#23 | Delivering AI Solutions to the Enterprise with Conor Twomey - Co-Founder & CEO at AI One

How can businesses leverage AI to create data-driven solutions and scale efficiently in today's competitive FinTech environment?

In this episode, Frazer and Conor discuss:

> How AI is transforming business innovation by enabling companies to reimagine workflows, fuel growth opportunities, and expand market reach beyond mere cost-saving

> Why a strong data strategy is crucial for leveraging AI to drive precise and confident decision-making

> How generative AI is democratizing advanced technology by lowering skill barriers and making it accessible for businesses of all sizes

- - - 

Conor Twomey is an accomplished executive with over 15 years of experience in addressing complex data challenges for leading global corporations. He is currently an AI Co-Founder at Stealth Startup. Conor is the former Head of AI Strategy at KX, a pioneer in real-time data analytics and decision intelligence. Under his leadership, KX successfully transitioned from a time-series database company to the Enterprise AI platform of choice for large-scale AI implementations. Before this role, Conor managed a 400-person organization encompassing Presales, Professional Services, Support, Managed Services, and Customer Success Management. 

Renowned for his insights on data and AI, Conor is a sought-after speaker and contributor on frontier technology topics, including Data, Analytics, Machine Learning, AI, and Generative AI.

Transcript

SPEAKER_02 0:10

Friends, welcome and uh welcome back, I hope, to the Best2Go FinTech Podcast. I'm Fraser Anderson. It feels like it's been a while, but the fall is here, and um the guests will be a coming, so thank you for your patience. Today I'm extremely excited to be joined by my friend Connor Toomey, co-founder of AI One. Connor Toomey is an accomplished executive with over 15 years of experience in addressing complex data challenges for leading global corporations. He is currently a co-founder of AI One. Hopefully, he'll be able to share a little bit more about that with us today. Connor is the former head of AI strategy at KX, a pioneer in real-time data analytics and decision intelligence. Under his leadership, KX successfully transitioned from a time series database company to the Enterprise AI platform of choice for large-scale AI implementations. Before this role, Connor managed a 400-person organization encompassing pre-sales, professional services, support, managed services, and customer success. Renowned for his insights on data and AI, Connor's been speaking at the NVIDIA conference and all sorts of places. He's a sought-after speaker and contributor on frontier technology topics, including data, analytics, machine learning, AI, and of course, generative AI. So without further ado, we will turn it over to my chat with Connor. Alright, Connor, take us back to the uh the early days of your career when you first started working in AI. What did that even mean when you started? How has it evolved? How do you think about the evolution?

SPEAKER_03 1:45

I suppose I was very lucky, Fraser. I I studied math in university, and pretty much everyone in my class wanted to be an actuary, which is how do you use data from the past to predict what's likely going to happen in the future, but in a very morbid sense? Yes. Like who's going to die and when?

SPEAKER_02 2:05

Yeah.

SPEAKER_03 2:05

And so I realized that that wasn't something I was particularly passionate about.

SPEAKER_02 2:09

Is is that like a uh University of Cork like fast like track for math? Is that still the case? Do they still want to do that?

SPEAKER_03 2:18

Yeah, we we've uh we have a deep heritage in Cork actually. Uh so George Bule, who is um the creator of Boolean algebra, the ones and the zeros that are the backbone, the bedrock of any computer system. George Bule famously studied and taught in University College Cork. So we've got a deep a deep heritage in mathematics and computer science as well. Um and so yeah, I I would I would have potentially explored computer science at the time, but I suppose the dot-com uh hangover was still in place at that point in time. Uh UCC actually dramatically ramped up the amount of places they had in computer science. But when I was in in high school, in in secondary school, as we call it in Ireland, I found problem solving to be something I really enjoyed. I played a little bit of computer games, I loved playing team sports as well. But problem solving was something I enjoyed, and I was quite good at math, and I would have trialed for the Irish math Olympiad team, didn't make the team, didn't get anywhere close to the team. But at the age of kind of uh 16, 17, would have spent Saturdays uh at the university learning about the sort of problems at the at the math Olympiad. And so I studied math then in UCC for four years and realized actually I didn't want to be an actory at all. And I was very lucky that a very good friend of mine, Finn Bardenihy, big shout-out pal, called me up and he said, Listen, I'm doing this master's course in computational finance. It's it's using maths and technology, but an understanding of business and finance to solve some interesting data challenges, and I think you'd really enjoy it. So off I went to Limerick for a year and uh was very thankful. I met my my girlfriend now wife who was in that class as well, which is great. But when I was in university then uh in Limerick, a lecturer pulled me aside and said, Connor, you're showing a great aptitude for actually using technology as a springboard to solve problems. And there's a very small Irish-based company that sits at the trading desks of world-class organizations and really helps translate the bits and the bytes of technology capability into what it means in the context of uh trade execution and risk. And so I joined that organization which which kind of promised uh almost a too good to be true uh philosophy. It was like a Willy Wonka's golden ticket of a side door into top hedge funds, banks, exchanges, regulators. And I was a little bit skeptical. Yeah, uh, I put in my CV on a Tuesday, I got a phone call on the Wednesday. Could I be at the head office on the Thursday? I got a job offer on the Friday, but only if I could start on the Monday. And so this was uh by the seat of your pants organization. It was incredible. And it delivered above and beyond all expectations. I spent 15 years with that organization, and it saw me living in New York, uh, London, and Tokyo, working for some of the top banks, hedge funds, regulators, and exchanges, solving the most demanding, in-the-moment decisioning and data-intensive analytical challenges. And the entire bedrock of AI is data. There's no AI strategy without a data strategy. And people can often conflate analytics with AI or machine learning in AI. And so without wanting to split hairs too much, the ability to harness data assets in service of driving more confident, more precise, more timely decisions for an organization is very much what the promise of AI is around lowering the barrier and lowering the cost of prediction.

SPEAKER_02 6:05

Can you talk about specifically what kind of got the trust barriers down and what were kind of the some of the early use cases that you saw that were really interesting? Because I would say today, and I don't know if it's because SaaS has proliferated to the extent that people are just kind of overwhelmed with the number of folks promising everything to them and under-delivering, but I would say I mean machine learning systems get a lot of skepticism when I sit in with entrepreneurs on customer calls and uh generative AI systems just kind of get fear uh and skepticism. So, I mean, I mean talk about that.

SPEAKER_03 6:44

Yeah, I think like in my own journey, half of my career was spent actually building out systems, so hands-on keyboards, sitting with the business, understanding what they wanted to do with their data assets.

SPEAKER_01 6:55

Yeah.

SPEAKER_03 6:56

The other half was almost as a CEO or a C-suite whisperer. So advising people on a better way to harness their data assets, either to monetize them or to reduce risk or whatever it might be. And so on that journey, I had over 600 customer conversations around machine learning. So we're we're going back to 2017, 2018, 2019. About two-thirds of those conversations were within the capital markets ecosystem, but one-third equally working with aerospace and defense companies, medical companies, manufacturing organizations, Formula One race teams. And I realized that the commonality was a recommendation engine. Like the outcome was a recommendation engine. Yeah. I have I have a thousand things to do today, recommend which ones are most likely to be the things that I should focus on. And so there was a a lot of skepticism because uh to to get access to that promised land, you had to have teams of PhD students who then understood how to write C or C, and ultimately Python, which became the de facto specialist labor, specialist tooling, deep algorithmic expertise and mathematical expertise. And so the accessibility of the recommendation engine, the juice wasn't really worth the squeeze. And so there were some fantastic use cases, like recommending Fraser that after you watch The Crown on Netflix, that you may enjoy another documentary about the monarchy. But it kind of failed on its promise, if I'm candid. And the barriers to entry were very high. And fast forward then to the generative AI era, and this is this compounding effect of now being able to harness the best of systems, the best of data analytics, the best of machine learning without specialist labor. Like the things that make generative AI so appealing and AI more broadly appealing now is that accessibility. Natural language has become the skill set that you need to be able to divine inspiration from your data estate. And so you combine that with the ability to bring online previously unsearchable or disparate data assets, and now you've got this double whammy effect. And as the actual models get better, faster, stronger over time, it's almost like a triple whammy effect. So you've got the barrier to adoption is low, the instant hit in terms of making all of these huge, um, huge amounts of unstructured data now searchable and discoverable into existing workflows, combined with the models actually getting more accurate and more cost-effective.

SPEAKER_02 9:40

Ironically, there's almost like uh it's like the S and B has an advantage because whether you're in an S B or an enterprise, every employee is familiar with Chat GPT at least. And so there's kind of a groundswell to bring these technologies into the organization. People are just doing it off their phones or their personal machine anyway. But with an S and B at least you don't have that much risk. With an enterprise, you have all this risk. And what I would like you to comment on is obviously writing prompts doesn't take any specialist labor, but but the data strategy that you need to execute to get the most out of it and at the enterprise still, I would argue, requires highly specialized labor and a lot of work.

SPEAKER_03 10:24

I think we saw this Fraser with um with the rise of cloud, cloud technologies, cloud computing, circuit 2016. A lot of enterprises almost used cloud, and I appreciate that a lot of what I'm gonna say is contentious, yeah. It's my own personal view. So if people disagree, please leave comments. I'd love to love to engage in conversation on it. But a lot of organizations in the enterprise use cloud almost as a forcing function to adopt a continuous integration, continuous deployment methodology. And so a different paradigm, a new way that you were forced to actually write and think about and conceptually uh live and breathe software within your organization. And during that journey, people stub their toe a lot on privacy, security, permissions, authentication, entitlements, data sovereignty, compliance rules, GDPR, California Data Acts. And so this entire AI ecosystem has been born, or the generative AI wave certainly has been born in a cloud-first world. So getting access to these models is incredibly taxing if you want to do everything on-prem in an air-gapped environment. Yeah. Where I'm going with that is for organizations that have been through the pain of successfully treating their private cloud tenant as an extension of their firewall, this technology isn't particularly difficult to bring in and to use successfully. I think some of the challenges that we see is from a crawl walk-run approach is that people have really high expectations of how this is going to deliver a return on investment.

SPEAKER_01 12:05

Yeah.

SPEAKER_03 12:06

And very often the place where an enterprise needs to focus is probably the first step, is the crawl, is on better search. So, how can I allow all of my data assets within the organization to be accessible for internal purposes only? And how can I use natural language as a way to express myself to understand what it is I'm looking for? And that that is a remarkable change from literal search, fuzzy matching, yeah, um, or regular expression search. So I think there's low-hanging fruit there. And then the kind of the walk phase is how am I then able to take a look at my existing products and services externally and actually augment those processes? So some of it is about automation, some of it is around reducing the time to decision, some of it's actually about reducing risk. And so, how can I actually embed this technology into my existing services and products? And then the the sort of the run phase is how can I create an innovation factory? How can I actually test and validate new growth opportunities for my organization and use it to purely ideate? So you've got almost like the efficiency, this is the improving existing products and services, and this is actually creating a factory for brand new innovation to occur.

SPEAKER_02 13:28

Curious how you're thinking about. I mean, you're we're not going to talk much about what your what your new company does, but it will be creating data as every company does. Oh, yes. Um curious if you could talk maybe through use cases or the way you're thinking about storing your own data to make it a truly AI-first business?

SPEAKER_03 13:51

Yeah, I think it's um I like to say whenever you're taking a look at a at an existing business model or an existing process or service or product, imagine you were to burn that down to the ground and rebuild it. So it's like out of the ashes rises the AI first Phoenix. Yeah. And then what does that allow you to do in terms of either uh digital labor expanding your go-to-market? So, how do you how do you make a sales team of three or four people feel like an army of three or four hundred using automation and hyper-personalization? Or how do you massively reduce either the time or the risk involved in making an assessment? So kind of pulling risk out of the system. And so for us, the way that we're thinking about that is access to the technology is ubiquitous through cloud. So it's not a huge capital expenditure to actually spin up an AI-based system.

SPEAKER_01 14:50

Yeah.

SPEAKER_03 14:51

The new models are coming out like every single hour. Yeah. It's almost like uh it's become nonsensical, these updates that come out with different parameter models all of the time. And so we'll we'll chat about those models maybe as well in terms of uh will they plateau at a certain point in time? But then how does one actually build a differentiated mode in a world where the cost of software engineering is going to zero, the accuracy of these models is is reaching a plateau at some point in time. And really, it's about understanding the domain expertise and having customer empathy for the enterprise. So understanding what the technology is in service of. How do I actually translate the bits and the bytes into new experiences for my own customers or new products and services I didn't think of?

SPEAKER_02 15:42

Well, I mean, talk about the models a bit because you mentioned sort of plateauing in terms of accuracy. Um, you know, another thing that people get through that that is thrown around quite a bit is sort of like small language models for enterprise use cases. I've literally never heard of anyone using one or seriously considering building one. You know, what are your takes? I'm sure you have some.

SPEAKER_03 16:06

There's a bunch of different ways to solve the solve the challenges around uh specific industries or sectors. If you're if you're in the business of medical devices, for example, and you're taking an off-the-shelf model like Lama 3 as an example, then the way that that general purpose model has been trained is is you know it's been trained on a lot of public information. And so when you apply a general model to a very specific vertical or domain, the level of accuracy that you're expecting is going to be impacted.

SPEAKER_00 16:39

Yeah.

SPEAKER_03 16:40

And so from there you've got a couple of choices. You either look to people who have uh created their own small model on a specific domain, or as you might apply fine-tuning as an organization, so you might spend money to actually augment the weights and biases of a off-the-shelf model. Uh, or you might look at something like retrieval augmented generation, which is a way for you to take your proprietary data assets and put that into a workflow that blends the best of large language models with your proprietary data assets. And there's no right answer, Fraser. There's no like this is the way it must be done. I think what people do think about when they're trying to understand what the best approach is, I think they at the enterprise level, there's really three key pillars that if I boil down every generative AI problem for an enterprise, it fits into one of three categories, and that is accuracy as the most important element. So if I were to ask the system a thousand questions, how many times will it hallucinate? How many times will it come up with something that looks and reads plausible but doesn't have any actual material behind it? And that is the top C-suite concern. And there's a variety of tactics that people have taken to try and improve accuracy. Graph rag as an example, uh mixture of agents, mixture of expert, different models that people have. But accuracy is number one. The second is the time frame or the latency, the response time of the system. So if, for example, I'm in a customer-facing workflow and it takes me spinny wheel 10 seconds to get a response, then you've you've lost the customer. And so understanding how quickly you can actually execute. And then the third is cost, and cost fits into two buckets for me. There's the initial setup cost and there's the runtime cost. And amongst all of that, you've got risk, the risk appetite as well. And so certain workflows have a far greater risk appetite if you're doing something creative, like hey, write me a sci-fi novel. If it makes a couple of errors, it's not a big deal. But if it's a recommend uh, you know, should should I operate on a specific cardiac patient, then your threshold from a risk perspective for an accuracy is very, very low. So I think that that kind of feeds into accuracy if I'm honest.

SPEAKER_02 19:06

Especially in financial services, the like accuracy threshold is so I think it's so challenging or has been so challenging to get people comfortable with. I do suspect there's some way to actually just build that into a workflow that might have a human in the loop for a while that will end up paying dividends down the road as your workforce is freed up to be more kind of productive, happy, strategic, problem-solving oriented. Um, not sure how you think about it because I mean, to your point, accuracy and hallucination is a bit of a nightmare. I mean, I I use these things all the time, and uh, I'm floored every now and then. Like yesterday, I was uh I was trying to calculate uh my kind of running pace and and it hallucinated, it was off by an order of magnitude. I mean, like it was like it's a very, very basic division problem. And then the other piece, which sorry, I'm asking you a hundred questions and mostly just rambling, but the other piece I I'd like you to think about is or talk about is I'm of the opinion that the sort of full cost is being hidden, like to the customer, and curious the extent to which you think it's being hidden. Um, so how much margin of safety do you need to build into your ROI?

SPEAKER_03 20:22

ROI is such a great word, right? Because it does have different connotations. But if I if I think about the C-suite conversations that I've been having, uh so much of the time up front is spent explaining to the C-suite that this technology cannot defy the laws of physics. And so we are not at AGI right now. Nor is this technology that you can just stick your head in the sand and it's a fad, and in six months' time the world will have forgotten about AI. And so there's a big Goldilocks seat spot in the middle. And there's some great statistics out there, and I'll quote Gardner specifically, who say that nine out of ten generative AI programs will fail to reach production in the enterprise in 2024, and they'll explicitly fail to reach production because of a lack of understanding of total cost of ownership and return on investment. And so, like a very simple framework for an executive to think about AI is this is where I am today, and this is the promised land of where I could get to. And so it's not about could I use AI as to replace an entire service or product, but there may be steps along the journey where it makes sense to do so in a way that's cost effective. That bridge to get from where I am to where I want to be needs to be something that can be delivered with certainty. And the costs come in around infrastructure costs, people costs, software costs, token consumption costs over time. But also it's the lift of what is my What state is my system in today versus what investment will it take to get it into a state that'll allow me to capitalize on that use case? So if if you start to view AI as something that can augment how you're doing business today through the lens of time, cost and risk, you then create a universal language, a taxonomy for you to consistently talk about AI to all people within your organization. And that's something that we encourage CEOs and C-suite executives to think about a lot. But from a cost perspective, yeah, it's it's how quickly can you validate a specific use case, what is the return on investment going to be, and then full understanding of fully loaded cost of total cost of ownership. So it's the hardware, the software, the people.

SPEAKER_02 22:52

Yeah.

SPEAKER_03 22:52

And getting me from state A to state B, and then understanding that you'll likely have a dual mortgage problem as well. You're going to have some time where you want your existing system and your new system to work in parallel as you gain more confidence that the new system is delivering the results that you expect.

SPEAKER_02 23:12

I'd love you to, and this is just a setup question, because you and I chatted about this the other day, but you articulated it so well. You know, if today, if today the barriers of entry to getting the recommendation engine everybody wants are too high, but now generative AI is here, and so now everyone has a uh hallucination engine or a word production engine of uh output or content that may or may not be valuable, may or may not be suitable, may or may not be sufficient to replace a human workflow. You know, how should how should the enterprise actually be thinking about the full opportunity set from an efficiency, growth, innovation perspective?

SPEAKER_03 24:00

So many people have failed to release AI into the enterprise because they're focused so much on the cost saving. It's like large-scale automation, which is a euphemism for layoffs, redundancies, operational leverage. Yeah. And so I think the way that people are viewing the challenges, they're focused so much on the bottom line, they should be focused on the top line. Yeah. So when you turn the conversation into uh not can I save 20% of cost, like Clarna letting go all of their support staff, it's like can I actually divert my precious and rare and beautiful and wonderful resources internally towards growing market share or towards you know doing more go-to-market outreach or actually innovating. Yeah, like the dream for a CEO is like this Jaws effect where you're able to actually take cost out of the system and then reinvest that actually in innovation. Yeah, you don't want to suffer from the innovators' dilemma where you know you've got an existing business that you want to keep the status quo of because the barrier to people breaking into uh a monopoly and creating perfect competition with brand new business model dynamics and pricing dynamics is gonna be fast. It's gonna be radical. So that's both petrifying, terrifying, but also exhilarating. Yeah. Does that answer question? Fraser, does that answer where we want to get to there?

SPEAKER_02 25:33

I think that answers the question. I think it does. You are spending a lot of your time now exploring latest technology releases, latest product releases. Where is your head? Where are your fingers typing away? What are you interacting with?

SPEAKER_03 25:51

I think there's a couple of interesting areas. I think for CEOs, the ability to give any CEO in an enterprise more confidence around their AI of decision making, I think, is something that might add some value. So giving them a framework for how do they view their existing technology footprint, skills footprint, data footprint to understand what the art of the possible is, regardless of sector or geography. I think that's it. That's an interesting thing that we're we're looking at, exploring. We're like, is that an interesting business model? If you take a look at the area of private equity, I think we get we hear so much feedback from private equity uh personas around this idea of IT diligence. And so we've got a part of our process that means when we're looking at potentially investing in a business, we must get this outsourced tech diligence, basically. Like, hey, is this a bag of frogs or is this a standard set of technology? Is this going to be unsupportable? Is it on mainframe? Is it written in cobalt? Help me out. And so the the value, the service that's provided is very much a check-the-box service. Yeah. And it's not actually providing a springboard for private equity firms to understand how or where they could capitalize on either technology or data assets as we enter the AI era. So that's that's an area that's like a little rabbit hole that we're coming up, coming down and coming back up out of. And then if you were to put a gun to my head and say, what do I think is going to be the largest disruption from a technology perspective in this generation? I think it's going to be the role of agentic software engineering. And so Which is the ability for someone using English language to express how an application could or should look and then using this idea of agents, which is it's the ability to give a specific persona and agency, uh, autonomy to a system to make decisions. And so you might say, in the most basic example, if you've seen Claude 3.5 Sonnet, you'll have seen examples where people write just in-time software. So, like, hey, I want you to build a React container that has a game of Tetris that's in hot pink and a little Game Boy controller. And the agent will break that problem down and say, Okay, well, if I was going to build a program like that, I'd probably think of this as the logic. And if I want to launch that into a container, a second persona might say, Okay, I can take that logic and actually do something with it. A third persona might say, Ah, I must test, let me test this all out. And a fourth persona might actually document the whole thing. So, what you're seeing is this orchestration layer where specifically tuned models are able to take on different parts of the workflow and then work in a team effort. Uh, most notably, a company called Cognition AI who brought out Devin and had a video of Devin. It was the first example of an enterprise scale uh software uh agent. And people find it mind-blowing. Uh, since then, there's been a bunch of companies, uh, a company called Cosign out of Y Combinator, yeah, have released, yeah, are releasing Genie. There's an open source program called uh OpenDevin. It's been re rebranded to Open Hands, and there's an enterprise company that's just raised five million dollars called All Hands that are going to try and bring an open source version of this sort of technology to the market. But what it means in terms of some of the what what does it mean when the price of software engineering goes to zero? What does it mean in terms of accelerating your roadmap, reducing your risk, uh migrating off legacy software, whatever it might be? I think it's a massive paradigm shift in terms of how people are actually building out code, the pace of coding, the pace of software releases is going to change.

SPEAKER_02 30:01

I mean, that does seem to move towards a future that you had articulated earlier, that like self-funding innovation. If you can remove the cost, remove the risk, remove the time required, now you have some very precious resources you can hopefully deploy.

SPEAKER_03 30:16

Exactly. Exactly. It's like a Robin Hood. It's like you're see you're stealing cost out of something that's non-core to your business. Yeah. And then you're you're freeing up capital in a very timely fashion to make that available for growth and innovation explicitly. And so that that's something that we're really excited by.

SPEAKER_02 30:36

Well, we're gonna have to have a longer conversation about AI one um at some point soon, I hope. But I think right now it's time to shift to rapid fire.

SPEAKER_03 30:48

Let's do it. Okay. If there's not a couple of running questions in here, I'm gonna be very upset. All right.

SPEAKER_02 30:53

Um I'm gonna have to make this up on the fly. Who impresses you?

SPEAKER_03 30:57

Fraser, you impress me every single day. Uh my co-founder impresses me. My wife impresses me. That's a tough question.

SPEAKER_02 31:06

That's the best answer I've ever gotten to that question. We're just gonna move on. Don't change a thing. What piece of media, book, podcast, movie, article, essay have you gifted or shared the most?

SPEAKER_03 31:17

Ooh, I love Andy Raskins, the greatest sales dick that was ever built. And it it talks through just a framework for how to articulate how you're going to partner with a potential organization to generate enterprise value. And so the format that it takes is um discuss a universal truth. So something that's undeniable, a change in the world, the introduction of agentic software engineers. Yes. Tease the promised land. So help someone in their mind's eye get to that promised land destination. So that's the second part. So what does it look like? What does it feel like? Then what is your organization? And this is the only time, the first time actually that you introduce your organization, but what does your organization do to help get the person to that magical state, to the promised land? And then what data artifacts can you provide to give the person confidence that you can actually deliver on that promise? And that's just like a very simple framework, no matter what service, industry, sector you're in. So that that's something that I share very frequently.

SPEAKER_02 32:24

I love that. What skill do you spend the most time developing outside of work? Or it can be work-related.

SPEAKER_03 32:31

That's a that's a tough one. I think there's a few a few things. I tried to be generous with my time and my network, and so I am kind of in phase two of my my career journey. I spend a lot of time mentoring younger people who are so gifted from an IQ and EQ perspective, and they've got talent to burn, but they they haven't had their eyes opened up to the aperture of what's possible for their career journey. And so I love, love, love spending time with people for 30 minutes or 40 minutes where I don't tell them exactly where they should go in their career, but it's like a little career compass that I point them just slightly in the right direction to go off and explore. So I get I get great satisfaction from that. I've got two young kids, so I absolutely love learning from them and spending time from them, even if they're waking me up by poking me in the eye at two in the morning. That's okay. And running, Fraser, all things running. If I am not talking about AI or mentoring somebody, you will see me running the streets of New York City.

SPEAKER_02 33:35

Yes, yes. Um, or making me run the streets of Boston or the forests of New Hampshire or uh along the West Side Highway. Um your journey's been inspirational, Fraser. Seriously, well done, pal. I'm uh getting a little better, a little faster every day. How do you cultivate flow states? I mean, you have tremendous focus.

SPEAKER_03 33:56

I think if um there's a couple of things that were influential for me, I suppose. One one was reading a lot of Malcolm Gladwell's stuff around just understanding of my own thought process. There's another author called William Poundstone, who writes these beautiful books about the intersection of behavioral psychology and business, and then James Clear Atomic Habits or Shane Parrish with his Farnham Street. You combine a bunch of these different things and you begin to learn a little bit more about yourself. It's one of the nice things about getting old, you get to kind of discover what motivates you, what are your rewards, your pseudo-rewards, and then how do you align those rewards to an end objective? And so I suppose the if I was to just parallel it into just running, I went home one one time at the holidays uh after living in New York for a little bit of time, and a very close friend was very, very kind to me, and he said, Connor, you look like a stuffed turkey. And I was after putting on quite a bit of weight, I was 245 pounds at the time. But it's um he told me not what I wanted to hear, but what I needed to hear, and it was from a place of kindness. And so uh from there, I I did a bit of soul searching into how I could change my own habits and how could I remove the friction from actually adopting a sustainable habit. And so I just started packing my running gear the night before work, and then as I'm leaving in a scramble to get to a meeting first thing in the morning, I just grab my running bag and instead of taking a subway home, I run home. So it's like I apply that same principle into anything that you want to accomplish. If you want to spend more time being technical, then roll up your sleeves, write some code, and get a co-pilot to help explain everything until you understand it better. If you want to get better at chess, you know, don't just play a thousand games of chess, start doing some of the lessons on chess.com. And so uh yeah, spend a lot of time thinking. I don't listen to music when or podcasts when I'm running, and I use that as a time for a lot of self-reflection and for me to, I suppose, massage through things that made me feel good during a day or things that really frustrated me during a day and get to the You gotta pet the snakes, man. That's it.

SPEAKER_02 36:10

Pet the snakes. Pet the snakes. All right, Connor. This has been a treat. I knew it would be. Thank you very much for making your time, and um you'll be back, I'm sure.

SPEAKER_03 36:21

I I hope to be back, Fraser, when we do the big reveal. Exactly.

SPEAKER_02 36:25

When we get out of stealthy mcstealth mode. Exactly. I can't wait. Thank you.

SPEAKER_03 36:30

Cheers, buddy.

SPEAKER_02 36:33

Hello again, friends, and thank you very much for listening till the end. One final thing before you go. If you enjoyed this episode, do you think you might also enjoy a monthly email from us where we share more stories, news, and perspectives from our early stage fintech ecosystem? If you think the answer might be yes, head over to our website, vestigoventures.com, and sign up for our Envisions newsletter. In addition to blog posts with our latest thinking and updates from our portfolio companies, we always include a short video interview with incredible people from our network. If you do sign up, I hope you enjoy it.