Dashboard: They Should Have Fired Themselves

[Music] Welcome to another 21 Hats dashboard. I'm Lauren Feldman and I'm here with Jean Marks. Hello Jean. Hello Lauren. How are you? I'm doing great. How are you? I am doing just fine. Gan, it's great to have you back here. We haven't spoken about AI in a little while. you've got a story that just came out in the Guardian um that I thought was really intriguing about how uh leaders are trying to leaders of businesses are trying to reduce their headcount using AI but they're not always happy with the results. Tell me about that. So this was a survey that came out. There was more than a thousand business leaders. It was done by some planning some software company called org view who I've never heard of before. Um but they um they they interviewed these business leaders and um apparently uh a great many of these business leaders had had laid people off Lauren um because to replace their jobs with AI. So they literally laid people off to replace them with AI and more than half it was something like 55% of them said that um they regretted the decision you know so they laid people off for AI. This is 2025, not 2035, 2025. They literally laid off their employees for AI. I mean, it's like that's all the employees biggest fear. Is is this across the board? Do you know what kinds of jobs these were or It was across the board mostly administrative and, you know, and and sort of, you know, marketing all back office jobs, right? So, it wasn't like they were being replaced by robots because, you know, people can't, you know, it's not like that's an affordable thing yet. So it was like back office stuff and and they didn't get anything more specific than that in the survey like what were these people doing because first of all you I listen I write about AI I'm doing this now like six times a month in Forbes and I and you know it's a big topic and people are like uh uh you know like I'm not seeing unless you're a giant company. I'm not seeing smaller companies laying people off yet because of AI. It's obviously everybody's biggest fear. Wait, it's not your biggest fear, Jean. It's your biggest fear. Not my biggest Wait, not my biggest fear. Excuse me. my employees biggest fear. It is my biggest hope. Um but it is it is a you know it when I wrote in the Guardian um I was like literally I questioned the sanity of the people that did this. You know I feel like like the people that the managers that laid off people they're the ones that should have been laid off for making such a dopey decision. And I'm going to give you the example why. You ready for this? So I when I first addressed this article um I wanted to um you know use it as a point where like big tech is duping you know duping uh people into believing that AI is so mature you can start replacing people with it. So I asked chat GP2 you ready? I said, "Give me a quote from three technology leaders where um where AI is already making businesses more, you know, you know, you know, where you can lay off employees and replace them with AI, right?" It gives me this quote back from Tim Cook, the CEO of Apple, right? I've heard of him. Yeah. The quote is, "AI is already making businesses more efficient, more responsible, and more personalized. It's a growth driver." That's what Tim Cook said. Okay. Um, and then I said, um, can you can can you give me the source of that quote? I mean, he's saying it's making businesses more efficient already and more responsive and all of that. Can you give me the source? Chad GPT comes back to me and goes, I apologize for any confusion earlier. The quote attributed to Tim Cook does not appear to have a direct source from a verified interview, speech, or official Apple communication. the chat GBT gave me a [ __ ] up quote and it didn't even tell until I told it till I had to ask it about it, right? And I'm like, you know, you're laying people off, you know, and replacing it with this technology right now, which like in the, you know, basic sense, it still doesn't work as, you know, reliably. Do you know what I mean? It was just it was just to prove the point that the people that are are thinking that like AI is so reliable and accurate and complete now that they can use it to lay people off. I mean, you're just you're you're out of your mind. We're nowhere near that. And by the way, don't get me wrong, it's it's great. I was going to say, I know you're you're looking forward to the day when you can tell everybody to do precisely this. Yes, I am. But that's, you know, I'm looking forward to the day when they I can actually regrow hair on my head without it being like a, you know, with a health concern, you know. I mean, they're all things that we have that we have visions for. I mean, I use chat GPT. It helps me find hotels or translate or, you know, fix, you know, give me advice for fixing something or whatever. It's great. It's it's all fine. But that's all like and and most of the clients I know that are using these these, you know, AI platforms are doing just that. I mean, you know, it's like reviewing contracts or, you know, writing an email better, you know, or creating a blog. It's all good. It's all good, but but but replacing your employees yet, you know, because of, you know, AI, I mean, that is it's unbelievable. So, in my guardian piece, you know, I just wanted to, you know, just make sure that everybody knew that why this is not reality. I mean, number one, this technology like doesn't exist right now that's going to replace employees. I mean, Microsoft in Wait a second. I mean, there are things that AI is already really good about, and you and I have talked about them, and and you know, I've done other podcasts with the the regulars on this uh podcast who, you know, they use it for things like creating a proposal in, you know, in 10 seconds that would have taken three people two days uh to create. Um, so I'm I'm sure there are times when people find ways to use AI that either frees their people up to do something else or to work somewhere else. Oh, exactly. It is a great productivity tool. You know, now when you look at like Microsoft right now is rolling out this year their AI agents and a bunch of them, a bunch of these, you know, uh, business platforms are doing the same thing. And um I'm warning anybody listening to this conversation, these AI agents, I mean, don't go firing employees because you think these AI agents are going to replace them. I do think one day they're going to replace people. I really do. I think we're a few years off from that, but it's coming. In the meantime, though, keep your people and make sure that they know how to use these agents so they can be as productive as possible and we'll see what cream rises to the top when it comes to your employees. But right now, these agents, these, you know, these tools, they're immature. They're limited. They're unreliable. I just, you know, you're not laying people off in 2025 and replacing them with AI in your back office. You know, secondly, um, you know, AI is going to have an issue and continue to have an issue with just security and privacy of data. you know, I mean, to really leverage all of these things, you're basically allowing Microsoft and Google and OpenAI and all these software companies to grab your data to let these agents process it. And I I have a lot of companies that are very hesitant to do that. Ultimately, I guess they're going to have to just, you know, drink the Kool-Aid, but there's just there's just a hesitancy for that, you know, and, you know, and and building these systems are super expensive. I mean, I'm just saying that there is there there's big things coming in the future and I get that and AI is um is going to make people a lot more productive, but um right now this is not it's all hype right now. I mean, you just, you know, all we should be doing is, you know, watching and, you know, playing and testing it, keeping an eye on it. Um, but again, we're we're still a little bit of ways before we're going to start relying on this stuff to like do our core business operations and then lay people. So, it just surprised me and it didn't surprise me, excuse me, when these survey respondents were like, "Oh, yeah, we laid people off to replace them with AI and now we're regretting it." Like, yeah, you dope because that's You don't do that. Did they talk about how they figured out that it was a mistake? Was No, I mean, again, the survey didn't reveal that either. I mean, I guess maybe because people are just too ashamed of themselves for actually admitting that they just made such a ignorant decision, you know. So, I just, you know, I mean, you know, again, you know, it's funny in the Guardian article because I I cover this stuff. So, I've been writing about these companies like Boston Dynamics that have these, you know, unbel, if you ever really want to see some crazy stuff, go on to YouTube and look up Boston Dynamics. They're they do robotics. So, they have these very human-like robots. And by the way, Nvidia and you know uh you know Tesla are developing their own robots very humanlike um with Google type technology where these robots can sense and see things around them and really do some amazing stuff. It they're all still first of all they're expensive right now. there's still a couple hundred grand a piece and there's still um you know we just we have a ways to go but when you look you know when you look down you know the pipeline you can definitely see these robots coming into warehouses and cleaning offices and uh uh you doing tasks that you know businesses are are you know people are doing right now the other thing is I wrote um in Forbes yesterday as well this past Sunday there's a company called Aurora um they're now rolling out auton autonomous trucks in Texas and they've gotten permits to do it and they've got autonomous trucks now driving down highways. Um, they get to a a populated area and they pull over and then a human gets in and picks it up from there because you can't trust them to do that stuff. But that that's massive. I mean, you know, truck drivers are expensive and, you know, their insurance issues and, you know, healthc care issues and all of that. And this technology is it, you know, it it's getting closer and closer and I think within a few years we're going to be seeing a lot more autonomous trucks on the highways because they'll just be safer and more reliable than a human. Um, so all this is coming. I get that. I just get that. But the point is like in 2025, um, I was just blown away that people are actually still thinking they can replace humans with AI. It's just not reality right now. you you mentioned the uh AI agent uh thing that Microsoft is rolling out. Yeah. As I understand it, the the cell with that is that businesses all use 10 or 12 platforms, you know, for one thing or another. And with this kind of agent AI, you can bring it all together and just have one platform that'll do your CRM and your payroll and, you know, your word processing and everything else that you do on all these different platforms. Bring it all together in one place. Simplify everything. Make it more efficient. That's not my take on it, Lauren. That's not my That's not my take. My my my take on it is that they're they're still going to have different platforms. I think the the the mecca of business software is that we just do have everything under one roof and it's going to take a little while, you know, to even get there. That's been a complaint of clients ever since I've been in this business. Microsoft, for example, still has like two different accounting applications, you know, under its Dynamics umbrella. Then it has uh CRM applications and Outlook and Office. It's got data everywhere. It's still a disaster under the hood, you know what I mean? Um the agents though are specifically for dynamics. So and for their financial dynamics. So you know the agents are supposedly like you tell your agent, hey um uh you know you um create a purchase order for this supplier um with this range of pricing that I will allow. And then the agent just goes out and does that. um when you know when when the end of the month comes their financial agents will just close the books, make the recurring journal entries that an accountant would make um and then you know check the reconciliation between the cash accounts and bank accounts. These are all repetitive uh you know kind of you know uh easily replaceable tasks that a bot can do as opposed to you know a human doing. Do you know what I mean? Yeah. And the same thing like both Dynamics and Salesforce are introducing agents where you're going to you know you talk to it and you say go and you know go and you create a campaign for everybody with blue eyes and green hair who's an Eagles fan um and send them an email that tells them this and then the agent just picks up the ball and runs with it. That's what that's what these companies are saying that their agents are going to be doing. And uh it makes me shudder when I think about that because like you know there's like you know there there's it's we have a long way before we can rely on agents to do that right and then they're sending out emails to our customers you know in our community. Can you imagine? I mean it's just you know I just see that I I see a lot of problems. So you're not going to be one of the first adopters? No I'm not going to be adopting that right away. But I think it'll get better and better um with it. I think what what the key to it all is going to be the data, Lauren. you know, they're they're um you like I don't I wouldn't rely on a Zoho agent to send out emails to my Zoho database because my Zoho database is still, you know, screwed up in so many ways. But I think it'll get better at validating stuff like the agents, the next generation of agents um before it sends out the email will check the data in my Zoho database and then validated against external data somewhere else uh email addresses or mailing addresses or name spellings or whatever. Um make corrections and then go forward with the campaign. Do you know what I mean? Yeah. And I and I think as it gets smarter and smarter, I think businesses will then uh be able to rely on it to do those kinds of stuff. You know, the other thing I want to just say is um you talk about like the different databases and and I'm talking about agents. So, Microsoft, Anthropic, Google, um they've and you know, a few other of the big tech companies have released their agents that control your device, you know. So, you can see this on YouTube as well. Um there's you you you tell it there's an example that um one I forget if it was Microsoft gave but you just tell your computer your your laptop you're like hey um go out and find uh 10 you know here's a sorry here's a list of companies that are my competitors um go find their websites find out who their CEOs and their VP of sales are get their contact information and put it into a spreadsheet and then you watch the video and it just starts like you probably starts like just clicking through screens, going to websites, grabbing the data, and then putting it into a spreadsheet, and then it all finishes, and then here's the completed spreadsheet just like you would have like a high school kid do, you know, on like a separate project. That's what the agents are doing right now. I mean, they've got agents that do that right now. And and um you know, that's what they plan on rolling out for businesses to use. It's exciting stuff. Um but it still will need some testing. So, I I guess I got my wires crossed or my stories crossed. I' I've certainly read about agents the way you've described them. I thought part of the uh the pitch was also the idea of bringing everything together under under one roof. May maybe I hallucinated that. Um I hallucinated. You mean you were wrong? Yeah. Uh but that is one of the ultimate goals though, isn't it? Are we getting closer to that? Yeah. I mean inching in that direction. Um you know it's even Zoho which we sell. They say they're in this one business one system but they still have different databases and they and they're positioning themselves as an integrated you know one under one umbrella kind of thing. Um most most data most systems that are out there these ERP systems um are still cobbling together data from different places. The only way you can really get there is if you just drink the Kool-Aid and buy into one big platform and and you know suffer the features that they might have that you don't like or lack of features because you want that integration. That's the choice that you make. Like for example like say um I was just at the Epicor conference. They're like a big distribution and manufacturing you know software company, right? So very good. They've got great great modules for distri, you know, distributors and manufacturers and and that's fine. Um, but that, you know, they also have CRM modules and they also offer a little bit of HR. Now, the CRM modules aren't as good as Salesforce and their HR offerings probably aren't as good as some other companies that just focus on HR, you know, and they offer some project management, but it's probably not as great as like an ASA or a Monday. You know, there are companies that really do it. That's what they do, you know. So, but if you're a company, you're a customer of Epicor, you might be like, you know what? Um, all right. They're not going to be as great as some of the greatest applications out there, but they're good enough and I'll I'll, you know, I'll I'll give that up so that I can have everything under one hood. Do you know what I mean? Um, and that's a choice that a lot of businesses are going to have to make. Isn't somebody going to try to do it like chat GPT? just try to bring you you can bring your own platforms together um on under our roof and our agents will allow you to use Salesforce plus Rippling plus whatever else you use. Yes. So this is a I forget if we talked about this before but it's worth mentioning I think we did but it's worth mentioning again because these platforms um there are many tools that are available right now that didn't exist a year ago. Um there are integration tools like Wiccado and Zapier which can take data from different different softwares um and bring them all into a large language model that you license uh from OpenAI for example or Llama or you know Enthropic and then there are other development tools to turn that data tag it and and turn it into unstructured data basically into like a giant chat GPT large language model which your people can then start querying and asking it to do stuff and then you can start running agents on top of that. You can do all that, but you know, I ain't doing that. You ain't doing that, Lauren. No offense, right? I mean, you can barely plug in your TV. I know you. So, you you know, I mean, you got to hire developers to do that stuff um and bring it in, but the good news is that all the tools are there. Um and and which a lot of these things didn't even exist a year or two ago. Um so, you can do it. You can build your own thing, but you're going to have to hire a $200,000 a year developer to bring all this stuff together and build it and synchronize it and support it. Um, if you really want some of that, you know, to be happening. Even the Epicor, I'm using them again because I was fresh in my mind. They provide their own way to do that. Like they could they because I asked them and they're like, "No, you can take everything from Epicore and bring it into this. We have a our own large language model and then you can use these integration tools like ricado to bring stuff in from your Salesforce or bring it in from your HR or whatever and build your own thing out and which is great. So they're they're all doing this. They're providing these tools but um I'm like I'm not it's not going to be for me to do but you know it's there. You're right. We have spoken about that before. Um, and I suspect not too many people listening to this podcast are are ready to take that leap and hire a developer and do that. Although I did hear from some last time after we talked about it who who were intrigued by what you had to say and are thinking about it. All right. I I got another stupid question for you though. What what's the difference between that which you just described kind of creating your own smaller large language model and doing what chat GPT already offers which is kind of create your own GPT where I know people are already using that. I um someone on this podcast talked about using it for uh her public relations firm where she has created a her own GPT with brand information about one of her clients. So she can turn to that GPT and say create a proposal specifically or a campaign specifically for this client. You already know everything you need to know about this client. what's the difference between that and the more ambitious uh so chat GBT itself is um they're competing with a number of uh applications out there the most wellknown is called is Google's notebook LM very similar concept where you upload files is what you're doing you know and you are you know which is what you're the person you were talking to is doing um and that is an absolutely great thing for a business say I think we talked about this like you know a while ago where I started the process and then stopped only because I got distracted. I'm going to get back to it to upload all of our contracts and invoices and documents, proposals and estimates into Google notebook LM. Same as this chat GPT, you know, example, and then that way I can start querying it and having it do stuff for me. No, it's not. It's very similar and open AI because you don't need a developer to do what the version that I'm using. You will need a developer if you're starting to s starting to bring in data from other systems, you know, like just uploading files is one thing. Okay. Sure. But if you're like again if you're in an ERP system, if you're using Quickbooks, you're using Sage, you're using Epicore, you're using Dynamics, whatever. Um it is it it is not an you know for the for the uh faint of heart to to bring that data into your open AI system that you got to develop and in you know and and it's got to be synchronized and brought in in a certain way and that's that's the only difference. But no, the that that that that example that you're given, it's like a it's like a a first steps for people uh to get into developing their own large language model, but if you want to bring in data from different systems, I think it's you can still use that, but you would need to be hiring somebody to do that unless you want unless something that's up your alley. Do you know what I mean? So the it's kind of baby steps. You can use the create your own GPT system if you're just uploading your own files. Correct. Um, I got it. Have you done that? I have not. Um, but it's definitely I'm going to be writing more about it and I want to talk to companies that that are doing it. Um, or at least like stepping in. So, that's something I want to cover. Let me know when you do because I want to read that. I will. Um, you also wrote a piece recently about Forbes that touched on some of this stuff. It it it gave uh a bunch of takeaways for business owners. Anything we haven't already covered here? No, we pretty much did. I mean, that's, you know, the the the that's the latest with AI. It's still a big company game. Um, you know, the the AI chat bots are getting smarter and better and which is all great. Um, you know, if you want to build something yourself, you can, but it's still relatively expensive and complicated, but it's getting easier. Um, so patience, my friend. You know, you know, you did mention one thing I wanted to bring up in in that article. Uh you talked about regulation a little bit and how you're not expecting much coming at the federal level, but states are looking at it and they're focused on a couple of things. Uh I think you know misusing AI in the hiring process and kind of uh duping customers with questionable AI bots that maybe promise things that they can't actually uh deliver. It it sounded like a a a good warning that businesses should be aware of, especially since this will vary from state to state. Can you tell us anything about that? Yeah, this and by the way, this comes I I wrote a piece on the the 2025 Stanford AI index. There was, you know, a bunch of researchers did a whole bunch of update on the the status of AI, you know, in 2025. So, I I pointed out a bunch of things, five in particular, but um one of them is exactly right is about regulations moving to the states. Um, you know, right now, you know, they say in 2026 there were just, you know, a handful of, uh, you know, state regulations. Uh, now it's it's it's going up to, in the past year alone, there's like 131 regulations at a number of different states um, around the country. And it does focus on people and hiring and you know uh not using AI uh in a way that could be biased and not using AI that can dup your customers um into doing stuff that you know could be potentially misleading. So, but you know my you know if you are starting to implement AI in your business and it's going to be either customerf facing or in the hiring process I strongly recommend that you run it by a labor attorney and make sure that you've got no specific laws in your states that uh you need to make sure that you're in compliance with how do you run that by your attorney if if you're using AI to evaluate resumeumés or something like that. What what's the conversation with an attorney like? Well, it's this. It's just like if I'm if if you look at the Stanford uh you're it's again it's called the 2025 Stanford AI index. They have a chart in there. I didn't include it my Forbes article, but it's a chart in there that shows all the states that have AI regulations. So, if you're in Illinois, the usual suspects, right? California, New York, Illinois, you know, if you're in like those states and others. Um, and again, you are using an HR platform that uses AI or maybe you're building out a you're you're adding a chatbot to your to your website that's using, you know, AI for automatic responses and things like that. Um, that's the that's your trigger for reaching out to, you know, a labor attorney and saying, "Okay, well, I understand my state has got some regulations about AI. this is how we're using it in our business. Are we are we cool? You know, that's how you do it. Got it. Anything else you're working on that we should be looking for this week? Um, yeah. I'm writing a piece on private equity, Orin Lauren, for uh for the inquire this week. That's, you know, I mean, I don't know about you, I'm getting like 10 emails a week about private equity firms wanting to buy out my stupid little business. And uh but I do have a lot of clients that are moving in that way. Private equity industry is just, you know, just ballooned in the past 20 years. they don't have the best reputation. Gene, are you going to deal with that? So that's why so I interviewed a couple of people in that business with some offering some advice for business owners that uh are are thinking of doing something like that. So it was interesting sort of um what to watch out for, how to pick the the right private equity firm. Yep. Yep. Yep. Yep. Right. To how do you pick them and you know how do you align with them and you know um and then if you do decide to go down that road, what can you expect you know it to be? So, um, you know, because sometimes people, you have to have your eyes wide open if you're going to go down that path. You're not going to sell your business out from under me anytime soon, are you? I am not go. My business ain't worth [ __ ] Lauren. That's a whole Why are they contacting you? I Oh, they're just bot emails. I'm getting I mean, they have no idea. AI. Another AI mistake. No idea. They have no idea. Gene Markx is a CPA who writes weekly on small business for the Guardian, The Hill, The Philadelphia Inquir, The Washington Times, the Chicago Daily Herald, Forbes, and Entrepreneur. You can also hear him on ABC Radio's Eye on the World with John Bachelor. Gene hosts two small business podcasts with Paychecks Corporation and the Hartford. Thank you, Jean. Thank you, Lauren. Have a great week, everybody. [Music]