and then and then he like rattled off that he was early in Stripe and Airbnb and all these things and we were like well that makes sense because we don't have a ton of money to invest but we want to do it and so so that seems like a logical path and so we kind of decided like let's move really slowly and make investments that we have very high conviction in and and build some track record so that eventually we could invest other people's money and so first one we ever did my my brother Stu was working at Teespring and Jack Alman was his boss and Jack one day said that he was leaving to start a company. We both had kind of gotten to know Jack, Steu, got to know him really really well obviously working together and we were like he is phenomenally talented and motivated and they and it was the earliest kind of stage. It was we were investing on on a safe before they went into YC and they didn't have a name yet. They weren't totally clear on what they were building and and it was a $4 million post post money valuation and and but we knew that we wanted to bet on Jack. We knew some of the co-investors who were were pretty impressive names and and so we we made that bet which is Lattis and now is doing like I'm not even sure how much but but well over 100 million in in ARR and so that's been an amazing story. And then the next investment, I was the first ever paying customer of Superhuman and I absolutely loved the product and like couldn't believe how fast the team was moving and how good the product was. Like I I got I started using it when it barely worked and then it kind of was progressing from there and that was was incredible. So I was just like, "Hey Raul, can we please put money in?" So we did that one. Then a couple years later, Mercury I I started my company Capiche and I had like had really frustrating experiences with SVB when I was working with with Jason and and we were using SVB and I and so like I started searching around online for like are there modern alternatives to to these banks and and found a post from Amad who said like I'm building an alternative and I emailed him and said hey can I use it?

[01:36:07]and he was like, "Well, you'll be our third ever customer and we're still kind of building the thing, but yes, you can if you want." And so I signed up and I was Mercury's third customer and I got to like see I have these email threads that from I must have been 2019 where I was like reporting a bug to him because I was trying to get a wire or I can't remember what it was but something and it's like 10 p.m. on a Saturday and then like 20 minutes later Amod's like I just I just pushed a fix. It should work now.

[01:36:34]And like so again like seeing the velocity and and how much he cared and obviously it's a huge opportunity. So then someone at CRV reached out to me to do like a customer diligence call when they were raising their round. And so then I emailed them out and was like I I just did customer diligence for you. I told them they'd be insane not to invest and now I know you're raising. Can I please put money in? And and so we did.

[01:36:58]>> And yeah and then we raised the first cough drop. We call it cough drop capital. My last name is now Peter Smith, but originally it was Smith because we merged, my wife and I merged Peterson and Smith to make Peter Smith.

[01:37:08]So Stu and I were the Smith brothers and there's like a 150y old cough drop company called Smith Brothers cough drops and and so the name Cough Drop Capital is a from the homage to that. We our our dad and his brother had a boat called the Cough Drop when we were growing up. So So that's where Cough Drop Capital comes from. And so we yeah, we set we raised the first fund which was about a million dollars and still has a ton of active companies that are doing great and and then the next fund is through a million dollars and and is like kind of currently being deployed today.

[01:37:44]>> Nice. And it's pretty small checks, right? Like 25K 50k.

[01:37:47]>> Yeah. Yeah, exactly.

[01:37:48]>> Nice. That's easy to slot into a cap table.

[01:37:51]>> Yeah. Today I have like pretty much zero time for for investing and so it's mostly it's like sometimes customers of ours that are building cool things and I'm like, "Hey, can I put 25K into this thing or or like I don't know." So it's it's like the the deals that fall in my lap that that are where I'm focused. I'm not like going to YC demo day or like going out and doing taking pitch calls with with founders. It's just like cool stuff that I'm using or or just excited about.

[01:38:19]>> Well, that's the thing. That's like what everyone always says why founders are good angel investors. They're just in the trenches. They, you know, people are asking them for advice or using new products or trying things. People are trying their products. So like you just see different stuff than like some country club guy who you know is retired and maybe was like early Oracle executive or whatever. Like it's just it's a totally different world. So >> yeah. Yeah. Totally.

[01:38:45]>> Uh well cool. This is a lot of fun.

[01:38:46]Thanks for taking the time to do it.

[01:38:48]>> Yeah. Thank you so much for having me.

[01:38:49]This was awesome >> and thank you for listening. Thanks again to Hanover Park for supporting this episode. Head to hoverpark.com/turner for the AI native ERP for funds integrating your fund admin portfolio management and LP experience all in one platform. If you missed it, make sure to check out last week's episode with Chris France on building loops to fix email for developers. Tune in next week for Eric Barnhartson at Moto building the AI infrastructure loved by developers. If you like this conversation, please like, comment, subscribe, name your next AI secretary after me, or just call it Tumas like I did. If you don't want to miss a future episode, subscribe to my newsletter, the split t.it link in the description with each episode of transcripts emailed directly to your inbox every week. Thanks again for listening. See you next time.

[01:39:43][Music]