The Rise of Tokenmaxxing: Analog Organizational Risks Just Got an Expensive AI Upgrade

Speaker A: You may have heard of the term looks maxing or the latest trend, balls maxing, neither of which I would encourage you to look up on Google. Um, but ah, essentially what they are is obsessively gaming features for nothing but pure vanity purposes. That's what they are. And now you may be thinking, why are we talking this about this on Future Focus? Well, we're not talking about looks maxing or balls maxing. We're, we're talking, we're going to be talking about this more corporate future focused related trend, which I don't know if it's officially termed, but I'm making it a term now and that's token maxing. And what token maxing is, is where corporate employees are obsessively gaming AI tools and usage to make themselves look productive. And this is the output of enormous pressure coming down from boards and just the industries in general. That's leading executives to feel enormous pressure and their response is to create and enforce arbitrary usage mandates that they then just push down the chain onto all of their teams and kind of this broad stroke brush that they push across the organization. I'm Christopher Lind and this is Future Focused and this week I want to spend some time talking about token maxing and why telling your workforce to just use more AI without tying it to business outcomes and other things like that, it's not going to lead to innovation, it's not going to lead to greater productivity, it's not going to lead to greater, you know, gains in your organization. All it's going to lead to is all of your people looking for spending time and finding ways to, to make the dashboards you create look green for them to survive. Because we are entering an age where we've actually moved past AI productivity and are, uh, now largely stepping into this age of just pure unadulterated AI theater. So that's what we're going to be talking about today. Before we get into it though, if you find my content helpful, if you like it, it would mean the world to me. If you like, subscribe, subscribe, share it with somebody if you think they would benefit from it. And um, yeah, also if you want to say thanks officially, you can go to buymeacoffee.com and look me up. You can say thanks there or check out my website. You can see how I help organizations, uh, with navigating really what I call the intersection of business, tech and people and solving impossible problems in ways that really raise the tide for everyone. Okay, so let's get into this because let me Share some of the backstory to where this inspiration came from. I've been watching this slowly start to bubble up, but it's really hit ahead and it was largely this episode. I've got a couple episodes lined up here, but this one in particular was inspired because I came across some articles about Amazon. Okay, Amazon is in the news right now, big time for this. Um, and if you've been following them, you've known they've been setting targets. They famously set targets that they wanted their developers, um, you know, 80% of developers to be spending time using AI doing their coding work each week. Uh, and to do so, they, they've been supported with the tech tools. They were given something called Mesh Claw, which is kind of like open Claw, um, to help automate and, you know, agentize all of these tasks, whether it's, you know, automating, coding, triaging emails for them, interacting back and forth with Slack, ultimately using AI to do as much as humanly possible. But where Amazon, you know, really is made headlines for it was they really were ramping up how they were doing this, they were gamifying this, which if you're not familiar with gamification or game based learning or game based, you know, anything, it's really about adding game elements that make things more addictive, incentivize things like that, which it has its place. But I've taken a couple graduate level courses on it. And let me just tell you, while gamification, let's gamify. This sounds brilliant. Um, let me just tell you, it comes with a lot more unintended consequences than people realize and often leads to more damage than good. But anyway, this is where Amazon went. They leaderboarded this, you know, they were token tracking teams, competitions for who's using the most tokens, and all of this stuff. And what broke recently, that is what inspired this week's podcast was now some of these Amazon workers are coming out admitting that they're basically just using AI by looping these meshclaw agents to just burn tokens, to just go through repetitive functions to run, you know, code, then recode, then review the code, then recode. Like they're just running cycles of AI on completely meaningless, useless activities. Just rework, truly just rework most of the time, um, for the sake of burning up tokens and looking like they are leading, they are ahead of the game with AI utilization. Uh, one quote that was in this whole, you know, interview thing that I looked at just said there is enormous pressure, there's so much pressure to use these tools A lot of people and of course it's always going to be downplayed. Most people aren't going to admit to it even though they're doing it. But they said, you know, some people, they're just using meshclaw to maximize their token usage. Oh really? This article's on Fast Company if you want to look it up. But it was also, they also were on futurism and other things like that. Um, but if you're thinking, well that's Amazon, Amazon, I mean we're not Amazon. We don't have our developers all using these AI tools. We aren't mandating everything to our employees. Well it's not just Amazon. Managers at Meta, Microsoft, OpenAI, Google, Shopify, they're all starting to do the same thing. Token tracking is kind of the big thing, which I'm saying token maxing, um, so token tracking is what the company's doing. Token maxing is the employee response to it is essentially how these things are working together. And it all is tied to this just measurement of AI success and productivity as a direct measure of AI consumption. Um, and it's making its way into performance reviews, all this stuff. Now you may say again, well Christopher, all those companies you just listed off, those are big companies. Or you might be listening to this going, oh, those things you just described are us too. Um, either way, when companies start making these moves, it continues to spread. It just continues to move, it continues to expand. And that's really where we're at right now is this is already starting to show its rifts in these big companies. But I talk to lots of smaller companies, lots of non tech companies who are following these exact same patterns now transitioning over. You may be thinking to yourself, oh great. Yet here's another thing, you know, huh, I can't deal with this with everything going on now I've got to learn about token maxing and how do I mitigate the strategies of token maxing and what do I do to get around all this? Well, the good news with this one is that this is not. I mean it's new in a sense, but it's also not new. Okay? This is old bad management rearing its ugly head. This is just a cycle. The new flavor of bad leadership coming into play or just short sighted leadership. I don't even want to say bad leadership. In some cases it's well intended, but it just goes off the rails. Excuse me, because really this just ties back to human psychology and its implications on corporate culture and the way organizations work. So yes, token, you Know, tracking and token maxing are new. That's the new thing. But this idea of arbitrary vanity metrics resulting in vanity gaming is not new. So let me give you an example that might hit home. Um, and it'll just be a little bit of a humorous break on this. I don't know how many companies I've been at where I've. We've done the same thing, but there's one in particular that comes to mind. And my guess is anybody listening to this who's been in the corporate workforce for very long has probably, probably encountered this at least once. But I remember the first time I experienced this, I was in a company and all of a sudden, in an effort to increase productivity and be able to more strategically manage our time and, uh, all the good corporate jargon that went around it, they wanted us to start time tracking. And so every organization, every team now had their time tracker. And at the time it wasn't anything too fancy. But basically you had to go in once a week, at least at the end of every week. And you had to go in and allocate how you spent your time that week. Okay, you got the list of categories that you could put in there that you had to pick from and you had to break it down and all this stuff. And there was a goal of being x percent effective. So you had to use this many of your hours each week and all this stuff. And let me just say that if you were to ask the question, did it make any of us better at our job? It did not. Did it make any of us better at managing our time? It did not. Uh, did we see any productivity gains out of this effort? It did. No, we absolutely did not. If anything, we went backwards because I remember being in these group chats and these group, um, messaging threads where everybody's talking about how they're doing it and what they're doing and how they're managing all this stuff. And, you know, when are you guys going in and putting in this so that we make sure it looks right and how do we make sure we're balancing all this stuff? And what I'm telling you was I specifically remember sitting there watching this going, this is absolute madness. Like, we are, we are wasting more time trying to game the system and manage all this stuff. And I was new to it, so I just didn't participate. Like, I was just going in, being honest and saying, you know, here's this stuff because I didn't know any different. And I'm watching everybody around me go, okay, we Got to do this. Hey, we got to make sure we hit this percentage. So, so, and so you make sure you put these hours here so we can balance the T. And I'm like, this is absolutely ridiculous. Um, but I remember the part that made it even crazier was we would, we would get email updates from leadership at the time and they would talk about like, hey, we're seeing too much utilization in this area. We really need to make sure we're spending more time, you know, on this or what, whatever. Like, these are the new goals, right? The new goals would come out, the new metrics would come out, and wouldn't you know it, on the surface, everything looked green because, wow, all of a sudden, especially because there was accountability behind it, those dashboards started lighting up green in all the right areas in no time. And you'd see these emails go out as they're telling everybody, you know, we're so happy to see everybody spending, shifting their time from this to that. And I'm like, I'm on the group messaging threads. Nobody's spending more of their time on that. They just change their practices. They're all communicating with each other to make sure the spreadsheets are filled out appropriately. Um, but again, it was just one of these things. It was a total nonsense burger because even with the accountability of, hey, we now, you know, these dashboards, these metrics are going up to the most senior levels of the organization. It was not making anyone better at their actual job. It was a complete waste of high value hours investing time in just feeding this nonsense machine to make certain people feel better about it. But the problem with it was it actually had this cultural rot effect that actually continued to bleed through. And this is something that I've observed on many occasions even since, because what ended up happening was everybody in the company, especially when leaders in the organization would celebrate these shifts that we all knew the employees all knew weren't real. It was nothing but vanity metrics. You know, people just gaming things as people watched and go, uh, you know, so and so gets an award. Like, I know they're the, the ultimate gamer, you know, oh, wow, this person's 200% productive. No, not. They literally just lie in the spreadsheet, you know, but they're getting awards for this and, and there's celebratory emails going out and there's new things happening. What actually happened was it completely eroded and corrupted trust in the organization because everybody just went, nobody actually cares about real value. And actually, I remember watching some of the actual, like High performers, like actual high performers get penalized, or some of them to not get penalized, would actually stop being high performers and shift over and adopt these gaming metrics. They'd learn from the bad actors and go, well, I mean, I used to be really good at what I did. Um, but being really good at what I did gets me a, a ding on my report card. And so I'm just going to shift over to basically, in this case, it was time maxing. You know, we'll just time max. And completely eroded performance in the organization, but it also eroded the culture. And so that's the whole thing. If you measure inputs, if all you do is measure arbitrary inputs, you get complete fiction. And so if you really want value, you have to measure outcomes. And you have to move away from this idea that, well, if we just measure the right inputs, then we're good. Now, to be clear on this, I'm not suggesting you don't measure inputs at all. What I'm saying is you can't put all the weight on that. And especially now there are ways to measure inputs that don't require additional effort. Um, so that you can subtly see, just so you can see like, hey, where are we baselining all these other things? But the accountability should not be on inputs, because if somebody can do it without maximizing the inputs, then that's actually oftentimes the person you should be learning from, not the person that has to maximize these inputs to be able to get the same outcome as somebody who's not. Um, so again, it's not throw out these measures. It's not abandon all this stuff. It's about saying, where are you actually putting the emphasis? And so again, the good news with this is how do you fight against token maxing? Well, don't put all your energy on emphasizing the token usage and the utilization of AI tools. That is m assessing and measuring the volume of time and energy people spend on using AI will not go well for you, I can promise you that. Now here's the thing, because I hopefully by this point have made the case that it's not good for overall outcomes, it's not good for productivity. You're just resulting in people, you know, wasting a bunch of their time. And, you know, you're eroding your culture, which, I mean, these are, these are all really serious things. So don't get to the end of that and go, oh, that's it, that's not a big net. That's not what I'm saying. But I'm saying those are all really, really serious things. But here's where AI adds a new layer. And so like I said, the good news is addressing it is, is not a brand new thing. But here's where AI, like I always say, is an amplifier, an accelerant of risk. Because this new token maxing, um, we're shifting from wasted time and eroded culture. You're going to get all that and on top of it you're going to get complete and utter financial destruction. And you might think, well Christopher, that feels a little bit intense. Really, financial destruction on that. Well, let me run you through some of the stats on these tokens because again, we gotta battle back this myth that AI is free. AI is not free. And just because it looks free to you, it is not. It is costing organizations a lot. And I had a conversation last week where I said one of my biggest concerns is, well, you'll, you'll, you'll hear where this comes in. But let just, this is real money going out the door. And again, so is wasted time. Don't, don't misunderstand just because you can't actually see wasted time. Hit a balance sheet. It is, it is money. But here's where token maxing is actually costing organizations tens of millions, hundreds of millions, potentially billions of dollars all in on this. So just trajectory wise, this year compared to last year, corporate enterprises are burning through 13x more AI tokens than they were last year. Okay, so here's the trajectory we're on. We're more than 10x what we were before. So that's increasing at an exponential rate. So token, token maxing is taking off. Token usage just in general is exploding, but that comes at a cost. So an aggressive AI developer, now this isn't just your general user, but in your organization. Somebody who's aggressively using a power user of unmonitored AI can easily get ready for these numbers, can easily rack up 100,000 to $250,000 a year in token consumption. Okay, you got that? It is not hard for an employee who power uses AI to burn through six figures worth of tokens in a year. Now you multiply that by an entire department, you can quickly see how we get into the millions, tens of millions of dollars. Now you think about the fact that if you start incentivizing, maximizing this, you are actually pushing everyone, you're incentivizing and encouraging everyone into this hundred thousand, two hundred fifty thousand dollar range. You're not just saying like, hey, my top users, my most AI effective users, the ones who know exactly what to Do. I've already vetted that. They're optimized. They're using it for the right things, like those people. Fine. Because they may be getting a tenfold return out of that hundred thousand. $250,000 might be worth it for your, your top players. And I don't mean top players in terms of performers. I mean people who truly know how to use AI well and are disciplined at using it Well. I don't want to discourage people from saying you should freak out about token usage and just say, hey, no more AI. I'm, um, just saying to now push everybody towards this category of a hundred thousand to $250,000 a year. In many cases, that is well in excess of their salary and benefits. Well in excess of that per person, by the way. So scale that up. But now, here's where this is only getting worse, okay? As these AI companies continue to advance their AI models, the advanced reasoning of these newer models, because of the rework and the advanced reasoning and all the cycles that they're going through by themselves, is costing 30% more tokens for the exact same prompt. So you take a generic prompt that you would have put in a older model, and you would think these would be getting more efficient, but in many cases they're not. Um, you would. You take the same old prompt, you put it in this model, you get this back, you put it in the new prompt, you got a 30% increase on that. So those numbers I just cited are getting more expensive. Now, to the, the people who are going to push back and go, well, that's only on the frontier models. Well, that's only for these kinds of things. I can tell you right now, the way culture is being incentivized and the way the AI companies are pushing, everybody is telling everybody you got to use the most advanced models, not to mention the tools themselves. They're largely designed to encourage you to maximize your token usage because they want you to spend this money. So I can tell you just from a change management, even if you sit and tell everybody, please make sure you're always. You're setting, you're switching between models, you're using, you know, Gemini, fast and only pro or thinking for advanced stuff. Most people, and I, and I've been in conversations with people with this, where they're convinced these advanced models are giving them better outputs. And a lot of times they aren't, but they're using advanced models for very basic tasks. So this reasoning tax, okay, that's we call it a reasoning tax, is now being stacked on top of that already six figure potential token usage that these cut you, you, your companies are pushing towards this. So now you got salary benefits. Now you're also going, hey, we want everybody burning at least six figures worth of, uh, worth of token usage which likely is going to increase by 30% just by the models naturally evolving. And oh, in an attempt to try and get better, they're just running through more cycles doing more complex reasoning which actually just burns through more tokens. Now here's the other part. Okay. Because if you're like, okay, wow, Christopher, you got my attention. That is getting expensive. We're not done. We're not done because here's this looming and nobody quite knows, um, what it should be or what it, what it will be. But a lot of the costs right now are, are still fairly low. Okay? A lot of these companies are, it seems, and that's why it seems free right now. But a lot of that is because these AI companies are being propped up by venture capital cash. A lot of these tech firms are being up, you know, they, they've, they're stood up on all this kind of borrowed funny money and, but now that's kind of coming to a close. The profitability window is slamming shut. More and more AI companies are facing enormous pressure to say you have to generate a profit. You cannot just keep a burn rate indefinitely. There has to be a path to profitability and it can't be just another funding round. That can't be your path to profitability. You actually have to generate revenue from your customers, which means, guess what's going to happen to those token costs. They're going to go up. They are going to go up and it won't just be, oh, uh, predatory AI companies trying to steal and exploit more cash. Yes, some of it's that, but some of it is also just realistic pressure of you can't be a business that loses money forever. Eventually going back to, I think it was last week or the week before where I said business physics still exists. To be a successful business, you have to generate revenue. And so given these AI companies are largely just burning through venture capital cash, eventually they're going to go, we got to make real money, which means we can't offer these subsidized rates on tokens. Imagine what that's going to do to your hundred thousand, $250,000 a year token consumption that you now have pushed and incentivized. Everybody in your company, max it out, everybody, max it. Run Meshclaw on repeat three times to try and you know, like you have now built this muscle in your organization saying totally max it out. People are competing with each other to be on top of the boards. They are, you know, you, it doesn't even matter if it's accomplishing anything. It doesn't even matter if it's accomplishing worse outcomes because hey, I'm at least at the top of the charts and have been forever. Imagine what that's going to look like once these advanced reasoning models continue to kick in and then the subsidies break out. I, I've run the math on some of these things. This has the potential to truly bankrupt companies who become over dependent on this. And so a lot of people think the whole fear of the AI bubble popping is like the AI companies collapsing. I think in some cases I don't use the AI bubble analogy. I say kind of the skyscraper, the top floors that got built too high come crashing down to where the, you know, the concrete set. Um, I actually think in some cases it's just going to be companies that they became too dependent on something, on subsidized costs and over incentivized usage, you know, unproductive usage, and that all of a sudden the bill came due and they now were dependent on and they couldn't undo it and it ended up completely flipping them. And I think this is very much what we saw with the housing bubble. But now this is just what's happening with, with people. You know, they're, they're taking out AI debt they can't actually afford, but it looks like they can afford it because everything's saying you can afford it. Just you know, know it'll work itself out in the end. You'll, you'll figure it out down the road. Well, what they don't realize is they're building all the infrastructure and internal processes to run on this current thing. So when it actually comes to what it actually will be, they can't sustain it. And so that's why I wanted to do this episode, to really go, hey, like I get, it's fun to kind of poke fun at token maxing and joke about it and go, huh, Ha. Look how everybody's gaming the system. But this is actually a real problem, not only just from a people, from a culture standpoint, but from a financial and operations standpoint. This could bankrupt your company. You go too far with this. You may not be able to afford to run the processes that you have migrated over to AI because of how they've been built. You may not be able to afford the amount of money you are spending on tokens once the actual cost sets in. So you might be going, wow, okay, well that went from entertaining to stressful. Um, so I don't, like always, I don't want to end with a, ah, hey there. Now I've got you all freaked out. Have a great week. Talk to you later. I want to talk about some tactical steps you can take to try and this week to try and either prevent token maxing from happening in your organization or getting your arms around it before it gets completely out of control. Now the first one is, do you know where those tokens are hitting your budget? Do you know and are they. Because this is one of the things where a lot of business leaders just kind of, they don't really look at the fine print of their budget. They don't necessarily see where AI SaaS and API token costs might be hitting the ledger. And in some cases they may not. It may be centralized budget, but believe me when I tell you it doesn't matter whether it's over in it's budget or in yours. When the bill comes due, you will end up having to help foot the bill. Whether the bill's over there or in your org, the cuts are going to come in the same place. But my exercise for you is go look. Where can you even find where these costs are hitting? Look for the decentralized credit card, you know, signups of these silent forces that are going to hit you. Like are they in your org? Are you paying for subscription services out of your own things, um, that are based on token utilization? Has it added an AI budget line in your thing that you know, you just kind of like it floated into your budget and you, they just kind of granted you air room in your budget to hold it. That we all know how these budget things work. When that number suddenly increases, they don't give you more space for that budget. They go, hey, you're over budget. And you suddenly go wait, what do you mean I'm over budget? I haven't even done anything. And they're like, oh, well your AI costs went up 30%. And you're like, I didn't even know I had AI costs. Well, that was this new line item we added to your budget. So that's one of my asks for you this week. Go, go find out where that cost is. Because it is hitting you somewhere. Some of it you may be managing, some of it may be hidden, some of it may be centralized. But dig into it and figure out where it is and what those costs are for your organization from There. Let me just tell you right now, if you have KPIs tied to vanity metrics, if you're tracking dashboards for inputs or token usage or, uh, you know, active hours using AI, kill it, okay? And I am not saying kill exploratory AI use. That is not what I am saying. I'm saying if you are incentivizing usage and holding people accountable to just usage, stop it right now, immediately. All, all stop. Full stop. Go ahead, tell your teams. I listen to Christopher. You can pin it on me if you want, but just kill it. Because this will create something you will not. You will not be able to put the monster back once it gets going. So put a full stop to it right now. And instead, because my goal is not to say disincentivize AI usage, kill AI utilization. Like, no, if you've listened to me for long, you know, I'm saying you should use AI, but use it effectively and bring people along for the ride. So don't stop AI experimentation, but reframe it. Encourage your people to explore. Go ahead, make it a goal that they have to find ways to use AI to effectively improve the way they work. That is totally fine. Going back to my earlier point where I was kind of jabbing on, you know, making it a goal, um, I'm not saying it shouldn't be a goal. Um, I'm saying how you make it a goal matters a lot. And when you mandate that, put in a strict 30 to 90 day window. There's got to be some range because figuring out, you know, an experiment takes more than just an afternoon or a week. And you need to put some time in, but there needs to be a clear, measurable outcome improvement for it to move forward. And that's the way you need to set these goes. Encourage experimentation. But then say, I want you to be focused on that. You need to be able to demonstrate that. And anything you're experimenting with, we need to have a hard stop where we evaluate and say, did it actually improve things? Because I can tell you right now, if you've been following my stuff for a long, you know, there's lots of things that are giving you these dopamine hits where you feel like you're more productive because things are happening, but it's actually not making you more productive. You thought this would be faster, but you're actually spending more time in delivering a lower quality product. And so you need to work with your teams to say, hey, anything you're doing with AI, going to put it on a, uh, clear window. That we're going to evaluate that. And only the things that work are we going to continue with. And the things that aren't working, we're going to. We're going to strip them off because we are not going to participate in token maxing. That is not something anyone should be actively design. I don't care how much of a trend or how viral it is right now. Um, it is. It is going to be the disaster for a lot of organizations. So please fight against this trend. Hopefully now you understand the implications of why I'm encouraging you to take it so seriously and the implications of ignoring this podcast and going, yeah, whatever, I think we can figure it out. Our people are smart enough. The rp. Our people don't do that. They would never do that. Yes. Yes, they will. Yes, they will. If this is how you do it, it's just. It's human psychology. So please take my word for it. And, um, with that, I hope you found this helpful. I hope you take it seriously. And, uh, by all means, send me your questions, send me your feedback. Uh, you know, reach out if you need help along the way. I just want to see people navigate this crazy time well. So with that, I hope you have a great week, and we will see you on the other side.