Investing with AI: From Chatbots to Agents (What Changed for Investors)

Welcome to episode one of Investing with AI Podcast for Financial Analysts.

We’ve spent years in investing, and over the past couple years, we’ve been deep in the weeds with AI. Testing tools. Working with firms. Trying to understand what matters versus what’s just noise.

For a while, most of it didn’t feel that useful but that’s starting to change rapidly. 

In this episode, we talk through what’s shifted, from basic chatbots to more agent-based workflows, and why that’s starting to matter for investors, analysts, and buy-side teams.

We get into:

• The difference between AI chat tools and agent workflows
• Why AI felt overhyped before and what’s different now
• Where AI is actually useful in investment research today
• The limitations that still exist (and there are a lot)
• How investors should start thinking about using AI in their process

We’re not coming at this as “experts” with all the answers. We’re in it every day, testing, breaking things, and trying to understand where this is going.

The goal of this podcast is simple:
Bring you along as we learn, and give you a clearer view of how AI is actually being used in investing.

If you’re working in equity research, hedge funds, or the buy side and trying to make sense of AI, this is a good place to start. 

Chapters (Timestamps)

00:00 – Intro: Why We Started Investing with AI
 00:21 – Khe’s Background (BlackRock → AI Consulting for Hedge Funds)
 02:06 – Brett’s Background (Hedge Funds → Fundamental Edge)
 03:30 – The Real Shift: From Chatbots to AI Agents
 04:17 – When AI Actually Started Working (2025 Inflection Point)
 06:11 – “AI-Pill” Moments: What Changed Our View on AI
 07:43 – What Are Agent Workflows in Investing?
 10:04 – Why AI Tools Failed Before (and What’s Better Now)
 11:43 – How Much of an Investor’s Workflow Can AI Handle?
 12:20 – Defining “Agentic” AI (Simple Explanation)
 14:42 – Data Accuracy, MCP, and Why This Matters for Finance
 17:19 – The Biggest Unlock: Using AI for Validation
 19:57 – Common Problems Firms Have with AI Adoption
 22:02 – Why Most Investment Workflows Are “Vibes”
 24:00 – Turning Intuition Into Process (Hardest Part of AI)
 26:44 – Expectation vs Reality: What AI Can’t Do Yet
 28:39 – How to Start Using AI in Your Investment Process
 30:10 – How We Stay Ahead in AI (Learning, Tools, Research)
 33:20 – Translating AI Into Real Investing Workflows
 35:14 – Why There Is No “Final State” of AI
 36:09 – What AI Means for the Future of Investing Careers
 37:30 – Outro: What to Expect From This Podcast

Transcript

SPEAKER_00 0:00

If you went into Chat GPT last year and asked the question, often the quantitative metric will be pulled from a block.

SPEAKER_01 0:07

All of a sudden, I migrated a 400-page website from WordPress to Next.js hosted on Versell.

SPEAKER_00 0:15

You know, no consultants or vendors do themselves a favor by overpromising around the capabilities of LLMs.

SPEAKER_01 0:22

How do all the pieces fit together? The data, the prompting, the custom instructions.

SPEAKER_00 0:28

Translate that to the investment process. All right. Welcome to episode one of Invest with AI. I'm here with uh my co-host uh Kay Hee, and we're really excited to start this journey. Kay, do you want to just uh kick off by giving a little bit of your background and explanation of what we're trying to do here?

SPEAKER_01 0:53

Yeah, thanks so much, Brad, and everyone listening and watching. It's great to be here. My name is Kay He. I'm based in LA. Uh I was a longtime finance guy in fund of hedge funds, most of it uh at BlackRock, where I oversaw manager, hedge fund manager research for 10, 15 years. Uh I left the industry, or I thought I left the industry at age 35 and had a 10-year period of doing various entrepreneurial things on the internet, a lot of which involved teaching, coaching, writing uh around kind of tools, productivity, self-improvement, workflow, and so on. And then I got the AI bug about two and a half years ago and kind of immersed myself. And uh since then, I started a company called Latour AI, which is a training and consulting business for small buy-side firms. Um, and we've worked with family offices, hedge funds, VC firms, and proprietary training firms. And we'll get into more of the details of what that work entails. So, what we're gonna do here is um Brett and I are we're in the weeds, we're playing with the tools, we're kind of uh AI pilled, as the kids say it, call it. And um, we are also working extensively with cutting edge firms trying to incorporate all of this. So we are happy to be here as your Sherpas, your beta testers. I won't say that we are your sages by any stretch of the imagination. And we recognize that there is no end state here, there's no steady state. So there's gonna be a lot of iteration, and uh, we're excited to have you on the journey with us. What I miss, Brent.

SPEAKER_00 2:37

No, that sounds uh that's that sound that sounds great. And my my my path in learning AI has been very parallel to Kay's. I was a investor, my you know, fundamental long short equity investor for 13 years at firms like Maverick, DESHA, Citadel, and Schoenfeld. You know, always through my career was at firms that were really insightful on the intersection of quantitative and fundamental investing. So the sort of theory through much of my career is one plus one equals three of bringing systematic signals into the fundamental investing uh process. Um, I retired from managing money in 2021 and started Fundamental Edge, which is a training academy. We train junior analysts to build models, meet we meet with management teams and deliver institutional great investment ideas to their portfolio managers, both to break into the buy-side, but primarily once you're already in the seat, how to do the job well. There was AI did not exist in our founding document of the plan of our business plan. It was something that wasn't even on my on my radar. And then come uh November 22 when GPT 3.5 came out, this became a cultural phenomenon and um just started that experimentation phase. For really three years, I was quite I was quite um uh unimpressed with the tools. And then probably summer 25, I started to see more institutional great outputs for um for these tools. And what really catalyzed this um this podcast was the agentic shift. We've gone from chatbots to agents, and I think um without trying to be, you know, we want to provide a nuanced argument of both sides here, but that has really catalyzed meaningful change in what investors can do with uh with AI. Okay, maybe walk walk me through your um your AI pill moment two and a half years ago. Like what what got you, what sort of gave you the conviction to to to you know shift your career and shift your focus with these two?

SPEAKER_01 4:44

It was yeah, it was it was funny because like you, I was pretty skeptical throughout 24. I kind of had to force myself to use it because you know I was creating content, writing, and so on. So it was definitely helpful for like lightweight marketing and writing and so on, but I didn't really have the moment. And then um, in fact, I interviewed Dan Schipper from Every, who's gonna be a future guest of ours. And my interview with him in the middle of 24 was is AI overhyped? That was the the theme of the episode. And so, Dan, if you're listening, uh I've been uh I've been converted. I'd say for me, the magic moment was in January of 25. And I had been playing around, again, mostly for writing and proposals and legal stuff. It was fine. And I took Nat Eliason's course called Build Your Own Apps. And I guess it was kind of right when Andre Karpathi, who's kind of a prominent AI researcher, released coined the term vibe coding. Now, I'm a failed computer science major. I got all C's uh in 2010, 1997 to 2001, and I never coded professionally ever. I don't know how to code. And I took this course and I was just kind of mucking around, and then all of a sudden I migrated a 400 site, 400-page website from WordPress to Next.js, uh hosted on Vercel. Like again, this might sound like foreign language to many people. It was to me at the time as well. And I had budgeted like 50k for that, 40K, and I did it in three weeks, and it's still up and running. And I was like, oh my God, this is this is no joke. So that was probably the first one. And then I would say the one that kind of tipped it with my clients was the release of uh the reasoning model. So 01 Pro, kind of in uh maybe February of 25, and then 03. I mean, 03 was kind of when I fell in love with AI. I was like, this thing is amazing. And I had started an email newsletter, a Substack that was basically trying to scratch my own itch, which was how can AI make knowledge work better? So probably a little bit more general than the approach you've taken, Brett. Uh, and all my most of my examples were finance adjacent, just given my own background. And immediately after I sent kind of like the first three issues, a bunch of my old friends from the hedge fund industry emailed me. They're like, come into our offices and teach us everything, you know. And that's how I accidentally started an AI training business. I want to ask you, Brett. Um, you had mentioned kind of the shift from chat to agents. And I feel like in 2025, the dates are so confusing now. In 2025, um, agents was the big word, right, last year. But I was always calling BS on agents. Like, no, the only agent people are using is deep research, um, where you kind of give it a task and you kind of walk away. Uh, and then, you know, Claude Code kind of came around in June-ish of last year. But I'm curious on your end, when did that kind of shift from this is a QH question and answer thing to, oh, this gets real work done. How did that when did that happen for you?

SPEAKER_00 8:10

Yeah, that's a great, that's a great um, that's a great question. My, my per I think we probably have the same personal Sherpa in the learning journey of Andre Karpothy. And he, I think he's done such a service to both of us in in terms of what he shared publicly on YouTube with many of his videos. And I just find him to have to sort of be the rare Venn diagram of technical fluency on these tools and also not having the right incentives, not trying to hype his company to raise valuation. Um, and the way he explains things is is um is very good. I think Kay and I both recommend his videos on YouTube as a good zero to one in learning what LLMs are. And I had a similar insight to you on 0103 with the deep research functionality. That was the first time where I had something in my research stack that I found to be institutional great, particularly if you pointed it towards open web information, uh, things where there was a lot of government data or something where you know the raw data was in the open arena. Deep research reports were wonderful at distillation. And for really the same time in the spring of 25, I found myself starting re research processes with deep research reports, which I found to be often better than cell-side reports and customized the specific thing. There were certain use cases. I come for the quant world where natural language uh processing is sort of a sort of a key fundamental alpha. And from a hedge fund world where we're constantly trying to understand the body language of management, that word selection really matters. There's information laden in word selection. So I think that was an initial, an initial uh use case that I found quite helpful out of chatbots. You know, did the company is the company using more bullish or more bearish language? But these were small, these were small pieces, right? Uh with my Sherpa on Andre, I think it was in October, he went on Dwarkesh and he called agents slop and said it's going to be a 10-year journey. And so I'm like, all right, well, if Andre says it, it's good, it's good enough for me. Like everyone talking about agents, I I never really understood the agentic thing in 25 either. OpenAI came out with their agent functionality, and it's just like, like, what is this thing? It just didn't work at all. Um, so I wrote it off as hype and followed Andre over Christmas as sort of went through the coding agent thing. And he came out in January. He's like, listen, like what I said in October is completely different, it's completely changed. Um, so I started going down the the um sort of the understanding of agents, and I have a full set sweat set of like evals and different things I want to try to do with these tools, and chatbots failed the majority of them. I think what when did Claude sort of come out with Cloud Code and Claude Co-work? But that was really like the big agentic shift of the Claude of the Cloud Code.

SPEAKER_01 11:03

Yeah, Claude Code was I think Q2 of 25, and co work was January of 26.

SPEAKER_00 11:11

Yeah, so the Claude Co, we like the December, January moment was like a big inflection in the agentic revolution. And even people talk about Cloud Code and an IDE, but even the things you can do in Claude Chat are now agentic. I can go build Excel, you know, build Excel things just directly in Cloud Chat. So Claude really, I think, took this big, this big step ahead from open AI. And you've seen more agentic tools. Um, it's really just the last 90 days, there was like call at 5% of my workflows, five to 10% of my workflows as an investor was the TAM for AI in 25. Now I feel like it's pushing up to 30, 40, 50, 60 percent. And the imagination you can apply with these tools now is well beyond, well beyond what you could do with um with chat with chatbots. Now there's still some issues that we'll explore in this. Many issues. Uh many issues, many jagged edges, context window limitations haven't gone away. Uh multi-document retrieval accuracy hasn't, you know, hasn't completely been solved. Um, but the pace of the pace of evolution just in six months is, I think, well ahead of the bullcase I envisioned, you know, summer, summer, um, summer 25.

SPEAKER_01 12:34

Can I ask you? Because I feel like the the word agentic gets used a lot. And what is how do you personally define the word? And then I'm happy to share my definition.

SPEAKER_00 12:45

Yeah, no, I'm curious. Yours. I, you know, I think about the investor toolbox as like workflows, kind of end-to-end workflows. Yeah. Hey, I want to go do an earnings preview process, or this company has a new CEO. I want to go and explore, you know, the you know, the background and the potential of this CEO. And within that, there can be 20 steps along the process. I want to go understand his his or her stock grant, past performance, um, do a background, you know, background analysis on if this person has, you know, nefarious historical affiliations, et cetera. In the past, I had to create each of those as individual prompts because, as you know, like a prompt past five or six pages started to lead to a garbled output with with chatbots. So I couldn't have, you know, I couldn't have build a workflow to do all of these things. And many of the, you know, uh tool providers tried to do that, chaining together prompts and building all this complex engineering around it. Now I can one-shot that with an agent, with this, with a skills file. Um, so it really just makes the tools much more user-friendly. It gives me an end-to-end workflow that still isn't perfect, but is like sh shocking in terms of the quality and caliber. So my 306 prompts have gone down to 30, you know, really like 18 orchestrated pipelines and then individual skills that I can pull off the shelf uh to dig in to dig in more deeply. So you're seeing this abstraction and the complexity of uh working with these tools. And there's been innovations around MCP and MCP and um and data as well, too. That um, you know, so sort of the one example you'd see that if you went into Chat GPT last year and asked a question, often the quantitative uh metric will be pulled from a blog somewhere, right? So if I went in and I said, what is the 10-year history of NVIDIA's earnings per share? These were exercises that a chatbot couldn't answer accurately. So it's like, if it can answer a basic question like that, like get out of my face with this AGI thing. But now I can plug uh Cloud Cowork into an MCP from DeLupa or Facet or FinChat, and that data is piped in through the MCP process with much, in my experimentation, much better accuracy. So it's not a solved problem, but there's been huge, huge innovation. How do you think about agentic workflows and what agents unlock? Okay.

SPEAKER_01 15:29

I think uh similar to what you said, so I hear kind of like end-to-end workflows being one versus kind of constantly steering the model. Okay, now tell me this, now tell me this. The analogy I use is uh it's like the difference between Waymo and Google Maps, right? They you know, Waymo, you give it the destination and then you just kick back. You don't really care how it gets there as long as it gets you there within a standard deviation of your expectations for time and safety, right? And so with the agentic, it's like, hey, I would like to, you know, um write this deck. I don't really care if you do slide seven first or this, but I just have a general uh intent of what I'm looking for, and then the model kind of figures it out, right? So that that would be kind of like the end-to-end, and obviously you mentioned the connections uh as well. The other thing that uh that I would add with um with agents is that they tend to like self-correct along the way. So you might give it a PDF that's too big, and then without telling it, it will run like a PDF extraction tool, right? Like I don't even know that the PDF extraction tool exists is a thing. I don't even know the PDF's too big. But the agent is making these decisions and what you know in AI language they'll say tool calling, right? And not just tool calling like Slack or Delupa, tool calling like these weird kind of utility help air apps to like chunk a PDF or to use what's what we would call command F, you know, is called grep in uh you know in computer science land. And so the agent not only can it get you answers, but it can find uh and solve its own problems by accessing new tools, which again gets to that point that you mentioned. It makes the end-to-end workflow more accurate and longer running. So in many ways, like kind of full agreement with what you said, I would add kind of the tooling element that that is kind of uh abstracted away that like I don't have to tell it to use the PDF extractor, uh, and then the ability for it to reason and act and self-correct in the process. Yeah.

SPEAKER_00 17:43

And I'd add probably one of the it's it's it's not the flashiest uh AI pilled moment, but one of the big AI pill moments I had this year was the usability of agents for validation. And so we we work in an industry where um you know accuracy is critical. We have a fiduciary duty to our investors um uh to conduct a rigorous investment process. And if you follow any of the finance benchmarks of LLM accuracy, they tend to sort of shake out in the 60 to 70 percent range. So I'm always like, all right, well, yes, for qualitative use cases where I'm putting in six transcripts that I can go and validate in the transcript, sure, right? But for quantitative use cases, that was a very difficult mental hurdle for me to get over. Like I don't want to unleash LLMs on financial models where it's 65% accurate. Like the exercise of finding 35% errors, like next word prediction, next token prediction, next word prediction, it's fine. Next number prediction is not fine. Like we don't we don't want a stochastic exercise to sort of make up a plausible sounding number, right? And in a few of these use cases, like LLMs are very good at making something look plausible, which means that where I've seen quantitative errors, sometimes a quantitative error looks more plausible than the actual number, right? And so this was something where I was like pump the brakes on quantitative use cases in 25. So what's changed is MCP, the data's gotten more accurate. But the ability to build agentic, I was using this like, okay, let's check these numbers, use these four LLM to check this LLM and go back and forth. They knew it was okay. But the ability to create agents to validate work now, it's exactly how you how you pointed out to sort of go in, tool call, check all these numbers, right? Even give me a hand like a checklist by hand to go in and spot check numbers that that could really be problematic for my thesis. That to me is a big unlock with agents that didn't exist in in in chatbots. Um I'm curious, Kay, to understand a little bit more about on the on your consulting side, are there three or four common themes you see come up from clients in terms of the problem space that uh the firms you work with have in general that you help them help them walk through?

SPEAKER_01 20:20

I think the one of the big ones, uh, believe it or not, is just I would say kind of a failure of imagination. Uh and what I mean by that is that you're kind of flashed with this, you know, you're you're presented this blank flashing prompt. And it's like, this is, you know, 200 Nobel Prize winners at your disposal. Go nuts. And what happens is people kind of anchor to the existing mental frames that they know, and that's like Google and chat. So immediately you start off on chat. So that would be kind of the first the the first challenge, which which we kind of help people shift into that agentic mindset. Um, and then you treat it like texting, where it's you know, one one one sentence prompts instead of a two-hour prompt, which again, prompting has shifted, but it's not like an iMessage. It's much more like a brief in many uh in many of these cases. So I think that's the the one of the biggest ones is to kind of like unlock what's possible, right? And so, you know, I joke with people, it's like you now have a data scientist in in your pocket, right? But what does that mean? That means you need to have the data that a data scientist works with. You need to understand the language to communicate with a data scientist, but more importantly, you need to look at your problem space and understand what a data scientist could be useful for. And if you've never worked with a data scientist, and you might have at some of the could at some of the firms you have, that is a completely different world. Right. And so I really kind of think of it as the expansion of kind of what's possible. That would be one thing. The second thing is usually just like how do all the pieces fit together? The data, the prompting, the custom instructions, the memory, the statelessness of a of an LLM, projects. You just kind of like, so I would kind of put that in kind of like marrying the workflow with the tooling and the features. And by the way, there's like, you know, Anthropic's been on a tear, but they've they've launched a new two new features a week since January of this year. And we're in April. So that would be another. And the third is, and this is actually a harder one, is I think I saw this quote the other day is so much knowledge work is just vibes and spreadsheets. And um people don't actually know what they do for work. It's the vibes. They're like, okay, you know, it could be anchored around thing like quarterly earnings. Like there's an there's a huge, they know what they do there. But if you ask them, like, what's your screening process? Like, oh, like sometimes I look at 13Fs, like we'd go to these idea dinners. There's not like an actual, it's vibes, right? Sourcing happens on vibes. And this actually isn't an LLM question. This is much more of a process. I don't even, I'm not a, I'm not like a formally trained consultant, but it's like a process excavation exercise for lack of a better term. Yeah. Or maybe workflow mapping, right? And I think that, you know, it's it's okay to run with vibes when, you know, without LLMs, because you can kind of like flow back and forth and you can kind of work in that kind of osmosis, a little bit reactionary way of doing things. But with LLMs, you need to be clear on your intent. You need to be clear on your steps. You really need to like deconstruct things. And um, that is that's the hardest thing with my clients because it it every every analyst is, every firm is different, any every analyst is different. Um, and mostly, you know, that is kind of uh it's a very kind of human exercise of like extracting out knowledge and then remapping it towards an LLM.

SPEAKER_00 24:22

That's such a that's such a salient point. And I think that's where you know I'm spending a lot of my time, and I completely agree. It's almost investing after you've done it. It's why it's hard to learn, right? Because the, you know, really great PMs will have a process, but the process leans more towards in an intuitive, uh, sort of tacit understanding than anything that's like overly robotic, right? Almost no PMs have a 24-page checklist of if I look on an idea, I go check off every single step. They've done it hundreds of times. Yeah. And like a master playing a piano, they look at an idea, you know, hit 72 Bloomberg functions, update the model, call the man. And they just sort of do it out of muscle memory, yeah, such that the workflow is tacit, not explicit. That's really hard to build an agent around. You kind of have to slow down and put document that process. And that's really hard to do because it's a nonlinear investing is a nonlinear process. Um, that's what we've done for four years at Fundamental Edge. Like we've tried to create a linearity out of it to bring to bridge that knowledge of the junior analyst to the PM. It's not perfect. Um, but uh I completely, I completely agree that um that that's a big part of the journey of enabling agents in your in your process, documenting your process, also documenting LMs are wonderful pattern recognition engines, like the mental models in your mind that you come back to over and over, how do you codify those such that an agent can think the way that you the way that you um think?

SPEAKER_01 26:10

And have you found, because right now, let's say there's a 20-step process and LLMs might be good at three of the 20 steps. Have you found that to be frustrating for people? They're like, it's kind of like the junior person, like, oh, I'm just gonna do it myself. You know, how do you how do you find that tension? Because I found that people, you know, you go in for the first time, like, I want this to like pull all this data, do this, do this, send me a report that does this, and then you know, send me a teams message every time the stock's down like X percent, right? And like, whoa, whoa, whoa, whoa, whoa, whoa, whoa. Like, let's pick the one like thread here that is this that lines up with the strength of the LLM. Like, how have you kind of has the expectation versus reality? I said like the Tim Urban kind of delta, how has that um uh manifested with uh in your client work?

SPEAKER_00 27:04

It's it's definitely a journey. Um, I think you know, no consultants or vendors do themselves a favor by overpromising around the capabilities of LLMs. This has been rampant amongst the vendor community to say, hey, our tool can do anything an investor can do. The firm, you know, splashy demo, adoption, disappointment churn, right? Has been this sort of cycle for three years. Um so we try to really be very transparent around we have a red light, green light, uh, yellow light framework of like what's possible today, what's sort of like close, like, you know, you know, maybe, maybe not, and then a red light. For example, an LLM cannot one-shot a complicated three financial statement model today. It can take a model that's you know five years out of date, update it, resegment it, find the three key drivers, uh, key in the thematic leverage, and then run automated research around those ideas and feed those, feed that those findings back into the financial model and compare my model to consensus. That's sort of like one of the holy grails of agentic systems. It's not possible today. It's not possible today. So, what we do is we try and chunk down that process to like the 17 steps you would need to sort of have the pillars of running one of those holy grail movements. And we just experiment. And like the frustrating, the frustrating thing with AI sometimes, and exciting thing is something that doesn't work today could work in three weeks. Or sometimes what works on Tuesday doesn't work on Thursday because you know, anthropic decides to throttle compute. Um, so it's this messy evolution, this messy journey of what's possible today. Now, you know, what we try to do is find those workflows today that pass that aha aha moment. Hedge funds are hard to impress. And chatbots, like for the most part, were not impressive. You're starting to see more workflows where the outputs are impressive. So I'd say, like, as you're thinking about any investor adopting these tools, you can't do it all at once. You can't rewire your entire investment process. This is likely an 18-month journey, nine to eighteen month journey. So find those individual workflows where you have the agentic, the agentic overlay, right? And for the first period of time, parallel process it. Don't change the way you're doing it. Just have an agent run in parallel, as sort of, you know, we use the analogy of bumper bowling. Like, you know, have have something, have bumpers behind that process that can catch any process, you know, uh mistakes that you might might make. Um, so I think it's a slow journey. It's not gonna, you know, finance is not gonna flip the switch overnight the way that engineers have, with this much more complicated game than building apps uh with coding agents. Um so um the the challenging part part of like you know why we try to provide value is like it takes a lot of hours just to stay up with this stuff, right? Like, I don't know what um I'd be curious actually, like what's your learning journey? Like you're you're you're sharing all this with the world via your your blog and and social. Like, what does your week look like in terms of what you're reading, watching, you know, listening to to stay up to date?

SPEAKER_01 30:30

Yeah, uh it's funny because we were texting yesterday. Um, I have anxiety that I'm behind. Uh I have two anxieties. I have anxiety that I'm behind, and I have anxiety that I'm not using enough tokens. Um, and so I would say I've probably, you know, I'm probably working eight to ten hours a day if you count um kind of consumption of YouTube videos and podcasts, like in that in that number. And it's a mix of, you know, a lot of my learning is is with my clients, right? And so that's part of it. I write two newsletters, I write a newsletter twice a week. And one of them I basically try to provide a use case or uh evaluate a feature. So I'm kind of dog fooding stuff as it comes out. Um, and then like the part of the excitement of starting this podcast is like, you know, oftentimes I'm paid to be the smartest AI person in the room. And I I don't want, I don't want to be. Like, I want to be the dumbest. Uh, so that I'm, you know, my baseline gets elevated. So, you know, again, a lot of conversations. You and I are similar where you will see someone interesting on Twitter with a unique perspective and just DM them, be like, hey, do you want to chat? Like, no agenda, just want to learn from you, hear what you're doing. A lot, I do a lot of that, as I know you do uh as well. Uh, shout out to our mutuals who have been on the receiving end of that. And um, and yeah, I would say, you know, uh there's a stable of newsletter. I mean, Twitter is a phenomenal source of uh of information. I don't love Twitter. It's kind of like a uh the whole the whole apparatus of it, but I think it's phenomenal um for AI and getting me angry at you know clavicular or something. Um, and then a few newsletters that I read and I listen to uh a lot of uh podcasts like Lenny's List, you know, Dan Schipper, Dwarkesh, uh Clairvaux, um, a lot of practitioners. There's also people that I'm on the lookout. Whenever they do a podcast, I just go listen to them. Like Brett Taylor, the chairman of OpenAI, would be probably the most common one. Andre Karpathi uh would would be another one. Yeah. How about you? Do you feel you feel uh you feel behind?

SPEAKER_00 32:42

Yeah, same. I mean, part of the reason I've sort of going down the AI rabbit hole is just to manage my own existential angst around the future, the future of the industry. And you know, I have three sons, 12, 10, and 8. And my 12-year-old son in particular is he's got the stock bug, and he talks about working on Wall Street. I'm like, man, I better like learn this stuff so I can shepherd him through this process. And I teach at Arizona State University to finance undergrads. I start class again in August, and I just want to be able to give them a framework for navigating through. Um, so step one is I have to figure that out on my own because I don't think anyone knows uh what's the right you know, career advice other than to go be a plumber or electrician. Um, we'll certainly need more of those. Um so I think we're trying to just figure figure this out together. And I think that's exactly right. Part of our objective here is to find the 36 um sort of most interesting people at the intersection of investing and AI and have conversations with them in a casual way that we Kay and I are individually having on a one-off basis and just share share that. And I think the mental model for me is anytime Andre Carpothy tweets something, I also really like Aaron Levy's uh content from Box. He's really thinking about enablement of AI at the organizational level. Anytime e either of those individuals tweet something, my mind starts racing. It's like, all right, here's the raw, like great insight. Let me think how to translate that to the investment process. And so our ultimate objective is to be a venue for people to come and learn together with us about how to take these cutting-edge concepts that you know, I think hopefully we're humble enough to not think that we know everything, but to sort of translate that in and play with those, experiment with those concepts, uh, to then translate that back to uh the investment process and hopefully give you know our investor audience lever you know time leverage on that because you know it's scaling Kay's 40 hours a week, learning AI, my 40, like that doesn't scale. I've been a PM, I've been at like you're trying to, it's earning season now, like it's uh trying to learn AI while dealing with earning season and incredible, incredibly volatile macro, you know, geopolitical markets. It's just a very hard thing to very hard thing to do. Um so that's really that's really the objective. And you know, we're trying to have people from AI people from firms and from vendors and consultants and academics and and push them on, um push them to sort of bring helpful things to the audience.

SPEAKER_01 35:34

One of the uh on that point, you know, one of the things I push back with a lot of you know potential clients conversations I have is this this idea that there's gonna be like a final state, you know, like there'll there'll be a button that's pushed and like AI is ready. And um this podcast is is kind of the admission that that's that moment's never gonna happen. No one's gonna ring a bell and say, like, all right, time, you know, wave in the AI, right? It's gonna be incremental, it's gonna be jagged, it's gonna be five steps forward, two steps back. Sometimes there's probably gonna be a five steps forward, ten steps back during this run of the podcast, right? With maybe it's like cybersecurity, right? Uh, we have no idea and we won't pretend to know that that we do. But what we kind of do promise you is that you know, we've got our ears to the ground. It we're getting paid to learn this stuff. And so we're gonna be here distilling it uh for you to give you, as Brett said, that time leverage.

SPEAKER_00 36:29

Yeah, that's exactly right. I mean, um great investors don't just pick a case and be like, I know that case is certain. They sort of think about a ring, a fan of outcomes like bull base bear. And like with AI, virtually every dimension of society and markets and Wall Street jobs, like the fan of outcomes is much wider. Um and so how do we navigate that fan of outcomes and understand what AI could mean to the future of Wall Street careers? What could it mean to uh the harvesting of alpha pools? What could it mean to the future of fundamental investing? These are these are things we can debate, but no one knows, right? Um you know, the pay the pace and rapidity of improvement has been striking to me. Um but uh yeah, I think that's exactly right, Kay. We're we're just gonna try and learn together with with everyone. Um so um yeah, so that's really what what we're doing. We hope you come along for the journey. This will be this will be fun. This will be participatory. Uh so um we're filling out the guest lists, the guest list, um, the guest list now. And um, yeah, we just want to sort of drive a service to all the other learners who maybe have the same existential angst about AI that that we have. Couldn't have said it better. Yeah, excited to be on this journey with you all. Awesome. Well, thanks everyone, and uh uh keep a lookout for more episodes soon.