Playing the Long Game in Business (Episode 138)

Jeff Bullas

00:00:24 - 00:02:35

Hi everyone and welcome to The Jeff Bullas Show. Today I have with me, Jay Bartot. Jay has an impressive resume. He's the quintessential entrepreneur. He's had multiple startups. He's the CEO and Co-Founder of the Seattle-based Zeitworks, an AI powered business process intelligence platform that provides enterprises with the data-driven insights and tools they need to continuously improve and transform their business process operations. Now before everyone starts clicking away here, I know business processes sound boring, but I tell you what, there's magic in the process because that's the difference between success and failure. So stick with us guys because we're gonna discover some magic here.

Jay is a serial technology entrepreneur and innovator with 20+ years of experience developing data and machine learning applications for a variety of verticals such as e-commerce and we know how big that's become over the last few years, online advertising, travel, medical informatics and consumer video creation. As a co-founder and CTO, Jay has built technology teams, innovative analytics, machine learning platforms and products for startups like Vhoto, Medify, Farecast and AdRelevance. These startups led to acquisitions from Hulu, Alliance Health Networks, Microsoft and Nielsen/NetRatings, where he played senior technology roles but there's more prior to Zeitworks, Jay was the CTO and Managing Director at Madrona Venture Labs where he helped to create and incubate nearly a dozen exceptional startups, providing support and expertise across a number of different areas of startup creation, including ideation, customer discovery and validation, data collection, design and procurement, product development and the list goes on. I won't go into all of them, Jay attended the University of Iowa where he studied music and we might find a little bit more about his passion for Music, Anthropology, we might talk a little bit about that as well and Computer Science. I think we're gonna be talking more about Computer Science’s application to business because business has changed a lot over the last few years. So Jay, welcome to the show.

Jay Bartot

00:02:35 - 00:02:40

Thanks, Jeff. I really appreciate it. Looking forward to our conversation.

Jeff Bullas

00:02:40 - 00:03:03

So Jay, you said you grew up in the period of grunge music and you attended University of Iowa where you studied Music, Anthropology and Computer Science. So let's talk a little bit about the music passion. How did that start?

Jay Bartot

00:03:04 - 00:04:09

You know, it started when I was a kid. I had older brothers and sisters who were, you know, growing up in the ‘70s and ‘80s, you know, had all the records of all the big rock bands and as a young kid like 8, 9, 10 years old, I loved the music, I was a little bit of a hyper kid so I was always kind of banging my hands on the table and kind of dancing along to the music and in school came up that we needed to all choose an instrument for the school band. And so I chose the drums, much to the chagrin of my family, I procured a snare drum and started beating it wildly at home, driving everybody crazy. But you know, it was a great way to get energy out and, you know, I've always had a good sense of rhythm and so I really became passionate about the drums playing all through middle school and high school and, you know, starting to play semi professionally in high school and into college and really thought that I was going to be a professional musician at the outset of my adulthood.

Jeff Bullas

00:04:10 - 00:04:18

So was there a hero that you wanted to be growing up as a drummer? Who were you inspired by a drummer?

Jay Bartot

00:04:19 - 00:05:19

You know, certainly several drummers from big rock bands from the ‘70s and ‘80s, John Bonham from Led Zeppelin, Stewart Copeland from the Police, geez, you know, Neil Peart from Rush, I mean, so many great drummers from that era, they were just incredibly talented and, you know, even when I started getting part time jobs, you know, in high school, you know, every other week I would pick up my paycheck and jump on a bus and go downtown in Chicago to the drum store and procure some new piece of equipment. So that kid is still inside me. Sometimes I just will get a wild hair on Saturday afternoon and head down to the music store. Of course, I'll tell my wife I'm going somewhere else because, you know, I've got enough gear as it is. But yeah, that passion for gear and music is still very much inside.

Jeff Bullas

00:05:20 - 00:05:49

Yeah, I totally get that because very coincidentally as we discussed earlier, my son was a drummer and I remember going to many drum shops and going, why do you need a $300 new piece of equipment? I've got this sound so much better. Really? I can't pick that. So I totally get it. And are you still friends with your neighbors?

Jay Bartot

00:05:50 - 00:06:54

I am. But you know, I've learned to find other ways to play music and express that level of creativity through other instruments as well. So, you know, in high school, even though as primarily a drummer, I picked up the guitar and got really passionate about guitar playing and then the bass and a little bit of keyboards here and there. But I like being a multi-instrumentalist and this kind of ties into some of my startup life because I very much consider myself a generalist even though I'm a technology person and have a lot of depth there. I'm really happy most when I'm wearing lots of hats, I'm wearing a technology hat, I'm wearing a marketing hat, I'm wearing a sales hat, etc. And singing with music. I'm happy when playing some drums and playing some guitar, I'm playing some bass, I'm singing and composing, you know, that definitely crosses over into entrepreneurship.

Jeff Bullas

00:06:55 - 00:07:28

Okay, so that's a perfect segue way to, alright, so you're passionate about music, thought you're going to be a professional musician. There’s always a 15 year old inside all of us. So where was the inspiration to start the entrepreneurial journey? Was there an aha moment or was it just a slow evolutionary process? Where did the inspiration come from?

Jay Bartot

00:07:29 - 00:12:18

You know, I think it was probably a bit of a slower evolutionary process. I'm a very creative person and I, you know, everything I do has some element of creativity to it, including entrepreneurship. Just bridging the gap real quick between music and computer science. Back in college, I was doing some music production on multi track tape machines and you got to remember and I'm sorry, I'm dating myself, but this is kind of late ‘80s and I was laying down guitar and drum tracks on these cheap multi track tape recorders as a pretty bad experience. They're pretty crappy and a college roommate said to me, you know, you should go down to the computer center at the university, they'll lend you the money to buy a personal computer where you can do more sophisticated music production with many sequences and keyboards and so forth. And I was like someone would lend me money, God, they must be out of their minds. But I marched down there and picked out an early Macintosh computer and I think there were $2,000, which was an incredible amount of money, but I could pay it off over time and I took it home and just kind of got obsessed with this thing as the ultimate instrument. I could compose music on it, I could write papers on it, I could draw and paint, you know, you name it even back in those days pre-internet. And so I started doing music production on them and you know, I don't know what happened. Maybe this was the spark that just sort of hit me like a lightning bulb. But at some point I got the idea into my head that I could program it. I could make up my own programs and make them do what I wanted to do. And so I started thinking about this and I talked to that same roommate again and said, you know, I'm thinking about programming this, what do you know about that? He said, oh, you can't do that. That's really, really hard. It's complicated and you got a lot of math and science and I was like, well, you know, I was focused on music in high school, but I like math, I like science and so I started, I bought up a basic language package for the Macintosh made by Microsoft actually, and I started programming this thing and then I really got obsessed. Again, I think programming and computer science is also very creative. As much as maybe people think that, oh it's, you know, it's all structure and math and so forth. It's very creative, you can just create all kinds of interesting programs and I got into graphics. So I'm a very visual person too. And so next thing I know I changed my major to computer science and my life took a really hard right turn and always stayed with the music, the music still exists today and I'm out buying gear and I'm up on, you know, all the latest music production, softwares and technologies, but the new computer science direction took me to a whole new place as I started my career, you know, I tended to bounce around a lot. I was a young impatient kid and worked at a medium sized company and I was like, these people moved too slow and I want to move faster and do other more neat things and ended up at small companies where I have a tenure of 18 months or so. I remember my dad telling me like, oh you gotta stay at companies longer, nobody likes anybody who moves around. But I think I was kind of the early generation of tech people who were bouncing around from experience to experience and you know, eventually really started settling into actual startups that were venture backed, venture funded and the internet in the mid-90s was starting to blossom and take off and so I got kind of caught up in that whole wave and really, ever since then I've been a serial tech entrepreneur and worked a lot with venture capital and just, you know, really learned that whole game. By the late ‘90s, I started getting interested in data and machine learning, which was also just kind of starting to blossom formerly known as AI technology, interestingly, it's called that again now. It's interesting to be old enough to see these cycles of this terminology but I didn't mean to, but I ended up kind of riding this wave of data. Machine learning provides some kind of transparency or value for either consumers or business users and I've just kinda been rinsing and repeating that ever since.

Jeff Bullas

00:12:19 - 00:12:46

So that raises the next question for me. So you've been bouncing around, you've had a ton of experience and the generalist Jay has seen a few nooks and crannies across all parts of the business. So how did you start your first business? What inspired that?

Jay Bartot

00:12:47 - 00:16:14

Well, the first startup where I could actually call myself a co-founder, I was paired up typically in the early days with business people. So I met a guy, I was introduced to a guy in 1998, I think, who was an advertising guy. He came from the traditional advertising world and he and his career was suddenly faced with buying ads and ad campaigns on the Internet. And so he found that in the offline world there were books you could buy and data you could get in printed form which would sort of tell you approximately who was advertising what and where across tv and radio and other traditional media. Well, no such thing existed for the internet. There was no way to know who was advertising, what and where on the internet. Again late ‘90’s and so being a software guy and a creative guy, I got introduced to this business person and we put our heads together, brought in a few other co-founders and built the first online advertising measurement service. We basically would crawl around to different websites, you know, looking for things, html objects and so forth that look like advertisements and we'd pull them down and then we would try to classify them. This is an ad about computer hardware or this is an ad about a service or what have you. And so we built this service and system and you know, got venture backing and started getting customers and next thing, you know, we were acquired ultimately by Nielsen NetRatings, first by another company called Media Metrics. Um,

But there was one kind of interesting thing that ended up being a bit of a theme in my career that I kind of stumbled into working on ad relevance and that is that we were providing transparency into, we're into an area where there was no transparency before with data and with analytics. What I hadn't expected was that when you provide transparency, you might, there might be a group of people who don't like that and benefit from there being a lack of transparency. So, you know, the story I always like to tell here is that at the time, who was the king of the hill and they had all kinds of pages and all kinds of advertisements from third parties, but also their own advertisements that they would run on their site and we would crawl around Yahoo! yanking down ads and calculating, you know, the frequency of the ads and what the ads were and so forth. Well they didn't like this. No, they didn't like it one bit and so they got mad at us and called us up and said, we want you to stop doing this and then we said, well, you know, we think we're providing a service to the business and community and they're like, no, really, we want you to stop doing this. And we got a cease and desist letter and then they tried to shut off our crawling technology, but we just sort of went around it and, but it was sort of a first lesson to me that when you're democratizing information and providing transparency into, you know, new areas that, you know you're onto something when you're pissing somebody off, when you find your pissing somebody off.

Jeff Bullas

00:16:15 - 00:16:41

Yeah, there's a lot of stories of small startups being bought all around the world. I'm trying to remember, I think was a TV series about it recently and the big end of town has got the money in the muscle and the pockets to scare people and scare small businesses. So how did you solve that? Did you get good advice from lawyers? How did you solve this?

Jay Bartot

00:16:42 - 00:18:56

Well, I think we, I'm not sure we ever solved it, but we persevered. There were sleepless nights though where we were scared. And I have another story about the airline industry and a company we did call Farecast that I co founded with a faculty member from the University of Washington. We're similar thing, we were providing transparency into airline prices and whether an airline price was gonna go up or down in the future, before the plane took off and it turns out that airlines use revenue management systems to optimize the price of an airplane ticket to sell it to the right person at the right time. And we developed algorithms that would predict based upon a number of underlying factors, probably things like demand and other things that were happening in the system to say, like if you want to go from Boston to Seattle on this particular date, the price is 500 right now, will it go down to 400 before the plane takes off? It's an inherently perishable product. So it's an interesting problem. And so we were pretty successful at making these predictions. Again, there were some airlines that didn't like it. They didn't like people reverse engineering their system and passing transparency under the consumer.

And so there was one airline in particular, a pretty big one that shut off our data. And there were some really scary nights where, you know, it's like, well, I think we can get by with this one airline shutting us off. But what if the rest of the airlines do this? What if they all go, yeah, you know, we're mad at these guys and we don't like the transparency they're providing. So we're gonna shut them off too. So those were scary. Those were scary times. And I would say that all the startups I've worked on have had sleepless nights where you're really a little dinghy boat riding a really big wave in the ocean and you know, you're not, never sure when you're going to be propelled up and you're gonna flip over and you know, have to hopefully, you know, grab onto a log or something or whatever, but survive.

Jeff Bullas

00:18:57 - 00:19:51

Exactly, right. Because you're sometimes living on the edge, sometimes you might be running low on cash. Yeah. I've got a mate who's running a global startup, he’s five years in. He's pivoted a couple of times and whenever we catch up, one of my first questions I asked him, I said, how's the runway? Yeah. He's been down to seven days with, you know, with a payroll to meet the next Monday and he's flown to the city with one of investors and just asked the question and managed to pull another few 100,000. So I'd say to him, how do you sleep at that?

Jay Bartot

00:19:51 - 00:19:53

What did he say?

Jeff Bullas

00:19:54 - 00:20:26

I think he has a few wines. I think that helps him. I just think, I think you have to be wired a certain way as well to live on that edge of danger. But yeah, it's a challenge. So I do love your story about, you know, trying to gain the ticketing prices for a flight. So I think one of the tricks I try and uses the incognito mode so they can't track me through cookies.

Jay Bartot

00:20:27 - 00:20:29

Yeah.

Jeff Bullas

00:20:29 - 00:20:34

But it's a challenge at the moment trying to get the right price because they're just going up and up.

Jay Bartot

00:20:35 - 00:20:42

Yeah, well I'm glad I wasn't running this business through Covid because you know, that would, that's what I'm sure is an exogenous shock to the system.

Jeff Bullas

00:20:43 - 00:20:43

Yeah.

Jay Bartot

00:20:44 - 00:21:14

And no predictive model based upon historical data would probably be able to reason with that. So you know, those are all risks too.

I heard that a lot of recommendation engines which are, you know, ubiquitous with internet services, you know, really struggled to suddenly see this very different change in demand and you know, needing to run around and fix those things was, I'm sure a challenge for a lot of companies.

Jeff Bullas

00:21:15 - 00:21:50

And that raises another interesting area of being an entrepreneur is timing can be almost everything. So are there any stories that you can share regarding good timing and bad timing. I'd be intrigued. Is there anything on your event horizon in the past that has, being, I got the timing really, really good or my God, I got it really, really wrong. Are there any stories that you share?

Jay Bartot

00:21:51 - 00:24:26

Yeah, I mean it's a great question. And it can be really tricky and I've been in situations where sometimes it felt like we were already too late only to realize later we're actually pretty, maybe too early. You know, the online advertising measurement business I think was timed really well because you know, the internet was exploding and e-commerce was exploding. When Vhoto was an iOS mobile app that would analyze consumer shot video, looking for good moments in video. So things that were exciting or where there was a lot of action taking place. And the idea was you would roll the camera and just record video instead of taking single pictures, which is a paradigm that comes from the beginning of photography in the early 20th century or even late 19th century. So with this new technology and these incredibly powerful devices in our pocket, why not just roll the video and then let an algorithm figure out where the good frames were. And so, you know, that one was interesting in that Instagram, you know, had some interesting timing of their own. There were known Instagram like apps out there before Instagram, but Instagram really kind of hit it big and, you know, mass adoption and the filters they offered and so forth were just right there at the right time and the right place. Where we came in and did our thing, I think we probably were a little bit too early. Later on, things like Snapchat and you know, apps with really advanced filtering and analysis capabilities came after us and hit it bigger. So I would say, you know, that's an example where just for whatever reason, our execution was just not quite right in a crowded field to you know, the other thing I'd say is that when you have competition it can be very validating. It's like this is not a dumb idea, other people are doing it, but you know, you're bumping elbows with those people in the market, whereas when you have something really new and it's out front, you're not bumping elbows with people in the market, which is nice, but you're also having to educate the public on what you're doing and why it's valuable, which is time consuming and expensive and risky. So you know, there's a spectrum of, you know, where you can be and what your timing is and where things are comfortable or not comfortable that can work to your advantage or disadvantage.

Jeff Bullas

00:24:27 - 00:25:27

Yeah, I saw a TED talk where they collected data on 200 startups, I can't remember his name but they found that timing by far was the most important element of success. I think it was over 40% and the rest were nowhere as large as that, so now that raises the most important part of it, you've just started Zeitworks which is about using AI and data driven insights to help transform business process operations. So I'm intrigued where this idea came from. I would say that it's maybe out of some pain points that you may have experienced with former businesses. Tell us where the inspiration for Zeitworks comes from?

Jay Bartot

00:25:28 - 00:30:31

Sure. Yeah. So in 2018, I was working in the startup studio that you mentioned earlier, Madrona Venture Labs and my colleagues and I. A small group of us have experienced entrepreneurs and operators. We had a pretty cool gig and the gig was to go out and find, look or think up new interesting ventures backed by technology startup ideas, do your homework on those ideas because that's one thing that most entrepreneurs tend to not do. Go out and talk to customers is like, is this really a pain point or is it nice to have, understand the market, do all these things before you start building and raising money because the sooner you find out that an idea is good or bad, the better. Most ideas are bad and so our goal was to come up with a lot of ideas and kill most of them because they're bad or you know had some issue associated with them and then keep the best ones and work on those. Find entrepreneurs to join us and help them build up an MVP product and find pilot customers and so forth and spin out of the lab. And so that's kind of what we were in this process of doing across multiple companies. We got some insight from our funder, a drone adventure group here in Seattle that they were really interested in The RPA Space. RPA stands for robotic process automation. With automation technology, people are sitting at their desks all day doing repetitive tasks. Well what if a computer could do those tasks for you? And so RPA was making big inroads in this time period helping organizations with repetitive tasks that people do for example processing a loan application or processing an insurance claim or onboarding an employee in an HR department and a large organization. And so the idea was, you know, find the repetitive stuff that people probably don't like to do anyway because it's really grunt work and have a little software robot do it for you. And so as I mentioned RPA became really big, companies like UiPath and automation anywhere in Blue Prism and many others suddenly started turning into unicorns and getting tons of business. And so our friends at the venture capital firm said go look at this space to see if there's anything interesting else to do there, maybe in the next generation of technology or what have you. And so we started talking with lots of folks, vendors, customers management, consulting groups and what we learned was more often than not when an organization has an operations team doing these repetitive processes, Nobody knows what the processes are and I found this to be really interesting. So you might have 20 people doing processing business loan applications, but everybody's doing it differently and you may have had it written down at one point like do this, then do this, then do this, but over time with changes in the business and changes in employees and nutrition and so forth, pretty soon, everybody's kind of doing things their own way. And so nobody knows what the steps are. And so that makes it hard to automate certainly the process, if it's automatable but it's also just really inefficient. And, you know, businesses are always looking for ways to optimize, you know, costs. And so what we heard was, you know, hey, I'm interested in automation and I think I have automation opportunities, but help me understand what these people do better on my team because I couldn't tell you, I don't know who's good, I don't know who's bad. I don't know who needs more training. I don't know which parts of the process should be rethought or re engineered. And so we realize that's a really big problem and it's a problem we can solve with novel data and machine learning technology. And so here what I like to say is that the data is hiding in plain sight, people are at the computers using their different applications, moving information from one system to another and part of the process. And so we build software that observes how people are interacting with their applications and basically finds the cycles for where a process starts, finds where it ends up being able to identify that piece of work in that process and then measure it. How long did it take? How complex is it, how many resources, human resources were needed to work on it? Whereas a lot of time is being spent in the process that seems inordinate compared to other parts of the process. And so again, this is another transparency play. I got this team. I got people doing work. I don't know how they do it. It's a black box, help me shine a light into what's really going on there so I can understand it and fix it and make it better.

Jeff Bullas

00:30:32 - 00:31:08

So you identify there's a paying point, you can save companies money and you can use technology to help them optimize their processes. Quite often when you start out building a business, the mistake can be you try to do everything all at once and that's a big issue and I had a conversation with a friend of mine and she's trying to build out a company and she says I need to do all of this first and I'm going, well, have you thought about just doing one thing? So where did you start? Did you find the biggest problem you thought there was and built a minimal viable product?

Jay Bartot

00:31:09 - 00:33:27

Yes, Yes. And these are tricky things. So investors primarily want to know right off the bat that the opportunity you're going after is ginormous. It's huge. The total addressable market, as it's known, is a really big number, billions and billions of dollars, but they also know that the startup, the phrase we use here is tries to boil the ocean that they'll fail because they'll end up not doing anything right. And so typically what startups trying to do is find some verticality to their problem and cut it up into chunks and solve one at a time. And so the verticality can be all kinds of things. Oftentimes it's an industry. So let's figure out healthcare first or let's figure out financial services or some said it's a consumer idea, some segment of the consumer population, let's nail teenagers before we go on to college students for something. What comes to mind with that statement is Facebook and Mark Zuckerberg focusing on college students first. He started at the single university level, he started going out to other universities, staying with college students and then of course, finally we all know, you know, he went very wide. And so finding those dimensions where, you know, you can start and prove out a concept and then start going horizontally is really the goal for most of these kinds of companies. And again the dimensions can be different things depending on what the business is and what the product is and so forth. For us, in Zeitworks, you know, gosh, repetitive business processes done by humans and operations teams. They're everywhere. They're in insurance and in industry and in HR and you know, financial services and healthcare but we can't do them all. So we've started in financial services with small to medium sized companies looking to solve their particular problems. And if we're successful here then we'll move on to other pastures.

Jeff Bullas

00:33:28 - 00:33:44

So I love your term, try and boil the ocean. So Zeitworks, where did you focus on initially? In other words, what pot did you try and boil rather than just boil the ocean? So what pot was it that you chose?

Jay Bartot

00:33:45 - 00:36:18

Yeah. Well, so our pilot customers because we're still in the pilot phase, we're still, you know, bringing on new customers and learning about their environments and situations and idiosyncrasies of their data, they're all kind of related to financial services with a bit of a bent on customer support. There's a lot of repetitive processes and customer support where somebody calls up and says I need help with this or that or what have you. And then the agent then goes through a variety of, you know, different types of work to, you know, try to answer those questions or fulfill what the customers need. And so we found a niche there and there's a little bit of variants and sort of the type of customer that you know, that we're servicing. So we have a bank as a customer. We have a company that does financial transaction processing. We have a large e-commerce site that we're working with that also does offline financial transaction processing. So this is sort of our little neighborhood right now, but I know for example that revenue cycle management and healthcare is a huge area for us, you know, there's all throughout healthcare are people on the phone calling insurance companies on behalf of providers. So calling payers on behalf of providers saying where's my money? You know, we had a patient, we performed these procedures on the patient. Here are the codes. They're insured with you, you owe us money, very repetitive, very arduous work and a place really where technology can help a lot. As a startup person, healthcare scares me, some industries are not as technology friendly as others or just that they move really slowly. So, you know, I've had some experiences in my career where you meet with a potential pilot customer and the healthcare industry and they're big and have a great name that they have on your website. You have a meeting with them and they're like this is really cool, we should do something together, let's meet again next month, nine months later you're in a meeting with them and they're like this is really cool, we should do something together, let's meet again next week. And that stuff will kill your company.

Jeff Bullas

00:36:19 - 00:36:57

Some of the things that intrigues me is that you need to collect data that means you need to get, you know, the term is a PI’s or the access to the back end of different systems and there's so many systems now. So I think years ago access to the data was harder. Is it? How do you access data because you might be collecting data across accounting systems, HR, finance, you know, so where do you start with collecting that data? Getting permission?

Jay Bartot

00:36:58 - 00:39:26

Yes. Yeah it's tricky. You know as I mentioned earlier when we were crawling websites, we were collecting somebody else's data. When we were building the airline prediction models, the airfare prediction models we were relying on the airlines data. And you know the list goes on and on so much that I came up with this term or this abbreviation SED which stands for somebody else's data and dealing with someone else's data can be tricky. Someone who's data that is may not want you to have the data or they may want to charge you for it or they may have all kinds of restrictions or stipulations around it. And so, you know, one of the things I found when I was talking with a lot of budding entrepreneurs, when I worked at Madrona Venture Labs was this notion that people wanted to build AI startups. So I meet young kids in their ‘20s who, you know, were fresh out of computer science schools who want to build an AI company. I want to be able to build a machine learning company. I'm really excited about this technology and what I learned to point out to them is that most of these companies are not machine learning companies, they're data companies and figuring out how you're gonna procure data in a reasonable efficient way, you know, that has robustness to it, is really the problem that I faced at most of these companies. It hasn't been about machine learning technology. Machine learning technology is generally a commodity. There's tons of machine learning technology and open source academia that is constantly publishing new machine learning technology. Google, Amazon, Microsoft, Facebook, they give away their machine learning technology platforms, why do they do that? Because they own the data and the data is what's powerful and important and is the mode for a particular company. So you know, investors are usually kind of like, well, you know, you may think you have some novel machine learning technology, but pretty soon everybody will have it. But what data do you have? And is it unique? Do you have exclusive rights to it? Is it data that you pull out of thin air that you own. All those things really end up being much more important.

Jeff Bullas

00:39:27 - 00:41:03

It is. That's why I asked the question because I noticed that you talked about basically mining Yahoo!'s data then it was the airlines data and I've watched friends who started social media marketing platforms and they've had to access the API of Twitter and Instagram and Facebook and LinkedIn and with just one pull of the plug, their businesses can be gone. So you're, I think a theme with what you've been doing, which is absolutely fascinating to me. I don't have fascinating to our listeners and viewers, but I'm sure it is, business has changed a lot. You're essentially a data miner that also on top of that adds a layer of process and smart. So you're both the minor as well as the producer. So you need to get access to the data and that's complicated. Number two, the next part is how do you make sense of that data and you've got to do this in a cost efficient way. So what happens with without revealing too much, I'm sure. But so how do you go in and say, okay, what are we gonna do with this data? We've got the data now. Now, where to connect the dots and make sense of it and then maybe put it in a format that can be used. How's that? Tell us a little bit about that journey.

Jay Bartot

00:41:04 - 00:43:07

Well that's, you know, it's a great question. I mean, that's usually where a lot of the work is, you know, again, thinking about all the young data scientists, you know, just come out of school or what have you. They want to build models. They want to train models, they want to do inference, they wanna do predictions, they want to do forecasting. But usually most of the work is pre-processing the data to put it into a form that the signal in it is obvious and those machine learning models can find that signal. And you know, a lot of this work is, you know, it can be a little mind numbing and boring, but it's also a lot where the art and creativity comes in and so, you know, there's an expression out there that, you know, real world data is messy. There's a lot of data sets that, you know, people who are learning about data science can use to experiment with and so forth. But when you're really pulling raw data out of a consumer or a business or a machine or what have you, oftentimes, it's noisy, it's missing values. It has all kinds of problems with it that first require a lot of manual grunt work to go in and and figure out, okay, how do I measure this data in a way and take those measurements and turn them into features from my machine learning model. And again, it can be time consuming. It's kind of an art. It's a little bit of a science but it's also an art and it ends up being where a lot of the work is. And I think, you know, I'm generalizing here, but I think most data scientists would probably agree that 80% of the work of building a machine learning model or a data science model is in data pre-processing and data prep. And that can be sobering to someone who realizes when they're in their first data science jobs like this is actually what most of the work is and not totally sure I like this.

Jeff Bullas

00:43:08 - 00:43:14

Yeah, it looks great in the lab. But yeah, now I'm actually in the real world and I've got to really deal with dirty data.

Jay Bartot

00:43:15 - 00:43:17

Dirty data, yeah.

Jeff Bullas

00:43:18 - 00:43:44

We've got a hard stop coming up for you to get into another appointment. So a couple of questions and then we'll just finish up, number one, what have been some of your biggest challenges? The first question to ask is, what gives you the most sense of joy and fulfillment. So let's deal with what have been some of your biggest challenges that have kept you up at night along your journey.

Jay Bartot

00:43:45 - 00:46:00

I think the biggest challenges, I'll just mention this first just because we were just talking about it is, you know, access to data and a fear that the result of my analysis of somebody else's data is going to result in someone saying, well you can't have my data or I've just decided that you can't have my data and so I'm gonna pull the plug on you. That certainly has been some big challenges. I mean, I think the biggest challenge for any technology entrepreneur is just time and time of course and money are, you know, deeply linked and so, you know, when you have some money, whether it's your own bootstrap money or an investor's money, you have, you know, only so much time to build the right thing, get it into the right hands, start monetizing your product or service and things always take longer than you want them to or expect them to customers move more slowly. Engineering problems turn out to be harder, cloud costs or other computer resources you pay for, turn out to be more expensive. And so that balancing act of having the right people on the right team and the right timeline and building the right minimum viable product to get the right signal for from the potential customer out of it and doing it all in the time frame that you have, which is always very limited is, you know, really the biggest challenge and it's hard to get it right. You mentioned timing earlier, you know, getting timing right is part of that equation of, you know, if you find that you bring something to market, but you're early, then you need more runway to be able to sustain yourself while you wait for the market to catch up with what you're doing. Sometimes you don't have that. And sometimes, you know, you may have to shut something down only to look back a few years later and go, the market's ready for my product now. But I got the timing off.

Jeff Bullas

00:46:01 - 00:47:06

I think that raises a very important thing for businesses that if you've got the ability to hang in there and persist and play the long game, that timing can show up as you mentioned. And I've been involved with a company Shuttle Rock in New Zealand here for the last 10 years and we pivot twice and I was on the board and we managed to be around long enough to stumble into an opportunity and that came down to just being able to play the long game and having the investors believe in you as well. And I think that's the important part, but getting, trying to work out what's important and where you spend your time, reminds me of a book I just read by Seneca and it's a little essay wrote called On the Shortness of Life, and he talks very much about spending your time on things that matter. So very short, but it's important and that's hard to work out sometimes.

Jay Bartot

00:47:07 - 00:47:09

Yeah, absolutely.

Jeff Bullas

00:47:10 - 00:47:18

So just to wrap up things, the second question, what's brought you the most joy and fulfillment as an entrepreneur?

Jay Bartot

00:47:19 - 00:48:32

Well, you know, certainly, hearing that people use the product and got value out of it I think is probably the greatest joy. I hear from people even now, 10 or 15 years later, oh yeah, I used Farecast, that was really cool, you know, that saved me money, you know, hearing those kinds of things, you know, brings me all kinds of joy working with and helping other entrepreneurs, especially young budding entrepreneurs, you know, is also a really great feeling again, I really think all of this stuff is creative and there's an inherent vulnerability around being creative because you're, you know, you're creating your product or technology or whatever it is, and then you're putting it out there and saying, you know, I built this, this is me, this is part of me and my team, and we hope you like it and get value out of it. And so when you get those signals back, which can be few and far between, but when you get them back with someone saying, yeah, that's cool, I really like that. That helped me. That's the greatest joy.

Jeff Bullas

00:48:33 - 00:48:55

Well, it's great to hear that, and I think you've obviously added a lot of value to the world, and to companies solving problems for them, and it continues. So, thank you for sharing your journey and your story. Jay, it's been an absolute pleasure. I have a lot more questions to ask, but maybe that's enough for a follow up chat.

Jay Bartot

00:48:56 - 00:48:56

So, happy to chat some more.

Jeff Bullas

00:48:57 - 00:49:08

Okay, thank you very much. Jay, it's been an absolute pleasure and enjoy the rest of your evening. It's this morning here, and I look forward to chatting soon.