Why Your AI Has No Idea What's Happening in the World with Campbell Brown, CEO & Co-Founder of PredictHQ

(Intro Narrator at 00:00:00) Today, we're talking to Campbell Brown, CEO and co-founder of PredictHQ, about the technology that allows AI to know what's happening in the real world in real time. You're listening to Joel Beasley, Modern CTO.

(Joel Beasley at 00:00:19) I was reading a lot about what you're doing with PredictHQ and all of this stuff, but I wanted to hear it from you. How do you explain what the company is and does?

(Campbell Brown at 00:00:30) Yeah. Look, I think you can give the melodic or pragmatic view of what we do, right, where we predict the events that are going to impact businesses. But I think the easiest way to explain it is to articulate how our customers use it. And so I always use Uber as a great example. So Uber's been one of our longest-serving customers. But next time you get in an Uber, ask the driver what sort of visibility they have around what impactful events are going to come up. Using our API, Uber serves up these notifications to drivers to help them be in the right place at the right time ahead of time. And so that's one use case for Uber. They use us in their advertising product. They use us in their forecasting product. But then you have the likes of a CVS who are trying to understand what is going to impact, let's say, their water supply at their stores. So festivals drastically reduce water because everyone's buying it, and they don't want to be out of stock. And so the easiest way to think about it now, especially as we go on, is models are really, really smart, but they struggle with real-world context. They struggle to understand what's happening around them. Humans do it. Models do it. But if you have this real-world context, you're able to adapt what's going on in the world, which is what we do at scale across some of the largest companies in the world.

(Joel Beasley at 00:01:48) And you're the co-founder, correct?

(Campbell Brown at 00:01:50) Correct. Yeah. So I co-founded this with my CTO, Rob Kern. We did this in New Zealand, so I'm a Kiwi. Started in New Zealand, and then our first customer being Uber, I thought, well, I can't stay in New Zealand. I need to get to America. So I moved my family over here pretty much six months after founding the business.

(Joel Beasley at 00:02:10) Oh, that's amazing. So Uber, you said Uber was your first customer?

(Campbell Brown at 00:02:18) They inbounded to us, and we were super excited. So we went over and above for Uber, and what we did is—when you're a headless business, when you're API-first, you don't really have a UI. So you've got to articulate your value in their own environment. So actually, what we did is we mocked up the Uber app on our phones. We flew over to San Francisco, and we said, "Oh, hey, look. Wouldn't it be amazing if your drivers could understand what's going on here two days out, three days out, seven days out?" And they were like, "Yeah, that'd be pretty cool." And we had a really, really great champion at Uber who helped us immensely. And they would just say, "Look, we've tried to solve this internally. We can't do it at scale. So we're going to take a bet on you." And that was the beginning. But that visualization of how you can bring this real-world context to life was just a real penny-drop moment for myself and my co-founder.

(Joel Beasley at 00:03:15) So what were you before Uber called you? What were you advertising? What product did you have in the marketplace?

(Campbell Brown at 00:03:21) We actually started with this—we thought the way the world would go is that people would want to use this UI to log in and have projects and products associated to all these events that are happening globally. And we're like, "Yeah, it's going to be a SaaS business." But we launched that SaaS business. We got a couple of sign-ups—actually, probably a decent amount of sign-ups. But then what we realized is everyone just kept on coming to us and saying, "Hey, look, this is awesome, but we just want that piece of intelligence in our notification system, in our forecast model, in our dynamic pricing." And we took a serious look at that. We said, "Look, we're just going to go all in on APIs." And this is nine, ten years ago when we did that. And so, yes, we still have a UI today, but it's more of almost like a demo tool in terms of what you can do with the power of what we provide.

(Joel Beasley at 00:04:09) Did you just say nineteen years ago?

(Campbell Brown at 00:04:11) No, no. Nine to ten years ago.

(Joel Beasley at 00:04:13) Oh, nine to ten years ago.

(Campbell Brown at 00:04:14) I was—

(Joel Beasley at 00:04:14) Oh my gosh.

(Campbell Brown at 00:04:15) That's the Kiwi accent kind of emerging, merging together.

(Joel Beasley at 00:04:19) Yeah. Well, I want your time machine.

(Campbell Brown at 00:04:21) Yeah. Yeah. Yeah. No.

(Joel Beasley at 00:04:22) Okay. So nine to ten years ago, that's when you first got started. You were building because you thought people would want to come in and see these events happening. So the first version of the product, you were tracking different festivals, different types of physical in-person events. Is that what it was?

(Campbell Brown at 00:04:36) Yeah, totally. And we were doing a good job of it. We had a travel business before this, right? And so we were doing car rentals globally. And this is actually where the genesis of the idea came and actually where I met my co-founder. And we saw these peaks and troughs in demand for car rentals, but we couldn't quite understand why until we understood it was a flood in Florida had completely decreased demand, or it was a baseball game or a soccer game in New York that had surged demand. And so we thought there's got to be a way to get ahead of this rather than reactive and always saying, "Oh, it was that event here." It's like, "Hey, this event is coming." And we built this really, really basic framework to get started, but very quickly we saw that we needed to then build an entire ecosystem around it. And that entire ecosystem could effectively consume customers' demand data and then actually predict what's going to impact them in the future. And so we do that at scale today. I think since we built this platform, the product called Beam, we have consumed around $7.2 trillion worth of spend through the system to just generate all these predictions every single day, across hundreds of thousands of locations across the globe.

(Joel Beasley at 00:05:52) So you just have teams—sorry, this is such a big thing that it touches so many industries. How do you make it manageable? How do you say, "These are the things we're going to try to track?" How does that happen?

(Campbell Brown at 00:06:06) Yeah. We've built a massive pipeline, and that pipeline consumes around 10,000 different sources. And so that is anything from event sources, venue sources, even persona sources—meaning Taylor Swift versus, I don't know, Joel Beasley playing down at a local pub. It's really dependent because that will then impact the amount of people that are going to attend because we predict attendance. We predict the spend of an event. And so we do all these micro-predictions as well. So we've got to harvest all this information. And that in itself is really, really difficult because you've got to be super accurate. Because if we're being put into a forecast, right, if it's not accurate, that forecast is going to be off. And so the most painful thing in this business has been the data quality side of things. So getting that data quality right to build the foundations on it. And then just in terms of how big is this problem, from the $7.2 trillion worth of spend we've consumed from accommodation, retail, travel, 60 to 65% of their variability, the demand they can't answer, we can answer about 60 to 65% of that. So it is related to events, meaning, "Hey, we see the spike in demand. We have no idea what's going on." 65% of that type of demand is caused by these real-world events. So you're talking hundreds of millions, if not billions and billions of dollars every single year that people have no idea about what's going to happen, when it's going to happen, and how they take advantage of it. And so we've flipped that on its head and allow these businesses to kind of see into the future so they can do better labor optimization, better pricing, staffing. You name it, we kind of help them adapt to the real world.

(Joel Beasley at 00:07:47) Oh, that's so cool. You know, there's two that I'm thinking of right now. There's Live Nation and then Outback Presents. They're both large event companies, and Outback happens to be in the city that Josh and I live in. And they're doing, like, Nate Bargatze—fifty, seventy thousand people in arenas. And I'm curious, are they using you guys to help plan when they're going to put artists in certain places?

(Campbell Brown at 00:08:11) It's an awesome question. You know, at the moment, it's not typically who we sell to. I think we've got a couple of event-based companies, and what they're using us for is very similar to what you say is, "I have Nate Bargatze, right? And I want to make sure I maximize the wallet share when he is in Nashville or when he's in Chicago. We need to make sure that it's not intersecting with another potential event that's going to take wallet share." And so that's what typically the event businesses that are using us for, that's what they do it for because it can go both ways, by the way. Meaning, "Hey, if there's a—okay, this is a bad example—if there is a dental surgeon conference with 30,000 surgeons going to it, and it's 0.6 miles away from where Nate Bargatze is, that's actually probably a good time to put Nate Bargatze on because people that go to these conferences, they want to do something at night."

(Joel Beasley at 00:09:04) And they have money because they're surgeons.

(Campbell Brown at 00:09:06) Dude, they got so much money. You know, they're just out there just throwing it around. But I think therein lies the thing is what we can begin to understand as well was what events actually positively impact these ticket sales and what negatively impact these ticket sales. Because when we consume demand, demand can be footfall. It could be ticket sales. It can be transactions. I mean, we even have people that are pushing 911 calls to us as their demand. So it's really what is pushing and pulling this demand for your business is what we help predict. And so for your example, it's a potential. And this is the thing. This is what keeps me up at night is I'm sitting going, "Jeepers, we could do this for the events industry. We could do this for healthcare. We could do this for government." And you've got to stay really, really focused. Otherwise, you'll blow yourself up. And so it's a tricky one.

(Joel Beasley at 00:09:59) I've got so many more questions now.

(Campbell Brown at 00:10:02) Yeah.

(Joel Beasley at 00:10:02) But before I dive deep into them, give me an understanding of—I know you guys are private, right? You're not publicly traded.

(Campbell Brown at 00:10:08) Correct.

(Joel Beasley at 00:10:08) Yeah. With what you can share, how do you talk about the size of your company publicly?

(Campbell Brown at 00:10:14) You can't necessarily—saying how many staff you have is not an indicator of how big you are because I guess we're doing a lot more with what we've got. At kind of fifty, fifty-five people that we've got at the moment, we're significant, is the way that I would put it. And so in the tens of millions—

(Joel Beasley at 00:10:37) Very cool.

(Campbell Brown at 00:10:37) Of ARR. And that's true ARR as well. This is not, you know, take what you get on a day and multiply by 365.

(Joel Beasley at 00:10:46) I 100% understand that. Okay. I just wanted some, like, a general area of, you know, are you two people in a garage right now with $100K of revenue? Where are you guys? I think that helps everybody understand that.

(Campbell Brown at 00:11:02) Tens of millions. And I think the cool thing about it, mate, is what we're doing now and the speed at which we're moving is—I mean, it's just so cool to have small teams delivering so much in such a short space of time. And I think the businesses that can really wrap their head around that small team ethos are going to do pretty amazing things because you see some of these businesses doing 8 to 10 million or 2 to 3 million ARR per employee, which is bananas because the standard used to be $200,000 to $300,000. And those days are kind of behind us depending on how they describe the ARR.

(Joel Beasley at 00:11:42) The first time I ever saw this, and I enjoyed it, was a company called 37signals. They're popular for doing Ruby on Rails frameworks. But they had a small number of people, and they have this company called Basecamp—

(Campbell Brown at 00:11:59) Yeah.

(Joel Beasley at 00:11:59) Or a product called Basecamp. That was an early project management software. But that was one of the first times that I saw you can have a really lean team because of your efficiencies with technology and have a high amount of revenue per employee. That was cool.

(Campbell Brown at 00:12:08) Yeah. Look, I think, you know, for us, one of the best things that probably happened in the past year or so for me personally is we became profitable, but we carried on growing at a decent rate of knots. And so, you know, that's a combination of, I think, a little bit of being beaten up during COVID a little bit and learning actually how to be lean and how to be really focused on what products actually matter to us, how to get our GTM right. But yeah, it's just so satisfying because then you're taking—if you can keep growing with that profitability as well and redeploy that capital to keep growing and growing and growing, it's just such a great position to be in because the decisions you make really matter. You know? And you can't be making fifty different decisions either. You've got to be like, "What are your one to two decisions that you're going to make this six-week cycle in your development run?"

(Joel Beasley at 00:13:05) Well, it also makes sense. As you were describing what you do, as an engineer for twenty years, for me, I want to find really refined APIs, people that eat, sleep, and breathe it that I can plug into and bring some new benefit to my application. And that sounds like—I don't want to go build an event scraping and gathering system because I'm a car rental business, and I want to try to predict stuff. But if the endpoint already existed and I could just tap into it and pay some type of usage fee, brilliant. I'll do that all day. And so I always look for good companies that I can rely on to bring in the features that I want to bring in without having to roll them myself because it's such a project to do it yourself.

(Campbell Brown at 00:13:48) It is. I mean, the whole DIY thinking, it seems legitimate on the surface. And every single one of our customers has actually tried to build what we've built internally. And typically, it just doesn't scale, and it's a tricky thing. So, you know, because the thing with events, as you know, they change all the time. New events are—artists come up, things are canceled, postponed, they're shifting, there's severe weather, there's natural disasters. All these things we take into consideration when trying to better predict someone's demand. But yeah, I think there's also a real consideration that—not only do you—you know, that we've come to realize is you also just can't expect—I can't go, "Hey, Joel. Can you just go to our API docs? Learn our API docs, and then you do all the work." And then I get angry at you for not using predicted impact area or predicted attendance.

Campbell Brown at 00:14:40
You know, all our really amazing features — actually, that's not your fault. You shouldn't have to learn all that stuff. It should be natural. You should organically have a conversation and be able to basically build the code or the notebook rapidly. And that's our next big development, this product called Bolt that you can have a conversation with. You can build forecasts from a conversation. You can do whatever you want rather than going back and forth, rather than getting on a phone and talking to a solutions engineer initially. You can build most things within a few minutes versus this back-and-forth manual process.

Joel Beasley at 00:15:17
So you've got Beam, and then you've got Bolt. What's with the Bs? You guys like that?

Campbell Brown at 00:15:23
Yeah, my last name's Brown. Maybe that's — maybe it's a subconscious thing, but I do like the Bs. Bolt is our effectively AI-native developer ecosystem or environment, and Beam is what we use to consume anonymized demand data to make these predictions that people then consume back into their models. Meaning, a really good example is when the Mets play the Yankees, right? Uber cared deeply about that because it moves 20 to 25% of people to and from that event. Airbnb don't care about it as much because Joel is likely not getting on a plane from Nashville to New York to go watch a local derby. And those nuances, you scale that up to millions and millions of times every single week, is actually really hard for anyone to understand at scale. So that's why we built Beam, so it can do these predictions all over the place, no matter what location you have, to understand specifically what impacts your business at your location at that time. And that was a lightbulb moment for me when we built that platform. So yeah, it's been an amazing revelation for us as a business.

Joel Beasley at 00:16:33
So I'm gonna do a little recap here just so I understand, and then I've got more questions for you. So you started in this car rental, trying to understand these troughs and this variance in data. You built this first product, and then you started to get customers, and then the product has expanded and grown over the past decade. That's kind of how we've got here, right?

Campbell Brown at 00:16:53
Totally.

Joel Beasley at 00:16:54
Yep. Okay. Now, you did mention something earlier about 911 calls, and I've done a couple episodes actually and visited the dispatch centers of the 911 emergency response facilities. And so I'm curious to know — when you said 911 stuff, that gave me two thoughts. One, are you doing predictions of events and emergency response that'll be there? Or two, do you have a custom event prediction platform where I can send you my data and you can predict events within my data?

Campbell Brown at 00:17:27
Yeah. What it is — this one in particular, the customer that's using it is actually to predict emergency room. I don't know. I can't — so this is the thing between New Zealand and America. Do we say ER in America, or do we say — yeah, we do. Okay. So it's the emergency room. What that means is they need to understand if there's a festival within, let's say, three miles of this particular hospital — because they have numerous hospitals across the country — do we actually need to staff up? Alcohol abuse, drug abuse typically increases the amount of 911 calls. So it's actually giving effectively better care to the public based on what they know is gonna surge, unfortunately, their type of demand. So that is the use case that we're being used in for that particular business.

Joel Beasley at 00:18:17
Yeah. No, that's great. I grew up in emergency rooms. One of my moms was — I have a mom and a stepmom, and I call them both mom — but one of them was a doctor and ran the emergency room in the city we lived in. So I spent a lot of time in there and seeing it. And it's very cool. I ended up helping nobody with life, but my brother became a doctor, and it's cool to see them care and treat people. And I can understand when you say you feel good that your technology is being used for that. I can understand how that would be.

Campbell Brown at 00:18:45
For sure.

Joel Beasley at 00:18:45
So look, I'm excited to talk about the larger models too. Are you doing anything with Anthropic? Are you helping those guys at all? Are they consuming your stuff?

Campbell Brown at 00:18:59
Not at the moment. I think the view that we have at the moment with them is more around what value can we provide to our customer within the new way in which their jobs need to be done. And so what I mean by that is we see a shift in the way in which people are using technology to help them out every single day. And that shift has gone from, "I'm gonna use Google or a spreadsheet as my kind of search base," to, "I'm gonna go now to Claude or I'm gonna go to ChatGPT." And so what we've built is effectively an MCP where you can switch on PredictHQ. So you can go, "Hey, I'm gonna do a search now." And then rather than you providing not really that much context, it'll actually hit our APIs and provide context back to you. For example, I am at Franklin Barbecue. I'm the store manager there. You could flick that on and go, "Hey, what events are gonna impact Franklin Barbecue in the next seven days?" It'll come back to our API. We'll predict how far out you should search around Franklin Barbecue and provide you with a ranked list of the events that are gonna impact you and how far away they are gonna be from you. And so that's kind of the way in which we're playing with Claude and with ChatGPT at the moment, because the thing is here, not many people know what good looks like. And I think herein lies a problem, not just for our business, but for other businesses. So you could be the manager at Franklin Barbecue, and you do a search, and Claude generates an actual believable answer, but it's completely wrong. They're pulling in events from 20 miles away, and you go, "Yeah, I feel pretty comfortable about that." But they're missing out on hundreds of events that are actually going to impact you. Now, if you wanted to scale that up across 100 locations, 500 locations, imagine how wrong it's gonna be. And therein lies, I think, the challenge with AI at the moment. You know, some of the believable answers it generates for you — if you scale that up, the inaccuracies actually compound. And so that's not for everything, but for really hardcore data work, you gotta be super cognizant of what are the sources you're using to generate that answer and what is powering those predictions at that time of inference. So that's kind of the way we think about it. These models don't have real-world context at the moment. I'm sure we'll probably work closer and closer with these large language models. But at the moment, we've decided to go the other side and help out our customers in their jobs to be done by integrating in platforms they use every single day.

Joel Beasley at 00:21:31
Two people are coming to mind right now that I've had on the show previously that worked for really large companies that were kinda consumer-based. One of them is Domino's. They were very technologically advanced. And that would make sense for them to use your data. And the other one —

Campbell Brown at 00:21:48
Customer. Do they? Oh, Domino's is a customer?

Joel Beasley at 00:21:53
Oh, okay. Hey, so that's good. Well, that's awesome. They're — I'm a solutions engineer, right?

Campbell Brown at 00:21:54
Yeah.

Joel Beasley at 00:21:54
And the other one would be HEB. They're a really large grocery chain in Texas area. Do you know them?

Campbell Brown at 00:22:02
Look, we want them as a customer. I think they should. So if you're listening, come talk to me. We'll sort it out.

Joel Beasley at 00:22:10
They were actually really cool. We had two people from their company on, and I just remember — you know, I do so many of these, Campbell. I'm almost at a thousand episodes, and there's some that I just really remember. I love when you see these somewhat on-the-surface boring industries, and you go in and talk to these people, and they're so excited. And they're pushing the boundaries of technology, and it's like, yeah, grocery stores. Who would have thought?

Campbell Brown at 00:22:37
So look, I can't name them, but one of the largest, if not the largest, retailer in America is now a customer. And talk about being inspired. So we went to their HQ. If I told you the HQ, you'd get where it is, but I went there, and they were just so passionate about what they were doing. But they are the most sophisticated forecasting team we've ever met. Legitimately insane. But the thing is, they said to us, "Look, we can be as good as we can, but we still don't know what's gonna impact our store, which has groceries and everything in there, right?" The store didn't know it was gonna impact it. And they said, "Look, there is a store that is in town from here. We had no idea why demand was surging." And what we figured out, it was a film festival that was happening 0.3 miles away. And no one was prepared. Staffing wasn't prepared. And then they told us a knock-on effect. They're like, "So do you know what the knock-on effect of that is?" I'm like, "No, I'm sure it must — you know, out of stock." And they're like, "No, no, no. People leave the job." I'm like, "What?" They go, "If they feel overrun, if they feel overworked, they are likely to leave their job. And do you know how much it is to replace that individual?" And I'm like, "Whoa." When you actually dive deep into the knock-on effects of these surges in demand, it's significant. But to solve that for them at scale across thousands and thousands of locations, dude, it's one of my proudest moments. I tell my kids about it, and they're like, "This is cool, Dad. I finally get what you do." Because otherwise, they're like, "Dad, I have no idea what you do." You know what I mean?

Joel Beasley at 00:24:10
Yeah. Yeah. Signing a little kid NDA first.

Campbell Brown at 00:24:15
Exactly. Exactly.

Joel Beasley at 00:24:17
You guys do some work with Expedia? I think that's one of the brands we can talk about.

Campbell Brown at 00:24:21
Yeah, we can. Yeah, they're awesome. I'm actually on stage with them in a couple weeks in Las Vegas, but they are a super progressive customer of ours. They started really small, right? But now they've just been expanding, expanding. They've got us — they're using us in their portal for hotels. So hotels can log in to their pricing calendar and go, "Hey, what's happening on this day?" It'll use our technology to show what's happening around their specific hotel. Should they change their pricing? They're actually using us in their app to provide educated urgency is the best way that I would frame it. Meaning, "Hey, Joel, you're looking at a resort in Miami at this time. Hey, just so you know, there's these events that are happening. Book now, avoid price surging," and all that sort of stuff. So there's all sorts of different ways in which they're using us. I think we're in about six or seven different use cases inside Expedia. And therein lies the unique thing about being API-first: you can be pushed and pulled anywhere. And so once people unlock truly what impacts their business, you can kinda go anywhere. You can go into notification systems, into forecasting, into customer support. They can take that knowledge and they can just expand it out to any surface, which again is what AI has been so good for us because AI, as you know, is just creating all these new surfaces where you can actually deploy our capability.

Joel Beasley at 00:25:39
Now help me understand the Bolt and the MCP server. Are those the same thing? Are they different? How do they interact?

Campbell Brown at 00:25:46
Yeah, they're different. So think of the MCP — that kinda serves the store manager or the revenue manager who just wants to know for their specific area what's gonna impact them, right? And they can ask a question and they can get these results back. Whereas what Bolt does is it's a sandbox or engineering environment where you can basically build the integration. So say you're Expedia, you can actually go into Bolt now and build a complete integration end-to-end with a conversation that you can then take the code and put it into your travel application. You can put it into your forecast model. So it's actually just speeding up the entire process of integration. Because what we've found over the years is that people go off-piste. You know what I mean? Because you can't — you're not out there to say, "Hey, look, you need to do it our way because we're the smartest at this," because that never works, telling people that.

Joel Beasley at 00:26:39
No, it doesn't. Dude, you know, I'm very interested.

Campbell Brown at 00:26:43
Yeah. And so if you can create — the real lightbulb moment for myself and my co-founder was, "Hey, rather than forcing them to go down our guidelines, let's give them a natural experience where they can talk with an AI, right? Where they can talk through an agent and they can get it to do tasks for them and can build it out." And by default, they're following our guidelines, but we're not forcing them to follow our guidelines. And that was that real moment for us to go, "Oh man, Bolt is the way to go," because people can get from zero to 100 really, really quickly, and it's not us pounding, going, "You gotta follow our guidelines." And look, when we presented this to the business — I presented this about six weeks ago — I used this example, and I've got the satellite phone. And this satellite phone is killer for when I go right out into the backwaters when I go trail running. And I never read the instructions, and they are really good instructions. They've got online, offline. I was like, "No, I'm not gonna read it." And effectively, I was this close to sending it back because I'm like, "It's so awkward to send a text message on it. I hate it. It wasn't very good." Little did I know, it's effectively — it just connects to your phone, and then you can send text messages on your phone. It was almost like a rebound, and I was like, "I'm such an idiot," right? Again, I didn't read the instructions, right? And it was actually a YouTube video. There's a group — there's someone called Outdoor Boys, and he goes off into the wilderness, gets lost, it's freezing cold, and he goes, "I'm just gonna pull out my satellite phone and connect it to my phone, and I can talk with my wife." And I'm like, "Oh my goodness, that is what I needed to do," right? And so I think what that's saying is if you can present it in their own environment, if they can do it in a conversational way, they're far more likely to build a really, really great application. You're gonna collapse the time to value, and everything's gonna improve. And the quality of their integration is gonna significantly improve because they're using all the tools that we know work really, really well, rather than expecting them to follow the documentation. Let's let them have a conversation with it. It's effectively the way we've done it.

Joel Beasley at 00:28:45
Well, that's kinda brilliant because I don't like to read instructions or directions. I think everything should be so amazing that if your product is truly amazing, it'll be intuitive. That's what I say. I'm probably wrong.

(Joel Beasley at 00:28:58) But yeah, I don't like to read, but I will talk with somebody. I will have a conversation with somebody who's used it before and talk with them. So I like this idea that you've got this agent trained on how to do it because, you know, as a developer, I've worked on so many different types of projects from financial planning to real estate software, you name it. And each world kinda has its own little things depending on what you're doing.

(Joel Beasley at 00:29:25) And so that experience of Bolt being trained on how to get you what you want, you're just making it easier. So it's doing two things. It's having the code already matching your guidelines, but it's also allowing the success rate to increase of them actually implementing your system, which is directly tied to revenue because they're not giving up. Like, you almost gave up on the phone, you know?

(Joel Beasley at 00:29:51) Totally.

(Campbell Brown at 00:29:52) Yeah. Like, I mean, the real beautiful fact, we had one of our solutions engineers—look, I just posted in our Slack channel, "Hey, who's had a good experience with Bolt recently?" And he just goes, "Look, it would happen today." We had, I won't go into details of what, effectively, there were these IDs that were wrong related to an analysis, and he was like, "This would have taken me ages to try and identify." He uploaded it into Bolt, and Bolt was able to rerun and identify all the issues.

(Campbell Brown at 00:30:16) He was able to package it up, send it to the customer inside five minutes before he went into the weekend. Like, if we had done that normally, we would have waited. That solution might have been to the customer by, like, Tuesday. And I think that is no longer acceptable. Like, all expectations have changed, and I think that's so cool that we can give the right answer with high fidelity in a really rapid fashion, but also in an environment where our staff can use it to be knowledgeable.

(Campbell Brown at 00:30:46) Right? We have built up so much knowledge over the past, you know, nine or so years, and we're now kinda just imparting it on this platform, but also controlling this platform from a point of view of improving it. And I think therein lies a little bit of the trickiness between, do you offload everything to an MCP so you can do this all in Claude and CUDA? And we're of the opinion that, you know, there's so much specificity that's needed for our business that we wanna control certain elements and then let them take it to CUDA or wherever they wanna take it, into notebooks, because that first touch experience where you wanna collapse that time to value is so critical, and you need to own that.

(Campbell Brown at 00:31:24) And that for us is just, you know, we were so focused on improving our API docs, but that's gone. Those days are gone. I don't wanna read through docs. Tell me what I wanna do or build it for me, and I'm gonna come back and I'm gonna refine it. I'm gonna make it better, and all my time and effort and knowledge is gonna go into making this a kick-ass app rather than get my token. It's just all this boring stuff. Like, just replace it. And I love the fact that, you know, we're able to do that now.

(Joel Beasley at 00:31:56) Okay. So I wanna just wrap up with a little bit more about your background, if that's okay with you.

(Campbell Brown at 00:32:01) Yeah. Absolutely.

(Joel Beasley at 00:32:02) Was this the first company you started?

(Campbell Brown at 00:32:05) This is the first company that I set out where it's just myself and my co-founder. Before this, I've done a startup where we had a daily deal website, which we got, like, 95% market share in New Zealand, but it was with a founder called Shane Bradley. He started it, and he goes, "Like, you gotta come with me and do this," and there was, like, four of us or five of us at the time. And that was where I kinda got the real bloodlust for startups is that I was seeing what he was doing. I was part of it.

(Campbell Brown at 00:32:37) And so I wanted to do something that was then mine, and I could then build from the ground up from nothing. And that's kind of the path that I've been on. Dude, but you know what it's like. Your paths are so weird. Like, I was doing GIS. I was a GIS specialist in London, right, for the Islington Council, and they fired me. That's how bad I was at that job. I was literally mapping out instances of all these—it was dude, it was so bad. But then you, you know, you find your way. And then, but I still have a love for spatial stuff, but it's just so weird how you get to where you go. I guess the one thing I would say is that I'm always curious, and I think I never really appreciated how curious I was until I started going down the startup path.

(Campbell Brown at 00:33:24) And then it just exploded, and then I just couldn't think of being anyone different. Right?

(Joel Beasley at 00:33:30) And one of the leadership questions that I like to ask everyone is if there's one piece of leadership advice that you've received that you've put into practice and you've kept with you for a long time, what advice is that?

(Campbell Brown at 00:33:47) Yeah. It's, this is gonna sound actually quite odd. But there was a friend of mine actually shared it with me, and it really resonated. And it was called "making your bed." And what it is is this captain in the army did this big speech around making your bed. Like, no matter how—so whenever you get up, just make your bed. Make it really, really nice because no matter how bad your day is, you're gonna have a nice bed to come back to, and you're gonna be able to reset. And then I combine it with something else. One of my dear friends said to me, "The darkest hour is just before dawn." And all that's meaning is, like, you're gonna get through it. You're gonna get through the other side. Tomorrow is gonna happen. And I think when you're really in the mixer and you got all this stress going on, just sitting back and kinda reflecting on, like, "Man, tomorrow is gonna happen. It's gonna be okay. You're gonna get through it."

(Campbell Brown at 00:34:44) And often that once you get through that really, really tough time, you kinda reflect on it, and you're like, "Man, why was I so worried?" Right? Yes, it's a problem. Yes, it's big, but if you're built for this, you will find a way. And I think that thing of, like, just like that making your bed is for me is, like, just make sure you're comfortable. Make sure you can control what you can control so you can decompress, and then understand that when you are right in the freaking mixer, it's gonna be okay. You're gonna get through it. And I think that's both from a personal point of view and from a business point of view because those things clash all the flipping time.

(Campbell Brown at 00:35:23) Right? All the time. And I think it's, you know, the thing I often say is just the ability—you get the superpower as a founder to eat stress. You know, you eat it. The key thing is being able to reconcile that stress as well. I reconcile it with exercise, gold panning, weather prediction. Like, there's all these things that I find to reconcile it. But, again, it's the thing. It's like, as you get more and more stressful situations, you become able to deal with it more and more. And so then you see these, you see other people going through stress, you know, "Oh, that's not that stressful. I've been through this, this, and this, and this."

(Campbell Brown at 00:36:02) And, but it's a really cool thing. The only thing I'd say is, like, it's not just called "eat stress." You gotta reconcile it. Otherwise, you get yourself into trouble. Hopefully, I'm explaining that well.

(Joel Beasley at 00:36:14) Oh, from one entrepreneur to another? Absolutely. Yeah.

(Campbell Brown at 00:36:17) Yeah. Yeah. Yep. And you find times where you just—dude, you can't see your way through it. But you step back for, like, an hour or two or whatever, you know, you can get around it.

(Joel Beasley at 00:36:31) Oh, yeah. Yeah. I mean, I built and sold a software company, and I mean, the number of nights that my wife and I sat in the kitchen wondering if, like, this is the last payroll we're gonna be able to make before it really took off. And then it would go up, and then it would go down. And, eventually—I mean, it never stopped going up and down. Like, it was going in the right direction, but it would go up and down in the right direction. And just years of it, just, like, a decade of that, and then all of a sudden you're like, "Oh, okay. Like, it's always gonna go up or down." And, like, it's always gonna work out.

(Campbell Brown at 00:37:05) How'd that make you feel, though, mate? Like, I'm not turning the tables here as well.

(Joel Beasley at 00:37:09) No. No. It's fine.

(Campbell Brown at 00:37:10) What interests me is, right, is, like, I feel that I can handle way more stress than when I was in my twenties or thirties. Like, I feel like I have this ability to consume. But, like, how do you find stressful situations? Do you see yourself going up that ladder of being able to handle it?

(Joel Beasley at 00:37:27) Oh, yeah. Well, what really rung true to me is when you said, you see other people having situations that I've previously been in, and my instinct is to laugh. Like, that's nothing. Like, you're on level two. Wait till you get to, like, level 30. Like, you're, like, you just get so—your skin just gets so thick. And then also you pick up better habits. Like, you pick up better financial habits. You pick up better things that allow you to weather storms. And it's like, "Oh, this is just gonna happen, and we just need to be prepared to this degree, and then we can ride the waves." You know?

(Joel Beasley at 00:38:03) Yeah.

(Campbell Brown at 00:38:04) I think it's recognizing as well what you can control and what you can't control.

(Joel Beasley at 00:38:09) Well, I spend all my time on the things I can control.

(Campbell Brown at 00:38:12) Yeah. Let go of the things you can't control because I think it's just one of those things that they're the ones that will keep you up at night. But focus on the things you can control, and I think that's been—yeah. And, honestly, just sleep. Oh my gosh. Right? Like, I was doing three or four hours and just working all the time, but it just compounds. And it compounds and compounds and compounds until a point where you're just an absolute horrible person to hang out with. You're making stupid decisions.

(Campbell Brown at 00:38:40) And people just underestimate how important good sleep is to just get all of that stuff out of your body as well.

(Joel Beasley at 00:38:48) That just described my twenties.

(Campbell Brown at 00:38:50) You know? Totally, right.

(Joel Beasley at 00:38:52) You're like, for a couple year, you burn out, you burn out, it doesn't work. And then you find, like, the one, two, or three things that you need. Like, my whole calendar is organized into, like, three areas. Like, if I were to tell myself that fifteen years ago, I'd be like, "You're crazy. You have to do way more things." It's like, no. You gotta do these three things, and you gotta do them so well. You have to make sure the sales—spend time in sales, make sure the product's undeniable. Like, those are, like, your two pillars, you know, and then you gotta take care of yourself. So yeah.

(Campbell Brown at 00:39:21) Totally, mate. And I think the other thing you gotta recognize—like, the other thing I'm cognizant of is not, you know, hindsight's twenty-twenty, and always gonna remember, like, hindsight is not knowledge. You know, when people are like, "Oh, yeah. Like, this happened because I did this." And that in actual fact, it's probably pure out of pure coincidence you actually did that thing that led to this other thing. Right? And so for me, I'm always cognizant of, like, was this an active decision or was it something that just happened because I made myself available to that? And I think that's what I'm trying to, you know, I try to convince the founders. Like, your job as a founder is pattern recognition. Don't take everything verbatim because there's people giving you advice that are actually in a very, very different situation.

(Campbell Brown at 00:40:00) They're in tens of millions of ARR, a hundred millions of ARR versus starting out. And so that pattern recognition for founders is really important. How can you apply that, you know, what's the patterns you see in the advice you get that you can then apply to your situation that you can control? Because so many people just listen and are just like, "I gotta replicate that exact thing for this exact moment for my exact situation," and it will not work.

(Joel Beasley at 00:40:25) It definitely does not work. Campbell, I think we're friends now. I like—I think we are. This is great. This is fantastic conversation.

(Campbell Brown at 00:40:35) Yeah. No worries. That was good, mate. Like, I think, you know, it's great to talk with founders as well who get it. And, you know, it's lonely, mate. No one's gonna get you, but I wouldn't have it any other way, the things that you go through, man. Like, it's just—it's such a—do you know what I worry about? I worry about, you know, not having the stress in my life. What am I gonna do? Like, if, you know, like, you worry—like I think it's a healthy amount of stress. But, anyway, maybe I'm being rose-tinted glasses looking at it that way.

(Joel Beasley at 00:41:05) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email, [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.