I had a conversation just this week with somebody who said that ⁓ at some level, it feels like the folks at Microsoft know that and they are acknowledging it and trying to collapse their ecosystem right now of copilot because every copilot product is a different product. What's inside of Excel versus what's inside of Outlook versus what's inside of Word versus what's on their website and what's on SharePoint. It's all different. I feel like that opening line for every one of them is

So I read an article. I I read lots of articles. Man, that's bad. But yes, I mean, like that's how I read articles now. I mean, it feels like the entire internet is Buzzfeed, but Hey, you know, I, I get, you know, there's some wall street journal articles that I'll dig into, but no. So I skimmed an article. ⁓

You scrolled an article?

And I didn't go behind the paywall completely, but I did find it in a few different places. But there was this medium article out there that said, we really need to kind of fact check it, but 84 % of people on the earth still haven't used large language models.

Yeah, that's easy math. there's like, how many people, Zulu tribes in the middle of Africa are... My parents are from Bangladesh. That's 150 million people. How many of them have used AI?

Well, I mean, don't they have the opportunity? mean, with smartphones, how many people have smartphones?

Yeah, like one like like the one thing that Bangladesh has to their credit is this one Nobel Prize from the guy who invented the whole microfinance thing. So there's one woman who has a phone in a village who, you know, that creates a micro, you know, economy for that village. So, yeah, maybe there's one lady in a village of 400 people with a smartphone. I don't I don't think she's on chat GPT.

There's not smartphones all over the place. Yeah. Well, I'm feeling a little bit of shame. Like I feel like you, I feel like you totally took me to school on my, uh, I feel entitled, I guess is what I'm saying. I feel like I came across as very entitled just now and uh, I'm a jerk. I apologize. Yeah. Well, I don't even know the right thing and the wrong thing to say a lot of times. So, uh, but you're, you're probably right. mean, like there's, mean, there's just a lot of people out there that don't have access to it. even if the numbers off by half, that's still a ton of people.

Well, right. And I think the point, like the sensational side of the article versus kind of pivoting into all the people who should have access to AI, what is the percentage of those people who have not touched it or have not adopted it in any way? So today, we're talking a little bit about fluency, the idea of are people even thinking about this in the terms of I have to start building a sense of fluency with these tools. ⁓ The same level that if I hand somebody a phone, they have enough fluency to know there's an unlock and there are apps and it's a touch screen. They're not going to go all caveman on it and ⁓ not understand what the thing is.

Yeah, I actually had a conversation with a CMO friend of mine and trying to get their team to increase their use of AI just generally. It's not going well. Then, you you start talking to them and there are a lot of tools that they aren't aware of either, you know, and I mentioned some of the ones that we talk about on here, like Manus and which I think Manus is pretty well positioned for mid-market if they can get it, if they can get their head around security because it's a true end-to-end kind of solution. I think it's a bit more expensive from what I can tell than some of the others, but it's definitely turnkey and it makes you smarter. But I think everybody's struggling with this because I keep going back to intellectual curiosity. I feel like that is the... That is the differentiator right now. I had a friend, sorry, I had a friend tell me, you know what? I like to think of it as intellectual grit because change is so rapid. I don't know if I fully, I think I like mine better, but I'm biased.

I think the curiosity comes first, right? Because the grit doesn't matter if you're not curious, right? If you're curious, you're going to get in there. The grit is sticking it out, right? I mean, you and I have been, you know, we started this episode, you know, joking a little bit about like maybe an entitled worldview. But we are entitled to the point that you and I can spend time every day, every morning.

Every company should be digging into this, I feel like, a lot harder than they are. The tools, as we've talked about before, the tools are available right now that do really, really cool things. And there are still really, really large companies that don't have the fundamental infrastructure in place to build on top of it with these tools. That's what's baffling to me.

Well, mean, that's big piece of it is do they have the infrastructure to, or do they have the curiosity? And do they have somebody who sitting in the seat ⁓ who has been given the hall pass, right? To kind of move it forward. And so, I was talking to a colleague of mine earlier this morning, we were talking about how you cannot do these types of things ⁓ in a health care setting sometimes just because of the risk attached to it. ⁓ even though there are hundreds of really great use cases for application of AI tools, it's really hard to get the actuarial analysis to work out or the risk analysis to work out. ⁓ at an event last week, last week, two weeks ago, and it was a group of CIOs and there was a panel discussion. There was a gentleman there who was with a national, international automotive manufacturer and he's an innovation officer there. And he was representing kind of their, that seat of innovation and what they're trying to do. And he set up a story. It's a great story. It was fun for me to hear him tell the story because this was a client of mine 10 years ago and the story actually goes back 10 years. And so the story he told was one of the fact that this automotive manufacturer has not done a great job with customer data from a structural... ⁓ lineage and definition of what a customer is and what

What's the approximate market cap?

This is billions, nah, but it's billions of dollars. They are an automotive manufacturer, right? Like you drive past the dealerships that they sell these cars. And so they're a big company. They're not just a startup. They're one that's been around for 50 years. But what the use case that the ⁓ fella brought up was the fact that they had ⁓ duplicates in their customer data. Difficult to wrap your head around is sometimes you have to really look at customer data and when you get to be this type of like, you know, fortune 100 type company, the definition of what a customer is, is hard because for this, car manufacturer, there's the customer who bought the car, the buyer. There's also the customer who financed the car, right? Because if your dad bought you the car, I mean, maybe the car's in your name. He's a cosigner. He did the financing.

Cosiners. My first job out of college was collecting on car notes. So I'm well aware of the tangled web.

Somebody is the driver, which is an interesting component now because ⁓ the radio, you and I grew up with Pioneer and Alpine stereos, and it's not a stereo anymore. We call it the head unit in the car.

Yeah, you've got to take that out before you leave because it'll get stolen. And I used to carry my CDs with me as well. Yeah, because the CDs would get stolen. The head unit would get stolen and the CDs would get stolen if you were in the wrong place.

I'm just saying that in my third generation 4Runner that I drive, one of the things I really wanted when I bought that was to put a nice stereo in it. I've got a Pioneer stereo, faceplate comes off. It's so good. I don't want somebody to steal my stereo. But that head unit and stereo, the driver is important because I'm selling

it off somebody's gonna steal it

to the driver products and services as you drive. And so that is a customer of the company as well. And then we could even go way down the rabbit hole over like broker relationships with leased cars and stuff. So who are all these people that are in their database? How many people are represented across multiple slices of their database because they are both a driver and an owner or a financer? This is not a unique problem. And this problem actually has a solution that ⁓ is pretty well recognized in the industry, which is ⁓ who knows and maintains and has the most incentive to maintain customer data? It's financial institutions. And so the trans unions and the Equifaxes of the world all have services that de-duplicate data. ⁓ And they're great at it. And they know they're great at it. And it's a premium service.

Yeah, it's expensive and yeah, they are very good at it and that actually helps collecting on car notes, believe it or not. You can find them. You can find them in a lot of other people that know them.

over a decade ago was decided that was too expensive and they stopped ⁓ using that service. ⁓ Yeah, yeah, yeah, it was great. ⁓ And so ⁓ I knew about this ⁓ years ago as a ⁓ consultant ⁓ because we did some research into this and over the course of several years, they had a lot of drift in their data and ⁓ duplicates. At some level, people tried to solve it in... different verticals within their company. They tried to ignore data that was associated with finance, or they tried to ignore data that was associated with the dealership. ⁓ They didn't solve the core problem. And ⁓ what this innovation officer brought up was the fact that ⁓ he had to sit next to their chairman, ⁓ CEO, in a conference room with his laptop and put his name into the database and then show him that there's five of him.

Yeah, that's got to be an uncomfortable conversation. I would love to understand how they even got to that point. Unfortunately, coming from the world of finance, it sounds like a finance driven decision at some point.

Now we're at a place and kind of the where this project ends though is that companies and the services like ⁓ the Equifax's and the TransUnions of the world are still valuable services. But rather than diving into a new contract with them, we are past the point of fuzzy matches with the databases and with AI we can really start looking not just at the data that's in the database but how those entities have interacted with the customer or with the client ⁓ over time. And we can let loose an army of AI agents to go dedupe or reclassify this data. ⁓ And it's not the same type of investment, right? We're not getting our clients or our companies into multi-year locked in investments with external partners. We could start building and solving and maybe we get ourselves to the 80 % mark and then have to make a decision about the, you know, is the last 20 % worth the squeeze.

You you mentioned something earlier about, you know, letting people go. That's why I feel like it's a little different and maybe I'm wrong. I've argued both ways that AI and this whole, you know, revolution is different than the industrial revolution. It's different than web.o and you know, like, but it feels like there is an trying to say this the right way. It feels like there's an opportunity for small and mid-sized businesses to run a lot faster than big companies that don't have their, that are so worried about risk, if that makes sense. they're worried about, you know, personal information. So PII, all of that stuff. What are we going to lose? Back to your hospital example. However, they're gonna, the opportunity cost feels so big for not letting everybody at your company kind of be entrepreneurs and really start to build things that solve real problems that they've been facing all the time.

Well, and I mean, you've sat in the seat where you've had to justify the investment for, you know, letting people sit and be entrepreneurs and giving them the tool sets that they need or dealing with the case of people running outside because the tool sets are free. And yeah.

I don't know if I did the best job laying out the story. Like how do you lay out, know, cause now, mean, just this week, you know, you're seeing more and more articles about all of these businesses, you know, how do we budget for AI? How do we now, it is a line item on a P and L at this point, because the usage is going through the roof.

So I mean, how do you justify, ⁓ what do you think, I mean, as we're talking to our peers here and we're talking about ⁓ budgeting for tools that not only have a per member per month type cost like the way that we talk about it in healthcare at times, but also we have a token burn credit utilization in the same way that ⁓ like our AWS and Azure services have started. charging, you know, I have always charged based on transactions or, you know, cycles, processor cycles. ⁓

what he and I were talking about was, you you've got all these big, big products out there and I won't name names, but I've spent a lot of money in my time in multi-unit environments on facilities management software. So, you know, you put in a work order or somebody needs to come out, fix the HVAC, things like that. There is so much, so all these companies are now trying to retrofit their products with AI. I think that it's hard because there's so much feature bloat over time. Like if you buy one of those solutions, I can't tell you the number of times I've approved a solution and you probably as an IT professional has approved a solution. And then the company uses like 5 % of the solution.

It's part of my regular schtick when I'm talking to these private equity companies that have bought a lot of solutions in the name of addressing problems. Nobody got fired for buying IBM, and so then they go out and buy CrowdStrike plus some other security tool, and they're using 5 % of each, and there's overlap between the tools. Yeah, it's a big part.

Yeah. And so, right. And that's because three people asked for a feature early on and the CEO said, fast track it. That's our biggest client. Make it happen. You know, so there's probably like 12 users on some of these big softwares that use every little bit of it to run their company. You know, they went all in, but most of the other companies just want a work order system. That's not hard to build in a weekend right now. So if you don't need a you know, a Maserati or whatever you pick the car, right? All you really need is a bicycle and you can build that in your garage. It's like, let's just fix this one little problem. So that's, that's kind of what he and I were, we're talking about. And I think these smaller companies are better set up to do that because they're just trying to solve these really discrete problems. And they're not having to untangle a lot of a big data mess.

Yeah, I think what I've noticed and I've talked to other ⁓ CIOs about is 25 years ago, my job at this large company, ⁓ international company, they were huge and I worked for one division of theirs and ⁓ what we were doing was going around auditing whether anybody was using a Microsoft Access database. because that was the face of shadow IT 25 years ago because it was accessible. You could go to Barnes and Noble or maybe Walden books back then and go get a book and the dummy's guide to Microsoft access. yeah, a lot of people did and they built systems. They were fragile. they were uncontrolled, right? Which was a big deal for this company that I worked for. They were ISO regulated. And so you had uncontrolled data out there. But today it's a similar problem, but it's a lot easier to build and you can secure it in some ways better. ⁓ You can democratize access to it a lot better. ⁓ And I think ⁓ we're moving to a place where it's not just this one really smart person sitting in a cube in the corner of the office who's the only person who's smart enough to use the tool, right? I mean, everybody had access to Microsoft Access, just only 0.01 % people knew what to do with it.

Yeah, or took the time is still, it all goes back to being curious. If they were curious about the solution and they had a problem that they needed to solve, then they would spend their nights and weekends trying to figure that out. And then, yeah, they do become somewhat indispensable at that point. It's, you know, it's much harder to pull that out, that person out because they're the only one that knows how it works, or at least that's the fear. I think that's usually an overblown fear in a lot of executives minds. think a lot of things can be switched out a lot easier, but that's a topic for another day.

But you know, mean, for those people who are ⁓ doing that work, the access to the tool, which is what we've talked about, these tools and access to them and giving them the ability to run fast. And ⁓ now AI is getting baked into every product. So earlier we were talking about ⁓ one of my favorite topics that I do my hot take on all the time, Microsoft Copilot, just garbage. So ⁓ now, I... I had a conversation just this week with somebody who said that at some level, it feels like the folks at Microsoft know that, and they are acknowledging it and trying to ⁓ collapse their ecosystem right now ⁓ of Copilot. Because every Copilot product is technically a different product, what's inside of Excel versus what's inside of. ⁓ Outlook versus what's inside of Word versus what's on their website and what's on SharePoint. It's all different and ⁓ different capabilities, ⁓ different guardrails. ⁓ And I think ⁓ maybe it will get better. while that it might, using these tools, you had an experience just in the last couple of weeks as we looked at how I think OpenAI and Claude both have got like a plugin with Excel.

Yeah, and honestly, my thoughts on some of this have changed a bit. ⁓ Well, and at least how I use it has changed a bit. So my first experience and I posted about this on LinkedIn was, you know, I was like, hey, I'm going to use this Claude Excel PowerPoint plugin, see what it does. I asked it maybe three questions and my tokens were gone because I'm just playing with Claude at this time. I'm a little bit further into it now, and this has only been a couple of weeks, but I started playing with it and three questions in, three prompts in, I'm out of tokens and I was like, this is ridiculous. I didn't even ask it to do much. Since then they've upgraded and given double usage to everybody for a while. So that's been nice, but I'm still spending like, if I spend a lot of, let's say an hour, hour and a half on a financial model for a client, or my own businesses, I might spend $10.

a finance professional, and when you say you're working on a model, like, I mean, we're talking about multiple worksheets, like a complex entity. You're not just taking ⁓ rows and columns in a spreadsheet and trying to like pivot table and build,

No, I mean, like you've talked to most finance folks, pivot tables are for the weak. We don't, we don't like them. mean, I, I, you know, I, I, I know that somebody's going to hate that I said that, but that is a hundred percent true. I feel like accounting uses a lot of pivot tables. Sorry, accounting. ⁓ but you know, finance guys don't find enough flexibility with pivot tables. They, they lock you in. They're just, they're kind of kludgy. ⁓ they can help, you know, a little bit at the beginning of an analysis, but then you kind of get away from them. ⁓ Yeah. Yeah. Yeah. I've got, I've got a ton of those, ⁓ comes with having kids. So what I started doing and where I started to see the most joy though from Claude. And then now I think I sent you an email, ⁓ received an earlier release of this chat GPT and Excel plugin. So I've used both now. played with them. I'd say Claude is still better. You know, in my experience, it gets things right quicker and it goes a bit deeper, but that's why it burns your tokens. Seems to think harder. So trying to give a concrete example. So if you're working across multiple spreadsheets and you want to create a new way or a new view of your P &L or balance sheet, You can pretty easily tell Claude use this other worksheet as a template, fonts, shading, all these different things and recreate that. So use this P and L template and just get the essence of it.

Let me interject one second so you said that I mean like that is a very like like that's a like like that's a squishy kind of a like can it really do that? mean can it can it tell that that's a good one? Okay. Okay.

No, no, that's where, and that's where I was heading. So like in my head, in my head, yeah, don't, don't cut me off. No, no, people tell me I'm long winded all the time. This is fine. So I want the essence of it. So what I'm doing is I'm going into Claude and using like plan mode, maybe in Claude code, because it's a little bit, you know, less, you know, token usage. or I'm just using Claude or sometimes I use Chatch PT wherever I have the most context about the project, I'll dive in there and then I'll tell the large language model, hey, I'm using an AI agent. Here's what I want to do. And I need it to use as little tokens as possible. Give me the exact prompt for Claude in Excel. I usually call it AI agent because I don't know which one I'm going to use Claude or Chatch PT. And then I take that prompt, which is, I might say essence over here and then say something in my prompt, like ask me follow up questions if you need more clarification. And then I'll go through what I meant by essence was I want the shading to be like this. I want font 10.5. want et cetera, et cetera, et cetera. And the formulas need to be designed a certain way.

So here, we titled our discussion today to talk around AI fluency. Fluency is what you're talking about here. It's the fact that you have gone past the, this is a magic toolbox and I'm just going to ask it to go, I'm going to rub the genie in the lamp and tell it to do stuff. And now we're talking about some very specific, I am trying to control token utilization, partially because of cost. But also, I I think both of us have seen from a fluency perspective, given fewer guardrails, it's not just that the AI is going to go off and burn a lot of tokens. It's going to go investigate and try to figure things out that you, I mean, I think you've talked about this. It goes too deep, right? It gets lost in its own head in a way.

Yeah, like if I ask it, if I went in and I said something like I just told you, if I just chatted to it like a person and said, hey, I need you to replicate the basic look and feel of this spreadsheet and do it over here, but it's gonna be totally different data. It's the balance sheet instead of the P and L or statement of cash flows, go. It will start going, all right, first, let me read the entire spreadsheet. Let me read all of the workbook. to make sure I understand how everything fits together. And you will just sit there for five minutes while it churns. And then it'll come back and say, all right, I get it now. And then it might not get it. So if I say, I want you to build a sheet named this, and I want you to do it with these, like everything I was saying earlier, fonts, et cetera. And I want you to tie it out to this other worksheet. on this specific row, that is what gets you the best results when you're working in a model that is already built. If you're starting from scratch, then maybe let it create, know, let your creative juices flow. But if you're working in a model as an analyst right now at a company, I would say you want to plan and prompt for that agent in some other tool.

Well, and I think that's ⁓ a fluency component we've talked about on multiple levels is ⁓ there's a danger, I think, that we've both seen of going kind of all in on one stack because it's almost like putting all your eggs in if you just had one smart person that you hired, right? I mean, and you're not looking at any kind of alternative point of view from ⁓ or pushing back a little bit, but letting a couple of these tools. ⁓ work together by ⁓ pushing each other's work against each other.

Yeah, yeah, a hundred percent. think, I think we're circling something a little bit here and this is how I've been explaining it to people is we are still coding. The language is just English. So we're not writing in Python. We're not writing in Java or C-sharp, whatever, right? We're not doing all those things. We're writing in English, but fundamentally we're still telling a computer what we want it to do. the computer is just smarter about it. So you've got to speak really well to the computer and give it exactly what you want it to do. And you're going to get a better output. And I would say, well, leverage the computer, the cheapest tool you have to plan and develop, and then move over to the heavier hitters. Once you have that plan completely dialed in and you have a prompt so specific that it's gonna accomplish your goal in one shot.

And so what we're saying is that a little bit, we're not even saying a lot, a little bit of thoughtfulness is the lesson in this specific area of ⁓ fluency in using these tools where if you're scrolling on TikTok or on YouTube and you're hearing these young guys and ⁓ gals saying, know, like, this changes everything and this is, you know, like Claude destroyed Excel and... ⁓ The example is a one-shot type of a prompt, right? And the context that they're working in is very different than the complexity of these tools and models that you've built, right? I mean, you're a hot key jockey, ⁓ fingers on keyboard. ⁓

Yeah. And I'm not nearly as good as the guys that are out there now. I mean, you know, I haven't been my, my skills are rustier because I've been leading people and been less in Excel, but yeah, this is helping me get back into Excel though, because I can now tell it like I would an analyst, but I have to be more specific than I would be with an analyst. I could tell one of my best guys, Hey, give me the essence of this sheet. Literally, that's, I would be able to say to a human person, give me the essence of this sheet. They all need to like be matchy matchy. You know what our structure usually looks like. Thanks. Get it to me by Thursday. Here it is. I've got a high school student who is really, really good, knows zero about business or modeling. And I have to say, go to sell a 26, do this color it this way, you know, And you can speed that really detailed process up by using a large language model to help you develop that.

Right. I mean, the part that's not the changing part is that it's not a debate. What you say to it and how you say it and the prompt, and we're not talking about prompt engineering, we're talking about context for the agent. This is not a debate on which product does it better, is Claude better than ⁓ ChatGPT. This is about the fact that as you and I hit these real world walls, as you're taking real PNL samples and you're helping clients and new clients out, that you have to get into the weeds with talking to it in a way that's structured. Again, you're not writing Python code, you're not having to talk to it dumb. In a way, I feel like I talked to another person the other day who was saying like, well, you have to treat it like a child. I'm like, well. you've messed up too, right? Because that's not the point either, right? We're treating it like someone who just doesn't have context.

Yeah, I agree. But the more, you know, in some of these tools, you can, you know, the, more context you give it, but also, I don't know, it's not just context. It's explicit extra instructions. I mean, in some of these like programming environments and you know, when you're coding the more specific, mean, you taught me the other night, you were like, no, no, no, no, Jim, don't, don't say that. You need to. I know what you're trying to do, but you need to use the exact word that's going to get you a better result. And that's what I'm saying here. But you could, guess the point is, like, can speed that process up too, by talking with a cheaper tool. I mean, you can even download a local model and do your planning for free if you wanted and, you know, feed it the spreadsheet, let it learn the spreadsheet, give it explicit instructions, but plan. Take the... 15 to 20 minutes to plan the one prompt, and then you're gonna get an outcome. Now you really did only spend 45 minutes on something that's gonna take five hours. So that's my point on that.

The, ⁓ I think we like that that horse is dead. think the idea about fluency, fluency in the context of, you know, right tool at the right time, fluency in the fact that as we use these embedded tools, we have to give it a certain brand of explicit instruction, but we can't downplay that that explicit instruction set is

or self-touch

It's contextual to what you're doing. It's explicit to the tool that you're working in because we can't yet get to the point. We're close maybe, but we can't do the, know, bring me the essence of this document.

You can do it. It's just gonna take you a whole lot more usage and it's gonna keep going, yeah, here, try this. And you're gonna be like, And then you're gonna continue to give it more and more explicit answers, but you'll have burned through 20,000, 40,000 tokens.

Yeah, I think, you know, one thing maybe will land the plane today, kind of going from fluency to talking, ⁓ kind of landing the fluency plane around ⁓ one of our favorite guys, you probably hear us talk about Nate B. Jones on our podcast, because that guy is deep, he is deep in on all things AI related, and ⁓ maybe he will like us one day and ⁓ come do a three headed ⁓ monster. ⁓ podcast with us. That would be amazing.

I would love it, but it basically would be me learning from him. Like, yes, what you said, Nate. Nope. No notes. No further questions.

No notes, no notes. He talks a lot, he uses the phrase, ⁓ you know, having taste and that this is a differentiator and maybe one of the differentiators that we're going to be seeing as emergent in the use of AI, being fluent with the use of AI, being able to use it effectively, ⁓ both at the enterprise level and at the personal level is taste, you know. ⁓ And one of the pieces of it is the angle of just because you can doesn't mean you should choose what you're going to invest your time, money and ⁓ experience into because ⁓ for some of us, we ⁓ have a very ADHD approach to life and we bounce around a lot and ⁓ it gets us in trouble. ⁓ But this idea of taste though, plays out in the fact that we know what good looks like. We have worked with consultants from the McKinsey's and Alvarez's of the world. We've worked with independent folks. We know, I think, the two of us are pretty self-aware. And ⁓ we know when we've worked on something and we've phoned it in. And we know when we're proud of the work that we've done. ⁓ So I feel like we do have good taste. And part of working with these products is training them in a way to understand taste. And ⁓ so one of the things that maybe we've talked a little bit about, we've circled it a little bit, is that when we pit these tools against each other to try to refine our prompting or our instructions, One of the things both of us have found is that we are telling the system, no, I don't like that. I need it to look like this. Right? We are. Yeah. Right. And so this idea of fluency is partially in being able to be the adult in this relationship between you and a AI tool and pushing back to push it towards good taste. ⁓ And there are accelerants that we've talked about on previous podcasts about ⁓ having ⁓ rag systems and having vector-based databases that can pull information together or pull examples together. Those things will definitely enhance the ability for the tools to give us answers that have good taste. And at the same time, we have to push back enough that the final product that we generate out of these tools, whether it's a memo that we're sending to a client, whether it's a spreadsheet, that it's got some level of taste that is not just good taste, but is reflective of the work that both of us know that we're capable of doing.

⁓ yeah, a hundred percent. And I think AI guilt is like probably going to be a thing if it already isn't, but you know, just feeling like you can get things done so quickly. However, like I said, there's a lot of planning that you're not going to get the same. This goes back to what we always talk about. I'll summarize it this way. 25 years of experience plus AI. Awesome. You know, you can ask the right questions. You can get to the essence of things a lot quicker than if you don't. I mean, you're going to prompt naturally you will prompt better in your discipline than somebody who doesn't have any experience in your discipline. You're going to be better at IT prompting. I'm going to be better at finance prompting. That's just the way it is because you can ask better questions. That's where we are, at least in the evolution.

I think these products are, we've talked about the, it's the limitless pill, it's helping us move faster. And the danger that maybe we're closing today's session on is the fact that whether we want to call it a retained humanness, whether we want to call it a ⁓ quality of output, as people talk about AI slop and just generated content that's out there, in between that, where people are just generating content for the sake of content, and in between that and people just not using these tools out of some kind of self-flagellation of, I'm just not going to use the tools, I'm just going to, like a serial killer, open up Notepad and type it all in Courier font. In the between there is the application of good taste, the application of the knowledge and experience that we have to generate product that is the equivalent of our summed humanity and our experience. ⁓ But we're going to be doing it probably a lot faster, at a faster rate, and we're going to be able to be more accurate because we have been able to spend more time investing into the actual analysis rather than the where, know, should I use a semicolon or, you know, an Oxford comma here.

Yeah. Yeah. I mean, I think this is a good, I mean, again, I think this is a good place to land and in the future, I think it might be smart to explore skills and how you can get yourself, your knowledge, your abilities into skills, which is, you know, that's a new file extension now, right? So, so you can get these skills built.

That is.

that you're basically cloning yourself inside of a...

So, you know, as we talked about fluency this time, this next time we get together, we'll be talking about the, you know, the ability to ⁓ clone yourself or create ⁓ the ability to have taste, to have... ⁓

Yeah. we're going to talk about the curation of personal taste.

⁓ this is... Go. All right. Well, I mean, I guess ⁓ for our friends who've been listening and who stuck with us this far into a conversation and is that ⁓ I'm Veer.

This is And I'm Jim and we are the functional executives on most days. And we would love for you to subscribe and keep listening. If you like this content and also, you know, feel free to drop us a line.

All right. And we'll come back, come around next time to talk about the curation of tape. You're so much better at words.

I said that- I'm not better with words, but I said the curation of personal taste.

We'll see everybody the next time around.