Loading summary

Transcript28 lines

[00:00]
A
Foreign. Welcome to Reshaping Workflows with Dell Pro Precision and Nvidia, where innovation meets real world impact in high performance computing.
[00:20]
B
Welcome back to another episode of Reshaping Workflows with Delpro Precision and Nvidia RTX Pro Pro GPUs. I'm your host, Logan Mahler. Today, very interesting topic, something we, we haven't really talked about before, which is investor relations. So with that I have Kathleen from Adair that's going to go all into how AI is shaping investor relations. So Kathleen, welcome to the show. Maybe take a few minutes, a little bit of background on yourself, what you do at Adair, and then we'll jump right into it.
[00:53]
A
That sounds great. So I'm Kathleen Perli. I'm the co founder and CEO of Adair. My background's actually in linguistics, which is what I studied in college. I then did a Fulbright in the impact on phonetics and phonology and second language acquisition. And while I wanted to be a cool PhD from the start, and my parents had other plans and I had to get a job, so I spent a good chunk of my career doing propensity modeling in the healthcare ad tech space and have really been driven to this idea of the fact that two individuals can say the same things and be misinterpreted. And so really diving back into that, we built Adair as this intelligence and evaluation layer for capital market communication. And the short of that is, you know, every quarter CEOs, CFOs and other executives get on these earnings calls to talk to investors. And the way that they say things, the hedging, the tone, the structure, it all carries signals that machines are now listening for. And so we really help ensure that executives are getting credit for the value that they're creating within their organization. In addition to all of my fun work at Adair, I also am a professor of generative AI at the Rice School of Business and also work with the deans on helping them understand how to navigate, ensure that we're preparing students for the future.
[02:11]
B
You gotta play a better background than me. So, okay, so let's start kind of with investor relations, right? So I work at Dell Technologies. You know, every quarter we have our earnings call where CEO, cfo, coo, go on and talk about how did the company do, there's questions asked, et cetera. I think everyone's kind of familiar with that. But let's. You said something really interesting, right? Is that the way things are said can carry like, you know, two people can kind of say the same thing. And have a different meaning. Before we kind of dive into Adair, the platform, what you do, do you have an example of that? And I mean you don't have to name names or you know, give away anything proprietary or any like that, but maybe set the context of like where you've seen an example like that maybe an investor relations or you know, maybe an ad tech drawing upon your background.
[03:00]
A
You know, one key, one that I see oftentimes is with non native English speakers. So one of the things that we found in our early research was non native English speakers. When you look at their tone, their vocal burst, intonation, you would find things that would indicate uncertainty or lack of confidence. The street, the institutional investors were oftentimes pricing that in as a lack of confidence in the business or the underlying business strategy. But when you actually hear those executives talk, talk and their native tongue or their, their kind of first language, that actually the intonation pattern actually disappears. And so there it is in the margins, if you will. But they are getting penalized for something that's completely in essence out of their control. That is no reflection on the ability for the company to execute. And so what's interesting is, you know, we've built a ton of infrastructure around financial reporting for decades, but we haven't really built a ton of infrastructure on how that information is actually communicated out and, and interpreted. And what we've seen is, you know, we have internal IR teams still leveraging tools that are from the analog era, if you will. So they're still leveraging spreadsheets, word docs with track changes and comments. But the markets have moved heavily algorithmically. So we're seeing about 70 to 80% of all trades being algorithmic. About a third of all sell side analysts are leveraging AI as that first pass. And you're seeing activist engagement and firms as well as individuals that are looking at compliance constraints using AI as a way to help target and identify kind of that first list to take a deeper dive on. And so I think one of the big challenges is we're not all being scored equally first. And the second is our audience has also expanded.
[04:50]
B
Yeah, that makes total sense. Right, like, but help me understand like how does Adair kind of fit into that? Right, because my nascent knowledge of kind of investor relationship, obviously I'm not an executive, but you know, that call occurs, it's kind of happening in real time. Right. Does a dare kind of fit into the pre planning like the practice and the prep that leads up to it? Is it more kind of a middle layer in between translating as things are kind of happening live, or is it more kind of the post type reporting or is it kind of all three?
[05:23]
A
It's all three. We kind of sit as a layer across kind of the beginning stages of the prep, all the way through post analysis and insights for roadshows, if you will. So post call conversations and meetings and conventions that happen. And what we do is we treat communication like a system you can test, you can simulate, and you can optimize before it hits the market. And we do this in two ways. Given my background in linguistics and really diving into my own academic research as well as other peer reviewed academic research, we've developed a C3 communication index which really takes each speaker and provide them a score around clarity, which really looks at can the market process IT control, Are you owning that narrative? And the third one is around confidence? And, and that's really structured around are you signaling the right conviction? Now these three scores are optimized and are not generic. So obviously what your, your CFO will always have to be more balanced than your CEO. What you say in tech would be considered evasive in energy, for example. And then it also needs to be tuned to the financial fundamentals, because when you think about confidence and you think about the algorithms and how they're parsing language, it really needs to take into context what are the commitments that have been made in the past, how does this trend quarter over quarter, and what do the financial fundamentals look like? You know, we are seeing, you know, if you are reducing modals, for example, you are getting a little bit more of a kind of movement to the right on stock performance. But it needs to be taken into context of how you perform, because the worst thing that you can do is have weaker financial fundamentals and try to sound really confident because you'll ultimately just end up sounding tone deaf. And so that's one big component. On the prepared side. We also look at, you know, how this will land with the sell side analysts. So we actually will simulate through digital twins of each of the sell side analysts how they'll likely respond and what questions will happen before the call even occurs. And so this allows the executives and the teams to really think through the different potential pushback points. What do they might want to adjust in their prepared remarks versus really just have better understanding of how that needs to be communicated in the Q and A. And we look beyond just what questions have been asked in the past, what's getting asked on their peers, but really looking at how do they view macroeconomic components what is their consistency in terms of re rates and what usually triggers a pricing target change and how do they want to be communicated to. And so this allows the executives to have a little bit more confidence going into these earnings calls calls. And I think it's very similar to what you see in any business. We spend a lot of time trying to de risk every aspect. So whether it's the supply chain, whether it's financing, but the earnings call communication is that last kind of what I'll say, gate, if you will, that hasn't been de risked yet. And that's what we're really helping our partners do, is helping them de risk that and ensure that. And how we think about it is the company's execution creates the value, their communication within the earnings call and afterwards really dictates what they get credit for as it pertains to that. And so really trying to remove the gap between what's communicated out and what's understood is a huge part of what we do.
[08:45]
B
Makes total sense. I mean, the first kind of the example you shared is making sure, kind of the message from whoever's speaking matches the facts, right? Like, you know, for example, I like to think I'm a pretty confident guy and sometimes I'm flat out wrong, but I just communicated with a bunch of vigor and people usually give in. But, like, if someone actually knows the answer, which, I mean, obviously an analyst, you know, at some sort of investment company would obviously know that and would take that as like, you know, dodging or et cetera. But the point you made, I think is really, really interesting and I definitely want to dive in, is being able to create kind of, I mean, digital twin means a little bit different, right? It's usually a 3D representation, but creating almost a Persona of what analysts will be kind of on the call to be able to have whoever speaking during the investor relations call be able to kind of see how it lands, right? Like, I would love to hear a little bit more. I mean, don't give away anything proprietary. But, like, how do you go about doing that? Like, how do you create something that can mimic, you know, their thoughts, feelings, their views on, you know, the market or the specific industry that they're in or, you know, even how they would perceive something. Like, how do you go about doing that? That's so interesting.
[09:59]
A
I think there's a lot of that goes into, you know, as we think about it from a psychology perspective, behavioral economics as well, and really parsing. I think one thing that is oftentimes missed is you learn A lot about an individual through the way that they communicate and how they structure even something as simple as like sentences in terms of how they want that information to be fed back to them. Because oftentimes they're giving you that playbook and sometimes it's very explicit. So you look at the sell side reports or you see their latest commentary that they were spotlighted on from cnbc. And so we're aggregating a significant amount of data and then layering our own understanding of linguistics, behavioral psychology, behavioral economics on top of that to understand and build these Personas, as you mentioned. And what we do then from that standpoint is we're updating these in real time. So these are not static, but they're really understanding, you know, what questions are they asking on your peers that happened within 12 to 24 hours before your call. How does their tone within that question deviate? One of the things we look at is divergence. So if the written sentiment of a statement or question from a sell side analyst differs from the vocal, we identify that as a flag of something that might be unsaid. And so what this really does is help prepare them. We have about an 85% accuracy for question prediction for our partners. And what I think is more valuable and more interesting is that these Personas and these digital twins, if you will, we actually force them to produce questions that not only look at what is likely to get asked, but what are the low probability but high risk questions that might get asked. And what we're hearing is that about 30% of the questions our agents are producing are things that the internal IR team or the executive team hadn't maybe considered that they're really glad the agents are bringing to light, even if they never get asked on the public facing call.
[11:59]
B
That's so interesting. So I, I did not realize that's how the process worked. I mean, once again, not an executive, but I kind of figured that questions would have to be submitted in advance and, you know, approved. But like an analyst can really ask whatever they want. And you said 85% of the time you can kind of source that topic or something that they might ask that might have been not planned for. Basically.
[12:24]
A
Yes, absolutely. And I think what's interesting is, right, markets are creating arbitrage off of the nuance within your communication. They're looking to identify, like, what edge do I have over the next sell side analyst or the next institutional or the buy side? And so these small clarity gaps in executive communication widen that modeling dispersion. It can potentially erode some credibility and amplify some of the volatility. And right now, there's just been, to date, no infrastructure to evaluate that risk before it even happens. And so what we've really done, and you've seen a lot of this technology be used on the other side, so not the ones that are issuing. So the company side, which is what we focus on, you know, the ones issuing the statements are not really going into this depth, but you're seeing it more on the buy side or the sell side with their quants. And so really understanding that you are talking to two different audiences now is a huge component of knowing the fact that, you know, we often will see companies beat expectations and still underperform. And the gap is rarely tied to the financial fundamentals. It's often tied to the interpretation. And so we'll see sometimes the verb, like, if they're too verbose and you think about auditory processing and the unconscious friction that you're creating and sentiment that you might be creating with listeners to, you know, if you have a really strong fundamental, but you have a significant amount of hedging still involved, that might indicate, you know, there's another shoe waiting to drop so what's not being said. And so we really find that the earnings call is something that happens every single quarter. It's a repeating component, so we can actually test for it. And it's probably one of the highest stakes reoccurring moment in business that there hasn't been an infrastructure in place to really help support executives land their communication as intended.
[14:19]
B
Makes a lot of sense. Like, I mean, you're right, it is kind of quarterly, traditionally, and repeatable. And it's. It's something that's done. I'm curious. You know, I. I find it completely fascinating, you know, maybe more a little bit. I mean, that's. You don't have to go super technical on this. But, for example, you know, I'm assuming that there is more than you just pulling down the latest model from hugging face to do this right, there's got to be some sort of, you know, proprietary training models that you use to understand how the market will respond to. You know, for example, one of the examples you used was, you know, results were really good, but they were too long. I mean, they just spoke way too much and were way too boastful. And then the, you know, the. The street slash, you know, the market in general interpreted that as, you know, something negative. Right. Like, how did you go about kind of creating, let's say, the intelligence layer to be able to understand and kind of predict what you do in the Adair platform.
[15:21]
A
I think, you know, it's interesting because everyone talks about, you know, is this idea around, you know, building proprietary models and components of that nature. But I think what people often miss is that the value isn't in the engine itself. We given the security of what we do, we run on premise for our clients due to the data sensitivity. And so we've actually built and fine tuned open source models. That's actually where we started. But what we find is that oftentimes IT and security individuals within an organization have a lot of constraints around that and there's a lot of testing. And so the speed to value is a little bit longer. And so what we've done is we've really created a model garden, right? So it's more of a bring your own AI, if you will. And the value that we layer in is around our agent orchestration, our ingestion and our pipeline components of it we deploy through Dockers and Kubernetes. And so it can allow, and I won't say, because my engineering team will kill me when I say this, but it's not plug and play in the sense of, you know, if a company has a specific model they want to leverage, we are able to adjust, integrate that, do evals to ensure performance. Because the value of what we bring is more on the ingestion layer, some of the algorithms and some more components within that, rather than just strictly the underlying model itself. We see that being more of like, if you will, you know, a thermometer can tell you what the temperature is, but can it diagnose what's wrong within a patient or what we should do about it? And so what we kind of do is we actually utilize quite a few, a significant number of agents that are digesting certain components of information. Our agents will then create certain artifacts that then are used by other agents to then produce certain components of the different aspects, whether it's the prepared remarks themselves, understanding of the analysts and where they are, the quarterly backdrop, what questions are going to get asked, how to answer those questions, tone and pacing, things of that nature. And so I think when you think about the value, and it's one thing I push my students for, is the value today is not necessarily in the underlying AI model itself, but in the workflow and architecture around it. And I believe, and this might not be well liked by a lot of engineers out there, but it is a strategic role, not a strictly an engineering role.
[17:44]
B
You know, here's kind of a Question for you. So I get there's kind of, I mean kind of an agent based system in a sense that's looking at different components and, and that how do you handle something like. And this is Logan Lawler's point of view, this is not Dell's point of view. But you know, you look at the S&P 500 right now, I mean, I don't think it's at an all time high, but I mean I think we're pretty close to it at the time we recorded this. But the market feels kind of choppy. It feels a little, you know, weird that we've got a lot going on in the world which we won't dive into, but the market's at an all time high. Like how do you go about kind of predicting those? I mean, I'm not going to say unknowns, but it feels very weird, right? I mean this isn't the rise that we had in the mid 2000s where everything was just kind of up, up and up like, or not 2000s, but like the mid, like say 2014 after recovering from the financial crisis. You know, everything was kind of up, up and up and it just kind of followed that trend. I mean we're still going up, up and up, but it doesn't feel right, I guess. How do you even predict that or how would you even layer in that level of intelligence into a dare?
[18:48]
A
So we do look at things like the vix, we look at our, the peers in a certain segment, the S&P 500, their closest ETF for normalization and things of that nature. And I think one of the things that is really interesting in terms of we pipe in real time data as well. And so I think, you know, while there is, you know, we have a very strong market right now, there is a lot of uncertainty and quite a bit of volatility. And what we find and what we focus on is the goal of the Adair platform isn't to, you know, if you will, it's not about sounding better, it's about being understood correctly. And so when we think about it from that perspective, you know, we're not a magician. If you have a bad financial fundamental quarter, like we can't make it a great one as it pertains to how the stock moves, institutional investors, kind of what the sell side analysts will do in terms of RE rates and price target changes. But what we do is we really focus on the margins. Right. And so we're able to attribute roughly for most of our partners, what we've seen is about a 30 to 43 basis points movement to the right. And so it's really kind of like what you see in terms of like a volume adjustment. And so when you layer in market volatility and uncertainty, you need to control for everything that you possibly can. And so what we do is really focus on kind of adjusting that volume of that noise. So if it's a bad financial fundamental quarter or material miss, we turn that volume down. If it's a great quarter, we help you turn that volume up so that you get the most like you create the value. As I talked about earlier in your operations and your execution, the communication dictates what you get credit for. And that's what we're really focused on, is those margin components. You can think of it very similar to performance medicine for elite athletes. Elite athletes don't make it to the Olympics because they're bad at what they do. They lose in the margins. It's a shave of a hair here, a shave of there, a slight increase to speed or start that then changes their ability to podium or not. And that's what we're really focused on, is really controlling for those margins. And those margins at Fortune 500 companies are significant when you think about a 30 basis points movement and the ability to kind of control for that.
[21:13]
B
Here's kind of another question. This is, I'm kind of thinking about Adair and like how it operates, right? I mean, you focus kind of on investor relations, which is obviously very important, right. But like I could kind of see this going beyond, you know, just investor relations, right. Like for example, you know, I've been at a lot of conferences, you know, like a GTC and stuff. And based upon, you know, Jensen's keynote, for example, that might move the stock price depending on what he says or doesn't say, et cetera. And that's kind of the same for any big kind of event where a CEO is front and center speaking at maybe their own trade show or you know, an industry trade show outside of a typical investor relationship. Is Adair something that you can use for that or is that something you've thought about or is it focus only on that, like once a quarter, quarter closed investor relations call?
[22:04]
A
No, we really think about it in terms of every high stakes communication moment that has millions and billions of dollars on the line that will impact the capital markets. And ultimately each one of those high stakes communication moments will need to be simulated before it happens. And we've, we do support, obviously earnings calls is kind of what we find as our initial Wedge into talking to companies and CEOs and CFOs and getting them comfortable with the technology and understanding how important not only what you say, but how you say it. And we've been leveraged across investor days, M and A announcements, crisis comms. We've done a significant amount of work also within and are expanding our work within activist engagement. So we're seeing, I think 2025 had an all time high of activist engagement for companies and so really helping understand how executive communication, like you said, whether it's an earnings call, an announcement, developer day, how that will move the market and understanding and testing it before it even becomes a reality, I think is a huge point of where we need to go as a, as a linguist in a world of finance bros, if you will. I've always had a passion for turning language into math and this is just starting to validate. You know, we've always had a golden record for or source of truth for numbers. But the level of communication and whether or not we measure it and how we think about it and how we de risk it has never had that visibility or stage presence, if you will, in terms of importance. But what we're doing with Adair is really ensuring that not only do we have the ability to make the invisible dispersion visible in terms of what's spoken versus what's understood, but how companies can kind of help de risk that communication even further.
[23:58]
B
So you mentioned kind of golden record, right? Is there such a thing, you know, in a CEO or a CFO communication in the way is there like the perfect way to communicate something or is there, is there a range? Right. Like for example, I've, I've listened to a few investor relations calls. Some people are extremely passionate. Some people are just very, you know, monotone, give the facts and move on. Like golden record, obviously with numbers, I mean the numbers are the numbers and that's it. But the interpretation of the numbers. Is there a kind of a golden record in how to position and position the numbers in the context of like a business performance? Yes.
[24:37]
A
And that's what we're doing through our blueprint narrative spine and our C3. Because one of the things that we want to ensure is that the CFO or CEO, who's extremely passionate, when they start leveraging a data, the worst thing that could happen is this whiplash where all of a sudden they don't sound like themselves. So our goal isn't to ensure that everyone sounds the same. That would be actually counterproductive to what we're trying to help executives achieve. But what we're doing is really focusing on what does their voice sound like, what does their communication typically sound like, what is the best version of that, given the context of financial fundamentals, industry role and the macro favorability, if you will. And we provide ranges. So within our C3 communication index scores, each speaker gives a targeted range that is taking into account all of those components and where they've been historically and where their peers are. Because I think one thing that we fail to lose sight of is narrative is always taken in context of something larger. And when you look at standard generative platforms, you lose sight of the context or the nuance that it's required. And that's what Adair is really focused on, is not just generation, but this council layer that pushes it into a way that allows to create a repeatable event where executives can actually start to experiment and test and measure the impacts.
[26:07]
B
So a day or let's say, you know, you have a new client, doesn't matter who it is, they have an investor call and you know, they deploy a day or they're set up, is it really kind of a, kind of a two part question, Is it kind of a single pane of glass where all of that kind of prep occurs versus more the, the analog, like you mentioned, kind of spreadsheets and you know, shared files with track changes and stuff like that. And then how many times typically would you know, you more or less practice does? I mean how many practice runs does it take to get it right, I guess or get it to where you want it to be.
[26:43]
A
So I would say this, there is like this centralized hub for it. So it'll show you business unit data, It'll show you C3 scores across your past and your peers prior commitments that have been made so you can understand as you're formulating the next earnings call. What do we say last quarter? What is the street going to remember and hold our feet to? So there is that single kind of collection of this like operating system we tie into the finance team. So when they update a number, it's actually tick and tied into both the remarks and the Q and A and suggested responses. And we also maintain, so legal gets in there, communications gets in there, compliance so that they all of the commons, track changes, edits, revisions, everything can be seen. Adair actually often operates after the first pass of the prepared remarks as another editor that you can kind of select, copy and ask for input on. And what we find is that twofold one, this allows for a lot of the removal of back and forth, but really pulling into context the full, bigger picture of everything that goes into it and understanding how that operates. So one of the components is that Adair provides is like an auditable report post call where you can see every single change that has happened, which is really important given some of the regulatory constraints on this type of communication. Now the other component of this, which I think is something I'm really excited about and more excited than I probably should be, but this just tells you how much of a nerd I am is really looking at when the. When you think about an earnings call process today, not only is the materials oftentimes scattered, if you will, you're oftentimes not getting final numbers until 72 hours before you know whether it's looking for tax information. EPS is typically one of the last things that comes in. And so I always tell individuals it's this living, breathing exercise. So when you talk about versions, I think sometimes you'll go through. We've had clients go through 37 versions of prepared remarks before the final and it might be a shift in geopolitical components or they'll do a version. And what's great about Adair is the speed at which it helps provide these simulations and different test scenarios allow them to do okay, well if this legislation passes, that's going to change this. So let's do one. And we'll find out three days before the call. Right. So it allows for that kind of variability, but the ability to adjust fairly quickly as well. And so we do see a significant amount of that. We are also providing kind of in your post call roadshow or analysis, whether it's rate of speech, how to frame some of the communication. I joke. I love that book in college, eat this. Not that, but kind of from a, from a communication instead of this, say this. And we're also starting to integrate live real time Q and A support as well.
[29:43]
B
It's so fascinating. And you're right, like I'm a nerd too a little bit. I mean obviously maybe not as smart as you, but like I do find it quite interesting. I didn't before this fall know like what went into an earnings call when it makes total sense that there's like 37 or whatever you said, like numbers of reviews because things are changing up, you know, kind of by the minute. So you know, so far it's been a great conversation. We're getting kind of towards the end of it, like what I like to do is, you know, kind of to wrap up the call, give kind of the audience and pretend that they just kind of tuned in right now. What are kind of the big three takeaways that you would say about Adair and how it is kind of reshaping how any, I mean, I'll call it executive communications, investor relations are currently working. And if you want to brag on yourself a little bit and you have any, I mean, obviously not naming companies or anything like that, but anything you can share around proof points would be fantastic.
[30:39]
A
The first thing I would say is executives and their audit committees and their boards need to understand that they're no longer communicating to just humans. They're communicating to machines and algorithms. Whether it's from algorithmic trading, sell side analysts leveraging this technology, or you're seeing even more of the institutional investors and the buy side actually going directly to the earnings releases. And so knowing that there is a new audience and you are creating credibility leakage by not addressing it and not adjusting, you know, you're bringing, if you will, a knife to a gunfight from an analog to algorithmic component. So really the first thing I would say is understanding that the game has changed. The second is really around understanding risk and the components within that. So we've, whether it's communication, finance, other areas, we've already kind of made this commitment to de risking operational risk, financial risk, but no one's really taken a head on approach to mitigating communication risk. And it is measurable. And there is, not only is it measurable, there are adjustments that can be made that have a true ROI impact from it. So I think that is another big component that I would lean into in terms of something to resonate with individuals who are listening. Within that, from a metrics and performance standpoint, I would say, you know, we're seeing about 85% of our digital twins being able to accurately predict the question, or 85% of the question prediction being accurately predicted by our digital twins. We've seen roughly 30 to 40 basis points movement to the right from partners that are leveraging Adair on their earnings calls. And I think one of the other things I would say in terms of my last third takeaway is in a world of immense generation, whether it's utilizing different AI tools or other platforms, the value really becomes in terms of discernment and judgment. And so how do you layer this counsel into the technology? And so how do you move from efficiency when you think about what you're doing as an organization, into how do you improve the results? And at Adair, we're really focused on not just the workflow of how earnings calls happen, but how they are actually received and how the market will react.
[33:15]
B
I personally think people need to check you out. And with that, because this is just fascinating and one of the most interesting calls, I think, or one of the most interesting podcasts we've had. So with that, before we close it up, Kathleen, take a second. Tell everyone where they can find you if they want to contact you. You know, definitely tell them where they can go to find more about Adair, Whether that's on LinkedIn or your website, and how they can contact you if they're interested.
[33:37]
A
Yeah, I would say reach out to me on LinkedIn, KathleenPurley, and then visit Adair AI. I get to wear both hats as a operator and CEO. And a CEO, as well as an academic and researcher. So any questions would be welcomed. I'm working on some new research on language and how that is indicative of activist engagement and executive management cohesion. So if you have any interest in that, please feel free to reach out as well.
[34:07]
B
Awesome. Well, thank you, Kathleen. Really appreciate your time today. And so with that, you know, another episode of Reshaping Workflows in the book. I think this is one of our, one of our better ones. It's not a topic that we have tackled before with investor relations, but I think it kind of shines the light on that. You know, not necessarily AI, but, you know, the data science aspect is becoming. And the AI perspective is becoming more and more integrated into places and industries and workflows that might not have seen it. Right. I've seen it before. So with that, this is Logan with Reshaping Workflows with Delpro Precision and Nvidia RTX Pro GPUs. And we'll see you on the next one.
[34:53]
A
Do what you want. Do what you want. Do it.
[35:06]
B
This podcast was produced in partnership with Amaze Media Labs.

Summary5 min read

Podcast Summary: Fixing the Investor Trust Gap Using AI

Podcast: Reshaping Workflows with Dell Pro Precision and NVIDIA RTX Pro GPUs
Host: Logan Lawler
Guest: Kathleen Purley, Co-founder and CEO of Adair
Date: July 30, 2026

Episode Overview

This episode delves into how artificial intelligence is revolutionizing investor relations and executive communications, focusing on the trust and interpretation gap between what companies say and how markets understand their messages. Host Logan Lawler welcomes Kathleen Purley (Adair), who unpacks how their AI-powered platform addresses these challenges across the full lifecycle of investor communications.

Key Themes & Discussion Points

1. The Investor Trust Gap in Communications

The Issue: CEO and CFO statements on earnings calls are interpreted not just by humans, but by algorithms and AI models (03:00).
Insight: The inflection, tone, and even linguistic backgrounds of speakers can skew market perception in ways unrelated to actual company performance.
Notable Example: Non-native English speakers often have intonation patterns that machines pick up as uncertainty, unfairly affecting market confidence (03:00-04:10).

2. What Adair Does: Building the Missing Infrastructure

Adair’s Role: Acts as an intelligence and evaluation layer for executive communications—preparing, simulating, and analyzing company statements before and after they reach the public (05:22).
Unique Approach: Treats communication as a system that can be tested and optimized, similar to financial controls or supply chain risk assessments.

3. The C3 Communication Index

Metrics Explained: Clarity, Control, Confidence—tailored to role, sector, prior performance, and market expectations (05:30-06:55).
Quote:
- "[We] provide them a score around clarity...control...and confidence. These scores are optimized and are not generic...they need to be tuned to the financial fundamentals." — Kathleen (06:00)

4. Simulating the Audience: Digital Twins of Sell-Side Analysts

How it Works: AI-powered digital twins predict likely analyst questions, including low-probability but high-risk ones, enabling executives to better prepare for the real call (08:45-11:45).
Accuracy: About 85% success rate in predicting actual questions from analysts (12:24).
Insight: “About 30% of the questions our agents are producing are things that the internal IR team or the executive team hadn’t maybe considered…” — Kathleen (11:40)

5. The Workflow: End-to-End Integration & Real-Time Adaptation

Single Pane of Glass: Provides a centralized workspace integrating financial data, legal/compliance, communication edits, and all past Q&As (26:42-27:15).
Collaboration: Cuts down on manual versioning and ensures everyone from legal to comms is on the same page.
Flexibility: Clients often churn through dozens of scenario versions rapidly in response to late-breaking events (e.g., up to 37 versions in some cases) (28:00-28:50).
Post-call Review: Automated audit trails essential for compliance and continuous improvement.

6. The Value of Workflow, Not Just Models

Quote:
- “The value today is not necessarily in the underlying AI model itself, but in the workflow and architecture around it. It is a strategic role, not strictly an engineering role.” — Kathleen (16:40)
Bring Your Own AI: Adair flexibly integrates with clients’ models (open source or proprietary) while focusing on agent orchestration and ingestion pipelines (15:20).

7. Beyond Investor Relations: Expanding to All High-Stakes Communication

Applications: Earnings calls are the entry point, but Adair supports M&A announcements, crisis communications, investor days, executive presentations at major trade shows, and more (22:03).
Strategic Point:
- “Every high-stakes communication moment that has millions and billions of dollars on the line...will need to be simulated before it happens.” — Kathleen (22:10)

8. Shaping and Measuring the “Golden Record” of Communication

No One-Size-Fits-All: The goal is not to make all executives sound alike, but to help each find and retain the optimal version of their authentic voice within context (24:36-25:42).
Personalized Ranges: Recommendations are context- and persona-aware, accounting for industry, role, history, and macro environment (25:30).

Notable Quotes & Memorable Moments

On the shift in audience:
- “Executives...need to understand that they're no longer communicating to just humans. They're communicating to machines and algorithms.” — Kathleen (30:39)
On fairness and interpretation:
- “They are getting penalized for something that's completely out of their control that is no reflection on the ability for the company to execute.” — Kathleen (03:17)
On communication risk:
- “We've made this commitment to de-risk operational risk, financial risk, but no one's really taken a head-on approach to mitigating communication risk. And it is measurable.” — Kathleen (31:00)
On digital twins and prediction:
- “We've seen roughly 30 to 40 basis points movement to the right from partners that are leveraging Adair on their earnings calls.” — Kathleen (32:20)

Timestamps for Important Segments

Intro & Kathleen’s Background (00:52–02:11)
Analog workflows vs. algorithmic market (03:00–04:49)
How Adair fits pre-, during, and post-call (05:22–08:45)
Building digital twins of analysts & Q prediction (08:45–12:24)
Impact on volatility and credibility (12:24–14:18)
The Adair intelligence layer, model architecture (15:20–17:43)
Handling macro volatility and “the weird market” (18:48–21:12)
Extending Adair beyond IR: trade shows, M&A, crisis (22:03–23:57)
The search for communication’s golden record (24:36–26:06)
Platform workflow, auditability, and collaborative prep (26:42–29:42)
Three key takeaways & proof points (30:39–33:14)
Closing and where to find Adair & Kathleen (33:37–34:06)

Three Key Takeaways (as summarized by Kathleen)

(30:39–33:14)

The Audience Has Changed:
Executives must recognize that their communications are parsed both by humans and machines. Not adapting creates credibility leakage.
Communication Risk Is Real:
The “last undermanaged risk”—how you say it matters as much as what you say. AI tools like Adair make this risk measurable and improvable.
Proof in the Metrics:
Adair’s digital twins can predict 85% of analyst questions. Partners have seen 30–40 basis points of positive movement attributable to improved communication. The future is about judgment, discernment, and workflow optimization, not just AI models.

Where to Find More

Kathleen Purley: LinkedIn: Kathleen Purley
Adair: https://adair.ai
Kathleen invites outreach for research in language, activist engagement, and executive team cohesion (33:37).

This episode is a must-listen for anyone responsible for high-stakes corporate communications, investor relations, or executive leadership. It illuminates how AI-driven platforms like Adair are bringing the same rigor and testing to the art of company messaging as has long existed for financial performance and operational risk.

Loading summary

Transcript28 lines

[00:00]
A
Foreign. Welcome to Reshaping Workflows with Dell Pro Precision and Nvidia, where innovation meets real world impact in high performance computing.
[00:20]
B
Welcome back to another episode of Reshaping Workflows with Delpro Precision and Nvidia RTX Pro Pro GPUs. I'm your host, Logan Mahler. Today, very interesting topic, something we, we haven't really talked about before, which is investor relations. So with that I have Kathleen from Adair that's going to go all into how AI is shaping investor relations. So Kathleen, welcome to the show. Maybe take a few minutes, a little bit of background on yourself, what you do at Adair, and then we'll jump right into it.
[00:53]
A
That sounds great. So I'm Kathleen Perli. I'm the co founder and CEO of Adair. My background's actually in linguistics, which is what I studied in college. I then did a Fulbright in the impact on phonetics and phonology and second language acquisition. And while I wanted to be a cool PhD from the start, and my parents had other plans and I had to get a job, so I spent a good chunk of my career doing propensity modeling in the healthcare ad tech space and have really been driven to this idea of the fact that two individuals can say the same things and be misinterpreted. And so really diving back into that, we built Adair as this intelligence and evaluation layer for capital market communication. And the short of that is, you know, every quarter CEOs, CFOs and other executives get on these earnings calls to talk to investors. And the way that they say things, the hedging, the tone, the structure, it all carries signals that machines are now listening for. And so we really help ensure that executives are getting credit for the value that they're creating within their organization. In addition to all of my fun work at Adair, I also am a professor of generative AI at the Rice School of Business and also work with the deans on helping them understand how to navigate, ensure that we're preparing students for the future.
[02:11]
B
You gotta play a better background than me. So, okay, so let's start kind of with investor relations, right? So I work at Dell Technologies. You know, every quarter we have our earnings call where CEO, cfo, coo, go on and talk about how did the company do, there's questions asked, et cetera. I think everyone's kind of familiar with that. But let's. You said something really interesting, right? Is that the way things are said can carry like, you know, two people can kind of say the same thing. And have a different meaning. Before we kind of dive into Adair, the platform, what you do, do you have an example of that? And I mean you don't have to name names or you know, give away anything proprietary or any like that, but maybe set the context of like where you've seen an example like that maybe an investor relations or you know, maybe an ad tech drawing upon your background.
[03:00]
A
You know, one key, one that I see oftentimes is with non native English speakers. So one of the things that we found in our early research was non native English speakers. When you look at their tone, their vocal burst, intonation, you would find things that would indicate uncertainty or lack of confidence. The street, the institutional investors were oftentimes pricing that in as a lack of confidence in the business or the underlying business strategy. But when you actually hear those executives talk, talk and their native tongue or their, their kind of first language, that actually the intonation pattern actually disappears. And so there it is in the margins, if you will. But they are getting penalized for something that's completely in essence out of their control. That is no reflection on the ability for the company to execute. And so what's interesting is, you know, we've built a ton of infrastructure around financial reporting for decades, but we haven't really built a ton of infrastructure on how that information is actually communicated out and, and interpreted. And what we've seen is, you know, we have internal IR teams still leveraging tools that are from the analog era, if you will. So they're still leveraging spreadsheets, word docs with track changes and comments. But the markets have moved heavily algorithmically. So we're seeing about 70 to 80% of all trades being algorithmic. About a third of all sell side analysts are leveraging AI as that first pass. And you're seeing activist engagement and firms as well as individuals that are looking at compliance constraints using AI as a way to help target and identify kind of that first list to take a deeper dive on. And so I think one of the big challenges is we're not all being scored equally first. And the second is our audience has also expanded.
[04:50]
B
Yeah, that makes total sense. Right, like, but help me understand like how does Adair kind of fit into that? Right, because my nascent knowledge of kind of investor relationship, obviously I'm not an executive, but you know, that call occurs, it's kind of happening in real time. Right. Does a dare kind of fit into the pre planning like the practice and the prep that leads up to it? Is it more kind of a middle layer in between translating as things are kind of happening live, or is it more kind of the post type reporting or is it kind of all three?
[05:23]
A
It's all three. We kind of sit as a layer across kind of the beginning stages of the prep, all the way through post analysis and insights for roadshows, if you will. So post call conversations and meetings and conventions that happen. And what we do is we treat communication like a system you can test, you can simulate, and you can optimize before it hits the market. And we do this in two ways. Given my background in linguistics and really diving into my own academic research as well as other peer reviewed academic research, we've developed a C3 communication index which really takes each speaker and provide them a score around clarity, which really looks at can the market process IT control, Are you owning that narrative? And the third one is around confidence? And, and that's really structured around are you signaling the right conviction? Now these three scores are optimized and are not generic. So obviously what your, your CFO will always have to be more balanced than your CEO. What you say in tech would be considered evasive in energy, for example. And then it also needs to be tuned to the financial fundamentals, because when you think about confidence and you think about the algorithms and how they're parsing language, it really needs to take into context what are the commitments that have been made in the past, how does this trend quarter over quarter, and what do the financial fundamentals look like? You know, we are seeing, you know, if you are reducing modals, for example, you are getting a little bit more of a kind of movement to the right on stock performance. But it needs to be taken into context of how you perform, because the worst thing that you can do is have weaker financial fundamentals and try to sound really confident because you'll ultimately just end up sounding tone deaf. And so that's one big component. On the prepared side. We also look at, you know, how this will land with the sell side analysts. So we actually will simulate through digital twins of each of the sell side analysts how they'll likely respond and what questions will happen before the call even occurs. And so this allows the executives and the teams to really think through the different potential pushback points. What do they might want to adjust in their prepared remarks versus really just have better understanding of how that needs to be communicated in the Q and A. And we look beyond just what questions have been asked in the past, what's getting asked on their peers, but really looking at how do they view macroeconomic components what is their consistency in terms of re rates and what usually triggers a pricing target change and how do they want to be communicated to. And so this allows the executives to have a little bit more confidence going into these earnings calls calls. And I think it's very similar to what you see in any business. We spend a lot of time trying to de risk every aspect. So whether it's the supply chain, whether it's financing, but the earnings call communication is that last kind of what I'll say, gate, if you will, that hasn't been de risked yet. And that's what we're really helping our partners do, is helping them de risk that and ensure that. And how we think about it is the company's execution creates the value, their communication within the earnings call and afterwards really dictates what they get credit for as it pertains to that. And so really trying to remove the gap between what's communicated out and what's understood is a huge part of what we do.
[08:45]
B
Makes total sense. I mean, the first kind of the example you shared is making sure, kind of the message from whoever's speaking matches the facts, right? Like, you know, for example, I like to think I'm a pretty confident guy and sometimes I'm flat out wrong, but I just communicated with a bunch of vigor and people usually give in. But, like, if someone actually knows the answer, which, I mean, obviously an analyst, you know, at some sort of investment company would obviously know that and would take that as like, you know, dodging or et cetera. But the point you made, I think is really, really interesting and I definitely want to dive in, is being able to create kind of, I mean, digital twin means a little bit different, right? It's usually a 3D representation, but creating almost a Persona of what analysts will be kind of on the call to be able to have whoever speaking during the investor relations call be able to kind of see how it lands, right? Like, I would love to hear a little bit more. I mean, don't give away anything proprietary. But, like, how do you go about doing that? Like, how do you create something that can mimic, you know, their thoughts, feelings, their views on, you know, the market or the specific industry that they're in or, you know, even how they would perceive something. Like, how do you go about doing that? That's so interesting.
[09:59]
A
I think there's a lot of that goes into, you know, as we think about it from a psychology perspective, behavioral economics as well, and really parsing. I think one thing that is oftentimes missed is you learn A lot about an individual through the way that they communicate and how they structure even something as simple as like sentences in terms of how they want that information to be fed back to them. Because oftentimes they're giving you that playbook and sometimes it's very explicit. So you look at the sell side reports or you see their latest commentary that they were spotlighted on from cnbc. And so we're aggregating a significant amount of data and then layering our own understanding of linguistics, behavioral psychology, behavioral economics on top of that to understand and build these Personas, as you mentioned. And what we do then from that standpoint is we're updating these in real time. So these are not static, but they're really understanding, you know, what questions are they asking on your peers that happened within 12 to 24 hours before your call. How does their tone within that question deviate? One of the things we look at is divergence. So if the written sentiment of a statement or question from a sell side analyst differs from the vocal, we identify that as a flag of something that might be unsaid. And so what this really does is help prepare them. We have about an 85% accuracy for question prediction for our partners. And what I think is more valuable and more interesting is that these Personas and these digital twins, if you will, we actually force them to produce questions that not only look at what is likely to get asked, but what are the low probability but high risk questions that might get asked. And what we're hearing is that about 30% of the questions our agents are producing are things that the internal IR team or the executive team hadn't maybe considered that they're really glad the agents are bringing to light, even if they never get asked on the public facing call.
[11:59]
B
That's so interesting. So I, I did not realize that's how the process worked. I mean, once again, not an executive, but I kind of figured that questions would have to be submitted in advance and, you know, approved. But like an analyst can really ask whatever they want. And you said 85% of the time you can kind of source that topic or something that they might ask that might have been not planned for. Basically.
[12:24]
A
Yes, absolutely. And I think what's interesting is, right, markets are creating arbitrage off of the nuance within your communication. They're looking to identify, like, what edge do I have over the next sell side analyst or the next institutional or the buy side? And so these small clarity gaps in executive communication widen that modeling dispersion. It can potentially erode some credibility and amplify some of the volatility. And right now, there's just been, to date, no infrastructure to evaluate that risk before it even happens. And so what we've really done, and you've seen a lot of this technology be used on the other side, so not the ones that are issuing. So the company side, which is what we focus on, you know, the ones issuing the statements are not really going into this depth, but you're seeing it more on the buy side or the sell side with their quants. And so really understanding that you are talking to two different audiences now is a huge component of knowing the fact that, you know, we often will see companies beat expectations and still underperform. And the gap is rarely tied to the financial fundamentals. It's often tied to the interpretation. And so we'll see sometimes the verb, like, if they're too verbose and you think about auditory processing and the unconscious friction that you're creating and sentiment that you might be creating with listeners to, you know, if you have a really strong fundamental, but you have a significant amount of hedging still involved, that might indicate, you know, there's another shoe waiting to drop so what's not being said. And so we really find that the earnings call is something that happens every single quarter. It's a repeating component, so we can actually test for it. And it's probably one of the highest stakes reoccurring moment in business that there hasn't been an infrastructure in place to really help support executives land their communication as intended.
[14:19]
B
Makes a lot of sense. Like, I mean, you're right, it is kind of quarterly, traditionally, and repeatable. And it's. It's something that's done. I'm curious. You know, I. I find it completely fascinating, you know, maybe more a little bit. I mean, that's. You don't have to go super technical on this. But, for example, you know, I'm assuming that there is more than you just pulling down the latest model from hugging face to do this right, there's got to be some sort of, you know, proprietary training models that you use to understand how the market will respond to. You know, for example, one of the examples you used was, you know, results were really good, but they were too long. I mean, they just spoke way too much and were way too boastful. And then the, you know, the. The street slash, you know, the market in general interpreted that as, you know, something negative. Right. Like, how did you go about kind of creating, let's say, the intelligence layer to be able to understand and kind of predict what you do in the Adair platform.
[15:21]
A
I think, you know, it's interesting because everyone talks about, you know, is this idea around, you know, building proprietary models and components of that nature. But I think what people often miss is that the value isn't in the engine itself. We given the security of what we do, we run on premise for our clients due to the data sensitivity. And so we've actually built and fine tuned open source models. That's actually where we started. But what we find is that oftentimes IT and security individuals within an organization have a lot of constraints around that and there's a lot of testing. And so the speed to value is a little bit longer. And so what we've done is we've really created a model garden, right? So it's more of a bring your own AI, if you will. And the value that we layer in is around our agent orchestration, our ingestion and our pipeline components of it we deploy through Dockers and Kubernetes. And so it can allow, and I won't say, because my engineering team will kill me when I say this, but it's not plug and play in the sense of, you know, if a company has a specific model they want to leverage, we are able to adjust, integrate that, do evals to ensure performance. Because the value of what we bring is more on the ingestion layer, some of the algorithms and some more components within that, rather than just strictly the underlying model itself. We see that being more of like, if you will, you know, a thermometer can tell you what the temperature is, but can it diagnose what's wrong within a patient or what we should do about it? And so what we kind of do is we actually utilize quite a few, a significant number of agents that are digesting certain components of information. Our agents will then create certain artifacts that then are used by other agents to then produce certain components of the different aspects, whether it's the prepared remarks themselves, understanding of the analysts and where they are, the quarterly backdrop, what questions are going to get asked, how to answer those questions, tone and pacing, things of that nature. And so I think when you think about the value, and it's one thing I push my students for, is the value today is not necessarily in the underlying AI model itself, but in the workflow and architecture around it. And I believe, and this might not be well liked by a lot of engineers out there, but it is a strategic role, not a strictly an engineering role.
[17:44]
B
You know, here's kind of a Question for you. So I get there's kind of, I mean kind of an agent based system in a sense that's looking at different components and, and that how do you handle something like. And this is Logan Lawler's point of view, this is not Dell's point of view. But you know, you look at the S&P 500 right now, I mean, I don't think it's at an all time high, but I mean I think we're pretty close to it at the time we recorded this. But the market feels kind of choppy. It feels a little, you know, weird that we've got a lot going on in the world which we won't dive into, but the market's at an all time high. Like how do you go about kind of predicting those? I mean, I'm not going to say unknowns, but it feels very weird, right? I mean this isn't the rise that we had in the mid 2000s where everything was just kind of up, up and up like, or not 2000s, but like the mid, like say 2014 after recovering from the financial crisis. You know, everything was kind of up, up and up and it just kind of followed that trend. I mean we're still going up, up and up, but it doesn't feel right, I guess. How do you even predict that or how would you even layer in that level of intelligence into a dare?
[18:48]
A
So we do look at things like the vix, we look at our, the peers in a certain segment, the S&P 500, their closest ETF for normalization and things of that nature. And I think one of the things that is really interesting in terms of we pipe in real time data as well. And so I think, you know, while there is, you know, we have a very strong market right now, there is a lot of uncertainty and quite a bit of volatility. And what we find and what we focus on is the goal of the Adair platform isn't to, you know, if you will, it's not about sounding better, it's about being understood correctly. And so when we think about it from that perspective, you know, we're not a magician. If you have a bad financial fundamental quarter, like we can't make it a great one as it pertains to how the stock moves, institutional investors, kind of what the sell side analysts will do in terms of RE rates and price target changes. But what we do is we really focus on the margins. Right. And so we're able to attribute roughly for most of our partners, what we've seen is about a 30 to 43 basis points movement to the right. And so it's really kind of like what you see in terms of like a volume adjustment. And so when you layer in market volatility and uncertainty, you need to control for everything that you possibly can. And so what we do is really focus on kind of adjusting that volume of that noise. So if it's a bad financial fundamental quarter or material miss, we turn that volume down. If it's a great quarter, we help you turn that volume up so that you get the most like you create the value. As I talked about earlier in your operations and your execution, the communication dictates what you get credit for. And that's what we're really focused on, is those margin components. You can think of it very similar to performance medicine for elite athletes. Elite athletes don't make it to the Olympics because they're bad at what they do. They lose in the margins. It's a shave of a hair here, a shave of there, a slight increase to speed or start that then changes their ability to podium or not. And that's what we're really focused on, is really controlling for those margins. And those margins at Fortune 500 companies are significant when you think about a 30 basis points movement and the ability to kind of control for that.
[21:13]
B
Here's kind of another question. This is, I'm kind of thinking about Adair and like how it operates, right? I mean, you focus kind of on investor relations, which is obviously very important, right. But like I could kind of see this going beyond, you know, just investor relations, right. Like for example, you know, I've been at a lot of conferences, you know, like a GTC and stuff. And based upon, you know, Jensen's keynote, for example, that might move the stock price depending on what he says or doesn't say, et cetera. And that's kind of the same for any big kind of event where a CEO is front and center speaking at maybe their own trade show or you know, an industry trade show outside of a typical investor relationship. Is Adair something that you can use for that or is that something you've thought about or is it focus only on that, like once a quarter, quarter closed investor relations call?
[22:04]
A
No, we really think about it in terms of every high stakes communication moment that has millions and billions of dollars on the line that will impact the capital markets. And ultimately each one of those high stakes communication moments will need to be simulated before it happens. And we've, we do support, obviously earnings calls is kind of what we find as our initial Wedge into talking to companies and CEOs and CFOs and getting them comfortable with the technology and understanding how important not only what you say, but how you say it. And we've been leveraged across investor days, M and A announcements, crisis comms. We've done a significant amount of work also within and are expanding our work within activist engagement. So we're seeing, I think 2025 had an all time high of activist engagement for companies and so really helping understand how executive communication, like you said, whether it's an earnings call, an announcement, developer day, how that will move the market and understanding and testing it before it even becomes a reality, I think is a huge point of where we need to go as a, as a linguist in a world of finance bros, if you will. I've always had a passion for turning language into math and this is just starting to validate. You know, we've always had a golden record for or source of truth for numbers. But the level of communication and whether or not we measure it and how we think about it and how we de risk it has never had that visibility or stage presence, if you will, in terms of importance. But what we're doing with Adair is really ensuring that not only do we have the ability to make the invisible dispersion visible in terms of what's spoken versus what's understood, but how companies can kind of help de risk that communication even further.
[23:58]
B
So you mentioned kind of golden record, right? Is there such a thing, you know, in a CEO or a CFO communication in the way is there like the perfect way to communicate something or is there, is there a range? Right. Like for example, I've, I've listened to a few investor relations calls. Some people are extremely passionate. Some people are just very, you know, monotone, give the facts and move on. Like golden record, obviously with numbers, I mean the numbers are the numbers and that's it. But the interpretation of the numbers. Is there a kind of a golden record in how to position and position the numbers in the context of like a business performance? Yes.
[24:37]
A
And that's what we're doing through our blueprint narrative spine and our C3. Because one of the things that we want to ensure is that the CFO or CEO, who's extremely passionate, when they start leveraging a data, the worst thing that could happen is this whiplash where all of a sudden they don't sound like themselves. So our goal isn't to ensure that everyone sounds the same. That would be actually counterproductive to what we're trying to help executives achieve. But what we're doing is really focusing on what does their voice sound like, what does their communication typically sound like, what is the best version of that, given the context of financial fundamentals, industry role and the macro favorability, if you will. And we provide ranges. So within our C3 communication index scores, each speaker gives a targeted range that is taking into account all of those components and where they've been historically and where their peers are. Because I think one thing that we fail to lose sight of is narrative is always taken in context of something larger. And when you look at standard generative platforms, you lose sight of the context or the nuance that it's required. And that's what Adair is really focused on, is not just generation, but this council layer that pushes it into a way that allows to create a repeatable event where executives can actually start to experiment and test and measure the impacts.
[26:07]
B
So a day or let's say, you know, you have a new client, doesn't matter who it is, they have an investor call and you know, they deploy a day or they're set up, is it really kind of a, kind of a two part question, Is it kind of a single pane of glass where all of that kind of prep occurs versus more the, the analog, like you mentioned, kind of spreadsheets and you know, shared files with track changes and stuff like that. And then how many times typically would you know, you more or less practice does? I mean how many practice runs does it take to get it right, I guess or get it to where you want it to be.
[26:43]
A
So I would say this, there is like this centralized hub for it. So it'll show you business unit data, It'll show you C3 scores across your past and your peers prior commitments that have been made so you can understand as you're formulating the next earnings call. What do we say last quarter? What is the street going to remember and hold our feet to? So there is that single kind of collection of this like operating system we tie into the finance team. So when they update a number, it's actually tick and tied into both the remarks and the Q and A and suggested responses. And we also maintain, so legal gets in there, communications gets in there, compliance so that they all of the commons, track changes, edits, revisions, everything can be seen. Adair actually often operates after the first pass of the prepared remarks as another editor that you can kind of select, copy and ask for input on. And what we find is that twofold one, this allows for a lot of the removal of back and forth, but really pulling into context the full, bigger picture of everything that goes into it and understanding how that operates. So one of the components is that Adair provides is like an auditable report post call where you can see every single change that has happened, which is really important given some of the regulatory constraints on this type of communication. Now the other component of this, which I think is something I'm really excited about and more excited than I probably should be, but this just tells you how much of a nerd I am is really looking at when the. When you think about an earnings call process today, not only is the materials oftentimes scattered, if you will, you're oftentimes not getting final numbers until 72 hours before you know whether it's looking for tax information. EPS is typically one of the last things that comes in. And so I always tell individuals it's this living, breathing exercise. So when you talk about versions, I think sometimes you'll go through. We've had clients go through 37 versions of prepared remarks before the final and it might be a shift in geopolitical components or they'll do a version. And what's great about Adair is the speed at which it helps provide these simulations and different test scenarios allow them to do okay, well if this legislation passes, that's going to change this. So let's do one. And we'll find out three days before the call. Right. So it allows for that kind of variability, but the ability to adjust fairly quickly as well. And so we do see a significant amount of that. We are also providing kind of in your post call roadshow or analysis, whether it's rate of speech, how to frame some of the communication. I joke. I love that book in college, eat this. Not that, but kind of from a, from a communication instead of this, say this. And we're also starting to integrate live real time Q and A support as well.
[29:43]
B
It's so fascinating. And you're right, like I'm a nerd too a little bit. I mean obviously maybe not as smart as you, but like I do find it quite interesting. I didn't before this fall know like what went into an earnings call when it makes total sense that there's like 37 or whatever you said, like numbers of reviews because things are changing up, you know, kind of by the minute. So you know, so far it's been a great conversation. We're getting kind of towards the end of it, like what I like to do is, you know, kind of to wrap up the call, give kind of the audience and pretend that they just kind of tuned in right now. What are kind of the big three takeaways that you would say about Adair and how it is kind of reshaping how any, I mean, I'll call it executive communications, investor relations are currently working. And if you want to brag on yourself a little bit and you have any, I mean, obviously not naming companies or anything like that, but anything you can share around proof points would be fantastic.
[30:39]
A
The first thing I would say is executives and their audit committees and their boards need to understand that they're no longer communicating to just humans. They're communicating to machines and algorithms. Whether it's from algorithmic trading, sell side analysts leveraging this technology, or you're seeing even more of the institutional investors and the buy side actually going directly to the earnings releases. And so knowing that there is a new audience and you are creating credibility leakage by not addressing it and not adjusting, you know, you're bringing, if you will, a knife to a gunfight from an analog to algorithmic component. So really the first thing I would say is understanding that the game has changed. The second is really around understanding risk and the components within that. So we've, whether it's communication, finance, other areas, we've already kind of made this commitment to de risking operational risk, financial risk, but no one's really taken a head on approach to mitigating communication risk. And it is measurable. And there is, not only is it measurable, there are adjustments that can be made that have a true ROI impact from it. So I think that is another big component that I would lean into in terms of something to resonate with individuals who are listening. Within that, from a metrics and performance standpoint, I would say, you know, we're seeing about 85% of our digital twins being able to accurately predict the question, or 85% of the question prediction being accurately predicted by our digital twins. We've seen roughly 30 to 40 basis points movement to the right from partners that are leveraging Adair on their earnings calls. And I think one of the other things I would say in terms of my last third takeaway is in a world of immense generation, whether it's utilizing different AI tools or other platforms, the value really becomes in terms of discernment and judgment. And so how do you layer this counsel into the technology? And so how do you move from efficiency when you think about what you're doing as an organization, into how do you improve the results? And at Adair, we're really focused on not just the workflow of how earnings calls happen, but how they are actually received and how the market will react.
[33:15]
B
I personally think people need to check you out. And with that, because this is just fascinating and one of the most interesting calls, I think, or one of the most interesting podcasts we've had. So with that, before we close it up, Kathleen, take a second. Tell everyone where they can find you if they want to contact you. You know, definitely tell them where they can go to find more about Adair, Whether that's on LinkedIn or your website, and how they can contact you if they're interested.
[33:37]
A
Yeah, I would say reach out to me on LinkedIn, KathleenPurley, and then visit Adair AI. I get to wear both hats as a operator and CEO. And a CEO, as well as an academic and researcher. So any questions would be welcomed. I'm working on some new research on language and how that is indicative of activist engagement and executive management cohesion. So if you have any interest in that, please feel free to reach out as well.
[34:07]
B
Awesome. Well, thank you, Kathleen. Really appreciate your time today. And so with that, you know, another episode of Reshaping Workflows in the book. I think this is one of our, one of our better ones. It's not a topic that we have tackled before with investor relations, but I think it kind of shines the light on that. You know, not necessarily AI, but, you know, the data science aspect is becoming. And the AI perspective is becoming more and more integrated into places and industries and workflows that might not have seen it. Right. I've seen it before. So with that, this is Logan with Reshaping Workflows with Delpro Precision and Nvidia RTX Pro GPUs. And we'll see you on the next one.
[34:53]
A
Do what you want. Do what you want. Do it.
[35:06]
B
This podcast was produced in partnership with Amaze Media Labs.