582 10 Accounting Firm HIRING Traps

For most of us, how we employ other people and pull them into our team and, and, and build systems to make them as effective as possible. How we do that with other human beings and bump up against them each day, that will dictate the impact that you and I ultimately have on the world. But working with other humans, it is hard. It is, it's just. It's tricky, man. So hard that a lot of people are like, I'm going to go out and do my own thing solo, become a business hermit, which no shame. That's about 50% of my days right now as I wish I had chosen your path. But today I want to share with you the 10 biggest mistakes I made hiring inside of my accounting firm. The stuff that we got wrong, the changes we made that really improved things. But also because we got a jam packed episode today. We're going to be building an accounting firm. Tech stack for a non accountant who just built a firm is a true story. I'm going to give you this firm. No website, no social media. It's gonna be a fun build, but also an oops all AI edition of the mailbag because holy smokes, is there some wild stuff happening in AI right now. People have a ton of questions. We're dig ourself out of all the AI questions that I've got. By the end, you should have a clear direction on like should I even be using ChatGPT at this point? Got some great questions on everything from if you're an accountant, like where to start learning if you don't know a thing. If you're doing accounting firm rollups, how does this impact what you're doing? The best AI for PDFs, how does this impact the accounting ledger? We're going to cover it all. Okay, you've come to the right place. Get comfy. If you're new here. Welcome to the Jason on Firms podcast where we talk about all things accounting firms and ever so occasionally, not that occasionally, quite a bit AI because how can you not in literally any space. But first, hiring, man, the 10 hiring mistakes that I made and a number of them that I continue to make that can leave you just deflated, depressed and frustrated with having to coexist with other human beings. Mistake number one was expecting that every hire was going to work out. Every ever been there ever worked really, really long finding the right person, then finally found the right person, then gotten two months into it and been like, this is not the right person. Let them go. And you're now like 12 months into a process where you're like, my soul just left my body. I do not have the energy to go through this suffering again. I've been there, been there more times than I would like to count. And I don't know the solution, but I can tell you the closest thing that I've gotten to one is changing my expectation. Very, very practically, if I needed to hire three people for something, what I would have done before was gone out, interviewed a whole bunch of people and decided on three people. What I would do these days is decide on four or five people. Because the trouble with just hiring three folks is that in my experience, like, if I am hiring people in the US which is. I mean, all I can speak to is if I'm hiring three US Folks, there is no world practically, where three months down the road, those three people are still working for me. And yes, you can pick that apart a bunch of different ways. Why, like, wow, Jason, sounds like a real mirror into your soul, sort of tyrant. Are you. If you hire very many people, you know exactly what I'm talking about. So if I really, really need three people for something mission critical, the solution, it's probably not to go out and hire three people, is to go out and hire something more. And so if we acknowledge the fact that in all likelihood, a lot of times people just aren't going to work out, then we need to bias the process to overcompensate for that, because that's. That's just the reality right now. Mistake number two that I made was not doing trial days. That is going straight into hiring somebody to come and work with me without ever even seeing how they work or if they know what they're doing. In accounting, it's very, very hard to truly assess the competence of an individual before you get them in the doors. They can, they can interview well, they can have all this experience, but that doesn't really map to do they really know their stuff. Trial days were a massive improvement to my hiring process. And trial days are one of those things where it's very easy to dismiss them or come up with reasons for why they won't work. And it's just one of those things where it's like, if you're not there yet, it's just because you. You haven't found the version that works yet because there's a hundred versions of trial days that don't work, where it's like, how am I going to get somebody to come do a trial with me when they've already got a job? Like, yeah, there's ways to work around that. How am I going to get people who are hard to hire to begin with to be willing to like make this an added inconvenience part of the process? Like, how am I going to get anybody to do that for me? You can absolutely do that. Like these are all hard things about trial days that I would argue you can actually flip and make advantages of your hiring process. So trial days, which we still use in my company now, like is a two way thing where we're learning about them but they're also learning about us. Like that's a risky thing for somebody to leave their job and come to work for a new place. Wouldn't it be nice to just have a bit of a window inside? And I know next it's like, well, are they getting the real work experience in a trial day? Probably not, but it's a lot better than nothing, right? They're meeting some of the people they'll actually work with, get more of a peek behind the curtain. Historically I've done half day trial days because it's the sort of thing you can scooch to the beginning or the end of a day to accommodate people's schedules. Sometimes we split them over two days even. But I can't tell you how many times a trial day has absolutely saved my hide, honestly. We'll probably, we probably throw half of people out in a trial day that we would have otherwise moved forward with. And it's for like technical things, but it's also for things like, okay, when we start the trial date, a member of my team's going to hop on with you. Dan, do have you share your screen real quick just to get you started. And we are looking at things like we gave them a zipped folder. Do they know how to unzip a folder? We told them to install Slack before we met. Could they figure out how to install Slack? Like it's a lot of things where having that 20 seconds or 90 seconds where my team member can see how the person uses a computer. Like isn't this all intel that we want to make sure that we have before employing somebody to come work with us? Of course it is. Like once you do it, it's kind of ridiculous. Where it's like, geesh. Like we literally just thought that we were going to be able to guess how good of an accountant this person was by sitting down and having a conversation with them. I almost said an account on a, on a conversation. You know what I mean? We really thought we would figure out if they're a Good accountant by sitting down and just talking about accounting. Not without actually seeing them do the accounting. Trial days. Total, no brainer will save you a whole bunch of bad hires. Absolutely worth doing. Mistake number three that I made was hiring too high level. Oftentimes it's easy to become over reliant upon expertise when the underlying issue is actually a lack of systems, a lack of standardization, trying to help too many different kinds of people. This is the hardest thing about running a generalist firm is you may love the variety, but staffing variety, that's very hard, that's very expensive. And if you're blocked on like not being able to find that next person, the further down the skill spectrum we go, the easier it is to find folks. This is actually what sent me down the path of my whole sort of service library approach. I've got a bunch of con old, maybe year old content now around that, but around how we standardized work in our firm. I eventually like my hand was forced and we had to go and kind of revamp all of our workflows because we just couldn't hire the next really high level person that we needed. So we had to somehow figure out how to get the same work done with less expertise. And that standard process is how we got there. Mistake number four, I won't labor on this one too long. Not hiring offshore because patriotism is not a business strategy. And you are like what did it for me was meeting a couple really sharp firm owners and they're like oh yeah, no, this is just, this is what we do. Like it's, it's just normal. And I was coming from an 80 year old firm where that was unthinkable for all the reasons that it is for most firms still. And I was like if this is normal for them and it seems like they're pretty sharp and have things figured out, then I, I need to rethink my approach here. Mistake number five, not hiring admins. When I first bought my accounting firm, admins were the people that you hired to pick up the phone when it rang. And you, you hired the lowest level of capability and expertise possible to just be able to pick up the phone and say hello and complete whatever that request was. I like really after buying the firm, I spent an outrageous amount of money on admins in a way that was like unthinkable for the, the outgoing guard. And it's because we started using admins in additional ways. And so admins began providing more direct support to the partners. And they had tried this in the past with the Very low level people that they were hiring. But as you may expect, once we hired higher level admin folks who were coming in more as like ops people, they were vastly more useful to the partners. And at the end of the day, whose time in the business is most important? Like whose time should we be spending money to get back? The big bosses. Right. And the, the sort of double benefit that I didn't realize until later was that these admins who are very capable people, it turns out they actually make great accountants over time. And I don't know if you're like me in my career, probably half of the best accountants I ever knew were never accountants, didn't have any credentials, like never planned on getting into accounting. The, the most powerful example we had of this was in the pre accounting process that we rolled out for my bookkeeping practice. Had it was about a $1.5 million bookkeeping practice. And we had the same issue that everybody has where every single month you got to log into all these different places and get all these things and stuff breaks. And my professional staff is wasting time doing all these logins. Right. Well, instead of doing that and it being up to the staff person to go do that for all their clients, we centralize the pre accounting process. Every single item that was required for a month end close that all came centralized down to one place for all clients and a centralized team of pre accounting folks would go out and fetch all those details so that when my professional staff got in to do the month end close, everything was already there. We had what we needed. If there was something we didn't have, the client got notified on the 1st when we couldn't log into the bank or something like that. Saved my professional staff a tremendous amount of time because now these admins were doing the pre accounting work. They that was step one. Step two was the admins were like well just tell us where you want us to put these files. Like we see you do this weird file structure thing, like can you just train us on how to do that? And then they're like well just tell us how to name the files. And then they're like well, well just tell us how you want the files annotated. I see you like draw these boxes new this highlighting and before you knew it we had people that were like hey, can you just, can you send me on like a bookkeeping course? Cause I'm, I'm already doing like half the month end close. And these folks just, they figured it out really fast. And at a time when it was very, very Hard to hire professional book professional accountants and bookkeepers pulling in folks that were like ops first. I would say they were ops first, but they also had great people skills. So we're looking for admin folks that had very, very good people skills. Like better than most of the accountants on the team, but were like systems thinkers as well. That person gets some accounting expertise and buddy Bob is your uncle. So mistake number five was not hiring enough admins. Number six was not having a delegation path for every person on my team. You may recognize as the big boss the importance, the imperative of you having somewhere for your tasks to go, right? Like you can't do that next thing until you're able to offload some of the things that you have right now on somebody else. But the same applies for the rest of your team. If you think about your most senior team member. Imagine a client sending them 100 page PDF right now and them than wasting the next two hours chopping up that PDF and putting it all the places it needs to be and all that. That would super grind your gears, right? Because that's a largely administrative thing. Don't you want someone that they could then hand that off to? Like oftentimes we will cling to tasks ourselves because we feel bad delegating them when like your trash is somebody else's opportunity. That same delegation muscle that like none of us were born with, if you give that person someone to delegate to, they're building the delegation muscle and also the recipient of that stuff, they're getting upskilled faster than ever. That's a win for everybody. Your senior folks are focusing on the senior folk things and everybody else is getting upskilled faster because they're getting the scraps that are now delegated down to them. This is one where it absolutely requires you to lead by example. And when you're starting out, it might feel like being a snowflake where you're like, I've never asked for help before. Like I'm not, not above doing any type of work. That's not what this is. This isn't saying that you're above any of this work. This is saying that your time is super valuable and this work that you've got, it's an opportunity for somebody else to get pulled into that work so that you can do even more valuable things. That is the best thing for the business as a whole. But don't stop with you because the same logic applies to the rest of your team. And this is why Pod Structures became a really core part of my firm every Podcast had a manager, had a senior, had a staff level person, had a junior, had a mix of onshore folks and offshore folks, so that everybody had a crystal clear path where they could delegate things to so that they were operating kind of within the zone of what they were trying to grow into, rather than just getting mired in the tedium. So mistake number six, not having a delegation path for every person doing this right, just looks like having an org chart where that's really clear for everybody. Clearly spelled out mistake number seven, not hiring for your tasks. And most of us don't do this. What we hire for is growth. We hire to do more. We hire to maybe push a few of our clients over to that hire rather than hiring for. What are the things that right now only you can do? And this usually means hiring a higher level people, a higher level person than you ever have before. But all that like hiring for growth does is give you one more person to manage. And before you know it, you're just like hitting your head against the wall because you're like, geez Louise, I already had a bunch of stuff to do. Now I have a whole bunch of stuff to do and one more person to train and manage. Well, what are the things that you are doing that we haven't effectively hired for? Oftentimes it's trivial stuff like paying bills for the accounting firm, doing stupid stuff on the building. Gosh, I hated doing like physical building stuff. Such a distraction from the business. At the end of the day, your time is the most important time to hire for. But what most of us do is actually the opposite. We hire a staff person who can serve our lowest fee clients. Like we actually hire for the bottom of what our firm does, rather than hiring for the top of what our firm does. Start with the top, because there's nobody whose time is more important than yours. You doing tax work right now, you got a lot of 1040s you're trying to get out the door, buddy. Have I got the app for you. Is this episode sponsor Soroban? If you do 150 or more 1040s a year, you need an AI 1040 intake automation solution. Vastly better than Cognito Forms or the Paper Organizer or anything that a practice management system has on offer. Right now with Sorban, you just dump all the docs that your clients are giving you into Sorban. It goes through a checklist of all the things that you should have for that 1040 based on pro forma data, automatically checks off all the items that are in those Documents and gives you like an intelligent portal you can feel good about putting in front of your clients. That will save you a tremendous amount of time in organizing all the information that you're getting in and figuring out what you have and what you still need and following up with clients, automates, all of that stuff. A tremendous tax season, time savings. We are in the thick of it and this is where Sorband is going to save you a tremendous amount of time. I know tens if not 100 plus firms now that are very, very happy users. Learn more about Sorban. Check out the link down in the show notes. Can we talk about a very real problem with most offshore staffing solutions? Actually faced this one myself is the staffing groups that serve as sort of like a middleman, a middles person between you and your staff. Like you pay the middleman and then they pay part of that to the staff person. Did you know in most setups there's actually no way to get that staff person out of that agreement? If you're like, I'm actually tired of paying half the monthly cost to the middleman, like half the money I'm paying is going to this company, not going to the person. Can I just work with that person directly? Most of those contracts, there's actually no way to get that person out of that employment agreement. That is why this episode sponsor Team up exists. They're an offshore staffing solution where you just, you work directly with that person. Team up, they're basically just going to serve as the recruiter and then after that there's no ongoing fees. Half of the money's not going to somebody else. You're just working directly with a talented human being. In the Philippines, the progression honestly for most accounting firms is they'll hire, you know, three to five folks offshore, have good luck and they're like, okay, what does it look like to go to 10 or 20? And they're like, I don't want half the money I'm paying to go to this other group anymore. I just want to pay these people directly. That's what Teamup will help you do. Find you somebody great for a one time flat recruiting fee. Actually founded by a former bookkeeping owner too, if that's something you want to explore. Learn more about Teamup down in the link in the show notes. Very closely tied to this number eight, hiring as a band aid for bad strategy. Oftentimes we will hire to do things that we don't have any business doing to begin with, right? Like when I said that don't hire for the bottom of the business. Hire. Hire for the top of the business. You're like, well, then how does the bottom of the business stuff get done? The truth is, if you're not excited about it, it probably shouldn't be done still, right? Like, there's, there's these hangers on low value folks on the client list. Oftentimes where we're like, no, we, we just got to hire somebody to do that. Now the problem is not the client. They're great people, they're great clients. We just got to hire somebody to bring, bring those people along. And if you've, if you've done that before, you know, like just how much of a toll? 5 to 10 to 15 to 20 years, when the client list gets bigger and bigger and the employee roster gets bigger and bigger and those projects get less and less profitable. In many ways, the folks that refuse to hire, like, they're like, I'm only ever going to do this myself. They're running the most pure version of this thing because they have to make those trade offs. Like there are 24 hours in a day for solo people in a way that the rest of us pretend there isn't. We're like, okay, I'm at my 24 hours now I just need to hire more people. That's the solution. When oftentimes it isn't like, that's, that's not a band aid for the problem. The problem is we probably shouldn't just be serving those folks or we shouldn't be trying to do so many things like we used to. Two more to go. Mistake number nine, and this one hurts not paying for talent. Like, at the end of the day, there's a cost to every day that you don't have the right talent. And I don't like paying headhunter fees as much as anybody else. I mean, 25% of first year salary when it's a high level hire. That's a lot of cheddar, man. But we also have to consider the cost of not having that person. There's situations where it's like, no, those are not, those are actually not good clients and we just need to let the clients go. But there's other situations where it's like, there's some real opportunity here if we could just find capable people to serve that opportunity. And even if you get sticker shock at what that number is, at the end of the day, this is, this is, you got to do the maths like it is. What is the opportunity if we have these folks here if that were a piece of machinery is not the nicest thing to say about your team members. But if that were a piece of machinery that was going to cost you, I don't know, what's 25% of 200 grand, 50 grand? If you could buy a $50,000 piece of machinery, 50,000 bucks up front, that's it. And you were going to run 1, 2, 3, $400,000 of throughput through it, you would probably do that. Humans, I would argue, are better because like the relationship I can have with a human, you may not believe this relationship I can have with a human, it is deeper and more meaningful to me than the relationship I can have with a machine. And we're accountants, so we're not all that way. But as much as I don't like it, I will happily pay a recruiter if it means they're getting the right button, right chair. Mistake number 10 is not treating hiring as just another lead funnel, right? Because most of us, we have an abundance of work to do. Like there's the starting out period for sure. But most people get through that in like 18 months and they're like, okay, I think I got enough. Now it's a question of how do I get a better client or how do I, how do I hire to support the right clients? And so on the, on the client acquisition side, you may not need much of a lead funnel at all. That stuff is coming in. But then you go out to hire and you're like, man, I can't hire. And we think that it's like, well, it's the pipeline. It's, it's the universe. Like, there's nothing I can do about this. Like, poppycock. Even in a pipeline shortage, there's so many accountants out there. Like all you need to be is 10% better then the next firm down the street. And you can find folks. And this looks like building a talent acquisition pipeline. Not too dissimilar to how you would build a client acquisition pipeline. Like, I know we all hate marketing, but marketing for clients and marketing for staff, not all that different. Like, how do we go out there and talk about what we do so that we look like we got a really cool place to work that's for a specific type of person. Just like on the client side, the more we are out there and having those conversations, the more the right people will find you. And those are much, much higher level people than are sitting, hitting control, alt, refresh on there. Indeed right now that like, that is the solution to consistently not being able to find people is you need basically a lead generation machine for finding really good staff. It's not that you can't hire for most of us, it's that it doesn't hurt enough yet to try something new. And so we're serious about like, you know, I really need to go find more people. I need to find higher level people. It probably just means finding a different approach and in a perfect world, finding some other firms who have navigated the same thing and can show you the shortcuts for all the like downsides and challenges that come with employing other human beings. Still to this day, the deepest satisfaction I take from running a business not only in the accounting firm, but also in what, what we do now, which is a team of, I don't even know, 10 or 15. Now I take the greatest satisfaction from creating an awesome place for people to work and being able to see where I brought them in and where I left them. Just creating opportunity for people like that is the, that is the greatest satisfaction I take in running a business. Interestingly, on my internal team side, just as much if not more than on like the external customer side, the people that you're serving. And as hard as it is, still something I pride myself in. But we got a bunch of AI questions to get into. And next, a non accountant who just bought an accounting firm that needs help building their tech stack. Let's do it. Hey, appreciate all the podcast listeners just filling the comments last week with wow, great transitions, Jason. I don't know why out of the blue you all just started doing that, but just know how much I appreciate it. Just bought an accounting firm. I'm a new accounting firm owner and I'm not an accountant. Bless the firm does about a million a year tax and bookkeeping. They said their firm is running on Sage. Buckle up. Their firm is running on Sage 50 and Pro Series. So a non accountant just bought this firm. Sage 50 and Pro Series. They do bookkeeping, payroll, sales tax and personal and business tax. They have no website, no social media. They use Gmail. What? It's all word of mouth referrals. Million in revenue. I think I said they said they want to make it more efficient before they turn on the marketing slash advertising machine. I love this man. This is. People talk about like will there be a market for the legacy accounting firms out there? And I don't want to sound mean when I say this. What will the market be for legacy accounting firms out there? The market will probably be all the non accountants that are buying accounting firms because they look at the P and L, or they look at the balance sheet and they're like, well, that's a cool business. But the business is run on like, paper files and a bunch of legacy stuff. Good for you. Going out, buying a firm. I love non accountants buying firms. In fact, I think about doing more M and A content because I think accountant accounting firms are just such cool businesses. Sometimes I'm a little torn because I'm also the accountant that's on the inside that sees those people. And I'm like, well, you don't get it. And there's a little bit of like me pulling up the ladder behind me, right? But the reality is there's a lot of smart people out there that can also add things to the accounting profession by getting involved. Like Patrick Dichter, we've interviewed on the podcast a number of times. He was a non accountant that went out and acquired several firms. And before he was buying the firms, he's like, yeah, no, here's the plan. I'm going to buy one per year. This, this, this, this. And he's done a great job. I think he ended up acquiring three firms. But he's, he's like, yeah, no, I think I'm good for now. His exact words, I think were wanted to avoid the brain damage of buying. Buying another firm. And if you've been there, like, you just know. Anyways, this person just bought a firm. Very, very cool. And the best analogy I can make here is a guy I talked with who was a developer that bought an accounting firm. A developer who's like, I don't want to write code anymore. I said, I want that to be my main thing. And he's like, what would be an easy thing to do? I know I'll buy an accounting firm. His greatest finding, which I think is fascinating as he got into it, was he thought he'd be able to build all this custom software to automate how all the work got done. He said there was actually very little I could do from a workflow standpoint. Where the opportunity came was actually in building a sales and marketing machine. And so he used his automation, his tech talents, not on the workflow side. Because he's like, I tried and honestly couldn't really make much headway there, but had great success on the sales and marketing side. If memory serves, it was like a six or seven hundred thousand dollars a year firm. And he pretty much doubled it in like 18 months. And so this person is saying, I want to make the firm more efficient before I turn on the marketing and advertising machine. I would just say like for an outsider coming in, it's not always that straightforward. That being said, from a tech stack standpoint, absolutely there are improvements here to be made. If I can apply even more asterisks to this. I shared this on social the other day. If you saw it, I thought it was quite profound. What I thought would fix my accounting firm in order was first the tech, second the systems, third the clients. But what actually fixed my accounting firm was the the opposite. First, finding better clients, second, building better systems, and third, building better tech. So if you're coming into a new firm and this isn't an advice that this person has asked me for at all, but I'm rendering it, if you're coming into a new firm, the easiest thing that you can do is probably go through that client list and like find who are the folks you want to double down on. Who is the stuff that's been hanging around and isn't really profitable work? Probably start there. But on the tech side, the single most important tool inside of an accounting firm is your practice management system. But it is the brain, it is the operating system of everything. And if you're coming into a new firm, everybody's already scared because they don't know you got this like split family sort of thing where you're coming in even though you weren't the people that weren't the person that hired everyone that's on the bus now. And so if you just like do a rug pull on the entire tech tech stack overnight, that is going to send people spinning even more than they already are. So my, my, my advice here, you've probably heard me give it before, is what you're looking for is actually the quickest wins, not what is the perfect tech stack that I can swap in overnight because it's not realistic with a bunch of human, human beings doing the work. So instead we want to find the quickest wins to be able to build to that. The quickest win on the tax side in the US it is a 1040 intake automation tool. Right now the best ones are Stanford Tax, Sorban and Trust. And the easiest one on the bookkeeping side as a month end close automation tool. And at this point you probably just have to start with double for both of these tools, 1040 automation and month end close Automation. These tools are sold per client and the beauty here is that you can wade into it client by client and they're so good, like it's such a step improvement that people get into it and they're like, dang, yeah. No, this is actually way, way better. And then they want to roll it out across the rest of their clients. And when you're first starting out in this sort of change management struggle, that's what you want, is the easiest stuff first, as opposed to like practice management, which is time and billing and all of your work management and everybody has to hit the buttons the right way otherwise the whole system doesn't work right. Don't start there, start with the easy wins. So look at 1040 intake automation. Look at month end close automation. That if that wasn't clear, the tool I'm talking about, there is double for month end close automation. There are practice management systems that do a bit of both, like the practice management and the month end close automation, but with where you're at right now, like where you just need to get some more piecemeal wins. I wouldn't start with a big old practice management change. Once you kind of got your sea legs and you're maybe a year into using these tools, then look at the practice management system like that's probably what I would do next. If you've got workflow automation tools and you've got practice management, like you're 90% of the way there when you do the PM change, make sure you pull in an implementer as well. For all of the leading platforms, there are people out there who have helped tens or hundreds of other accounting firms do this rollout and, and I know it's not going to be at a time when you're like excited to spend more money, but they can shave so much time off the learning process. And when you're doing this rollout with a team, they can be a really useful foil for you to not always have to be the bad guy. If you're trying to make this big platform shift with your team, there's going to be fighting and there's going to be folks that don't like this or that. It can be really helpful to distribute a little bit of that on the implementer who can come in and say like, gang, listen, I've done this before, here's what will work, here's what will not work. We have to do it this way. Sometimes you just want somebody else to be the one to say that, right? And so that's part of the value of pulling in an implementer as well. So at that point, like you've done 90% of it, at that point it's just dialing in, you know, RightWorks. If you're a QuickBooks Online user you really need RightWorks. Ramp is a tremendous resource for just like bookkeeping workflows, getting more of your clients using Ramp. Biggest upside there is at the time of spend at the time of the transaction. Ramp helps you capture the receipt like just then and there so that when you go and then do the month and close, everything's ready. Like Ramp, you can literally set it up to turn people like turn users cards off. So like an employee in the business, you give them a card with a budget, you can set that card to turn off. They haven't submitted the receipts for past transactions. On the tax side, what are the other like fun toppings on the tax side? Probably Blue Jay. For AI tax research, they said they're using Pro Series for tax software, which is kind of in this like no man's land because Intuit has so many tax products and the only one they're pushing is ProConnect because it's web based. And so Intuit doesn't even really like Pro Series. But changing your tax software is like, requires so much muscle memory change. That's a ways down the list of priorities. Definitely not before changing practice management systems. Other thing I would say here, actually for somebody coming in to own a firm for the first time, there's a lot of other people that are doing this right now. Meeting them, finding out where they hang out, that's super valuable because you're going to have a whole bunch of ignorant questions and blind spots and all that. And to try to do that with other accountants, like it's better than nothing. But to find some other folks who are non accountants who are doing the same thing, like, that's really, really valuable. Building some of those connections will save you a lot of sleepless nights. Speaking of this, there's I think there's been two times in this current AI era where it's like it was hard to sleep and right now I'm having my third. The first was when ChatGPT first came out and like I was struggling to even find the boundaries of what it couldn't do. The second was when reasoning models first came out and research got so good overnight, like the models just got way smarter. Third is what's going on right now. Claude. Cowork open. Claude. I've had the opportunity the last three or four days to just go really deep on just using that stuff and doing a ton of testing and holy smokes, are we at like an entirely different level of what is now possible with accounting AI? People's heads are spinning, honestly, for good reason. Because right now what's beginning to happen is what like the AI bros have been saying was going to happen. When I say AI bros, I'm not talking about Twitter AI bros talking about like the Google folks, the open AI folks. I can see it right now in a way that I've never been able to see it before. And we got a bunch of questions on the subject to get through. Let's do it. Mailbag. Mailbag. This segment is called Mailbag. I record this as this week we'll be shooting a video on Claude Cowork and on openclaw I we're shooting them this week. They'll go out in the next couple few weeks because somebody needs to make a video showing where all this stuff's at. What's. What are the okay ways to use it and the not okay ways to use it. Accountants are busy, you don't have time and you shouldn't be spending the time right now to put open claw on your machine or anything like that. But you need to see just how good and just how powerful this stuff is now. Holy moly. It's a I it's a big step from chat GPT and having a conversation back and forth like we it's really come a long ways. It is. It's two things without going too deep. The first is just how good Claude's models are now. Anthropic Scott. You know, the last couple opus models are so much smarter than any other large language model we've ever had. And so that's just unlocking some cool stuff. But the second is, let's say the harnesses themselves have gotten good. So the harnesses being like there's this sort of meta layer that when I give it a request is handing out different jobs to different sub agents and sort of orchestrating a more intelligent approach to getting a job done. It's no longer as simple as just talking back and forth with a large language model. And so those harnesses, like the orchestration of getting something done has gotten vastly better. Combination of those two things, better models, better harnesses. The ceiling is so much higher now for what you or I could just walk up to an AI tool to do. The problem is open claw right now is not something anybody has any business putting on their work machine. And Claude Cowork, the other really powerful tool right now still in Research Preview and doesn't come with a lot of the compliance stuff that I would want to see for a tool that I will feel comfortable putting client sensitive information into now that could change tomorrow. Which unlocks a huge amount of use cases. And it's why it kind of puts me in a tough spot from a content standpoint. And where I've kind of settled is we're going to show a lot of this stuff off and be really clear where the boundaries are. And that most of this stuff you honestly shouldn't be touching. But I need you to see just how far it is so that when it is live, when it is good enough, when I can use it responsibly, you can see like just how many opportunities it opens up for us. That's a pretty big tease. I know. I'm very excited for those couple videos to come out. I'm pretty proud of like the ChatGPT YouTube videos we've done in the past. They're tr. They're humongous undertakings, like genuinely, probably three weeks end to end of work of the research that goes into building the examples to like the full editing process, which is a huge task. And it's, I mean it's not me, it's. It's five or six people that will have their hands in a video before it goes out the door. It's kind of wild how you go from making like 90 second loom videos for accountants to some of these videos. Now, like we will have tens of thousands of dollars invested just into the making of the video. But I want you to see like just how good this new stuff has gotten. Even though right now I wouldn't recommend putting any client stuff in it. It's that time of year, the time of year where your hosted tax software starts grinding to a just to a halt because everybody's trying to use their computers at the same time. This is because when you host your apps with most providers, like your tax software provider, the servers you're accessing, they're shared servers. Everybody and their mom are logging into those things, trying to get some stuff done at the same time. The solution is today's sponsor, Verito Man. Verito is gonna give you your own private server where all your tax stuff, even any hosting needs, QuickBooks, desktop, whatever you got, it's hosted on a server that only you're accessing. So it doesn't matter how many people are hammering that server, the only people hammering it are you. It's easier than it sounds to get started. I can actually transition you in just a weekend. And especially as we're heading into busy season, gang, this is the time to do it. Bardo. Vastly cheaper than the hosting you get from your tax software provider and faster like kind of a no brainer. Now Bardo can also do your whisp, the little security thing you got to do if you do tax work in the us they can also be your entire IT department. In fact, the cost of hosting your apps and them being your IT department is less than Thomson Reuters Virtual Office Now a lot of people are very happy Verito users. In fact, the average response time from Barado Support if you come to them with an issue is now less than 60 seconds, 24 hours a day, seven days a week. Like I said, now's the time to do it before we get too far into busy season. To learn more about Verdo, check out the link down in the show notes. Hey, real quick. Some very cool new updates were just shipped by one of the leading practice management systems in my app recommendations this episode Sponsor Financial Sense. So if you do bookkeeping work or client accounting work, this is worth checking out. They're building some month end click Close automation features into the practice management system. So stuff like handling uncategorized transactions, flagging transactions without payees, without class, without location expenses and bills without attachments, inconsistent expense classification, large transactions. It will now flag all of these items inside of a month end Close in Financial Sense because it's syncing the actual Ledger data from QuickBooks Online to Auto flag these items. Really cool. Some stuff that you've only seen to date really in like dedicated workflow tools. Financial Sense now, which is a practice management system, is pulling some of the best of that workflow into the platform so you can do it all in one place. Pretty cool stuff. And I got a discount code for you. If you sign up for their team or scale annual plans with two or more users, you can use code Jason10 to get 10% off. That is Financial Sense doing some cool bookkeeping automation stuff. To learn more about them, check out the link down in the show notes. I want to read you this is a tweet from Peter Gurness. Here it is. It's pretty short. I'm the VP of Marketing at Microsoft, responsible for Copilot growth. Copilot is Microsoft's AI thing. We just spent $8 million on a super bowl ad for Excel. You might have seen this. There was a Copilot super bowl ad. He says I need you to understand what happened. 130 million people were watching football. The biggest game of the year. Beer commercials, car commercials, movie trailers, Beyonce and us. With a 30 second spot about a spreadsheet, an AI powered spreadsheet. We showed two NFL recruiters using Copilot and Excel to quote, chart top linebacker prospects with 40 yard times under 4.6 seconds. Because nothing says super bowl like pivot tables. The ad costs $8 million almost to the end here. That's $266,667 per second to tell America that Excel can make a chart. Now, Excel has always been able to make a chart since 1985, but now I makes the chart faster, sometimes correctly. We need 130 million people to know this urgently. It's almost done. Don't worry. During the second quarter of a football game. Here's why. We have 450 million Microsoft. 365 paid seats. 450 million. That's more than the population of the United states. Of those 450 million, do you know how many pay for copilot? 15 million. 3.3%. 96.7% of our customers looked at Copilot and said, no, not maybe later, not. Let me think about it. Just no, we're almost to the end. So we did what any rational company would do. We bought a Super bowl ad for a spreadsheet. Satya Nadella called Copilot a true daily habit. That's a real quote from our CEO. On an earth on an earnings call. A true daily habit. Like brushing your teeth or checking your phone or ignoring Copilot suggestions in the bottom right of your Word document. Our daily active users went up tenfold year over year. That sounds impressive until you learn what tenfold of almost nothing is. Almost nothing. Slightly louder. I'm almost to the end. In July 2025, 18.8% of subscribers said Copilot was their primary AI tool. Six months later, in January 2026, that number was 11.5%. It went down in six months. While we were spending 37.5 billion per quarter on AI, the tool got worse at being people's first choice while we spent a country's GDP on it. Internal reports say weak ROI. Internal reports say resistance to forced adoption. Internal reports say what every IT director already knows. People open Copilot once, watch it hallucinate a formula, and go back to typing Vlookup like God intended. But I can't put that in a Super bowl ad. So instead of linebacker prospects, 40 yard dash times, chart the top three with leadership skills. Leadership skills in Excel. Okay, so I. I think this was parody, but there's a lot of very true points in here. In fact, you might have seen Microsoft's head of AI very explicitly called out accountants and lawyers. He said in the next. I think it was the next 12 months. This is Microsoft's head of AI said, In the next 12 months, AI will be doing almost all of what accountants and lawyers do. I think it was 12 months. I know it was like 18 or less. It was not a very long time horizon. And that's just funny coming from the head of Microsoft AI. Microsoft, who famously has yet to ship an AI product that is even worth using. You want to talk about like an own goal? It's incredible. If you go back to ChatGPT, ChatGPT comes out. It's a big deal. Two months later, Microsoft buys half of OpenAI. And at that time, I'm like, dang, Windows is going to slap in like six months. This is exciting. And look where we are now. Holy shamo. How is Microsoft fumbled this so badly? Arguably not as badly as Apple. But it is interesting that Microsoft's head of AI is like, yeah, in 12 months, accounts are going to be out of work and we're like, brother, we won't even touch your stupid little Excel assistant. Like, I'm better off taking my spreadsheet to an entirely different app and talking with that AI assistant about the spreadsheet than the assistant that you have built directly into the spreadsheet. And all this boils down to. It's very hard to know what to believe. And so it, like, it's why I just like went on my own little ayahuasca journey of just spending like $650. I think it was on cloud model usage just over the last three or four days. And there's a. There's a stream outside my house that four days ago was just full of water. And as I look out there now, it's just a. It's barely babbling. That's how much AI I used in the last three or four days. I came into that, like, not expecting a ton, like, thinking a lot of it was hype. I'm. If you can't tell, I'm pretty amped right now. There are just some really cool things that are possible with these tools, from automating technical review to the doing of the work to like, building your own little skills for it to do whatever you needed to do. And then the models, in a way that they never have before, actually follow those directions really well. It's very exciting. Let's get into the questions though. Have you or your team tested any of the accounting AI ledgers yet? I know you talked about digits, but any opportunity for count or puzzle, really interested, interested to see all these in action and See the possibility of moving over when we put out that video on the main channel maybe a month ago of like, what are the best AI ledgers to use right now? The winners were digits and kick. And it's still just for cash basis stuff. Like not big, ugly, complex class tracking, nothing like that. Yet it is for cash basis accounting. And some folks went out and tried digits and kick and they came back and they're like, oh, it doesn't do everything. And I'm like, yeah, no, that's why I said cash basis accounting. But there's a reason that those two are at the top of the list and there weren't others. There's a lot of platforms still that are just, they stink. Like, they're just, they're so hard to use. Puzzle, it's, I mean you can see some of my app recommendations, they are up there. They're just not as good as digits and kick right now in my opinion. They started a little bit more mid marketing, but then came more small business and now it feels like they're in a little bit of this limbo. They were built initially for startups and that's the one type of client where like if there's an argument to be puzzle first it is in doing your startup work. And so if I'm specializing in startups, I probably would go to puzzle. But right now, because most small accounting firms, the majority of your work is cash basis, high volume stuff, Puzzle is too much for that while simultaneously not being great for more complex and project work and all that. And so I've got digits and kick at the top of that list right now. Count, man. I just like, I just haven't had a great experience and don't know anybody that's excited about count yet. I want like, I'm their biggest cheerleader. Like I want them all to work. Like I put together like dinners for founders of these apps and like I bring all this stuff together and try to support these folks. But I will also be equally honest about where they're at and what I recommend, what I don't recommend right now. So yes, I hope AI ledgers get better. And to be clear, they're getting there. Which AI is best for making two documents work together? Especially if a PDF is involved for making two documents work together. I've been on chat GPT teams and have been really frustrated with its results, trying to read PDFs and work on another document. One tricky thing about PDFs and AI is it's not always clear what in a PDF and AI can see and not see. And so first thing to look out for is, is there actually text text in a PDF document? Because oftentimes what looks like text is actually an image in a PDF. And that can be a hard thing for a, a AI model to visually understand. Second is like tables, things that require an understanding of the layout for it to make sense of it. So, for example, the tax table, the way that the IRS presents a tax table in an IRS publication saying, like, if your income's in this bracket, it's taxed in this rate, and the next X of your income is taxed at this rate. At least early days, large language models could not make sense of that sort of layout. They're actually pretty good at this now. But it's, it is one thing that has made AI's understanding of PDFs difficult. And if you think about a tax return which doesn't follow any sort of logical structure, like, there's these lines and it's like, here's this box and then like 25 periods and then some, like, bars and a box that might or might not have a number in it. And oftentimes there's like a box next to the box and like several boxes there. Like, language models, by and large cannot make sense of tax documents just because they're like what they see. If they're just working with the text of the PDF follows no logical flow. Like it, like the way that if you, if you copy and paste the text out of a tax return PDF, when you read back the order of the text, very hard to make sense of that. Alternatively, you can give them an image, like they can give them an image of the tax form. It's just the processing requirements of seeing an image are vastly greater than working with text. So If I have 100 pages, giving it 100 images is not realistic. But 100 pages of text, that's like no big deal for a large language model. So PDFs like, they're just, they're not structured from a data standpoint in a very helpful way for an AI model. So it's hard to answer questions like this, to say exactly, and a certain situation why something didn't work. Well, I've had situations where I will literally just copy and paste all the text out of a PDF into a large language model and it will work better. Situations where I've had to do the opposite, where, like, if it's one concise thing, you know, a table, for example, I'll just take a screenshot of it and give the screenshot to the AI tool and oftentimes that will work better. I will say this has gotten better with really good models from Cloud Opus 4.5 and 4.6. And so this is inside of the tool called Claude by the company called Anthropic. Their models 4.5 and 4.6 are a really big step up on all of this stuff from anything else that we've had in the past. OpenAI kind of scrambled to get out GPT5.2 in response, but it still falls short of the cloud models. It's hard to recommend people to switch because, like, there's been other times in the past where one company will kind of come ahead a bit, but then it's like a month later. The answer is totally different. And you don't want to hop around between these different apps every single month. Right. So if you're selling, if you're actually, the last question here is like, is it time to ditch chat GPT? And so let's, let's go deeper on that at the end. Next question. Bot Keeper. Could barkeeper have stayed alive longer with this current era AI stuffer? Have we talked about Bookkeeper yet? So Bookkeeper Win under it was a company that's been around the accounting space for over 10 years, raised well over $100 million, maybe met a few hundred million. They've raised a lot of money. And the promise with botkeeper has always been we will automate your bookkeeping work in the platforms that you use, like QuickBooks, stuff like that. And so they had built a bunch of automation on top of common accounting tools. They also had a lot of offshore talent that were augmenting that as well. And there's been a lot of promises. They had a good marketing budget, so they were out there, they were visible. They were even putting on their own AI conferences there for a while. But it's a great example of where just having peers you can, you can like talk to about the legitimacy of platforms is really viable because I know tons of people that got sucked into the Bot Keeper thing and they were like, yeah, no, in the end, like then wasn't really worth the money for us. Wasn't. Wasn't really what it was promised. But if you think about Bot Keeper in terms of the whole ecosystem, you've got this player that's like, we're going to be the most automated version of this possible. Even pulling in offshore folks to kind of fill in the gaps. This company went under two weeks ago. They have all like the AI models. Like they have all of this stuff and couldn't make it work. And so for the people who today would tell you, like, oh, with the AI that's there right now, we can go out and automate all of accounting. Like, if that were really possible, these companies that had raised hundreds of millions of dollars, they would have gotten there and they just didn't. Now, that answer might change in the next three months, in the next six months. But, like, if, if we were really there, like, you don't think Bot Keeper was like, doggy paddling harder than anybody else to find? Like, how do we use AI to make this work that we're doing more profitable and be able to, you know, serve more, more firms with, with a smaller team? Like, companies are out there trying to do this and you don't want to dance on anybody's grave. But, like, if you're going to take anything from this one, it is like, pick tech that your peers will validate. And Bot Keeper was never that. It was really exciting and I loved that they were there and they were working on it and there was a time when it was like, yeah, you could see a world where somebody could actually make this work with the tech that's there. So first lesson, like, like, don't buy a thing unless you know somebody else that really, really likes it. But second lesson was like, we're not just plugging some AI stuff together away from fully automating things. Like, maybe that changes soon. But if that were the case today, you wouldn't see stuff like this. So bummer that Bot Keeper went under. Always a little bit for me. Not like, scammy is too strong of a word, but was always a company that was kind of in that gray area of like, nah, this doesn't super feel real. And like, in like, the, the most negative interpretation of this is like, maybe a bad example of how, like, VC excitement and funding can, like, prop up something that's like, it's not real yet. But I mean, to be fair, there's a lot of examples of other companies where it's like, they go out and they raise the initial round, raise some money, and they're like, no, it's not there yet. And the point of the money is like, we're trying to get there, and by the time they need to, they do get there and that takes off and all that. So it was a bet on a future that never quite panned out. Next question. I'm starting my accounting business and would like some courses on how to use AI to manage my work and grow my business faster. You know, asking how to use AI is kind of like asking how to use a computer. There's, there's a lot to it. In fact, using AI is kind of just an extension of using a computer. Definitely a new skill set for sure. That is not a quick thing to pick up. But I could tell you, I can tell you as someone who creates content, it is tremendously difficult to make anything related to AI because it changes so fast and so like, I mean it's, it's, it is a big project to put together a useful resource on how to use something like the logistics that just going, go into having something polished and researched and all that. It's just a lot and then a week later it'll, it'll be out of date. So going as far as like building a course, like something more involved, end to end. While there are kind of some fundamentals that you can build a course around, that is a tricky thing to be able to do in a evergreen way because that stuff just expires so fast. So I don't have a great answer for you, unfortunately. I have talked before on the show about Chad Davis's community. It's called Automation Town. It's a, it's a group of basically like AI champions from other firms, even some, some agency owners. So it's kind of the, the folks who have an abundance of time and, and energy to be able to work on like the coolest applications, building more sort of custom stuff inside of their firm. So if you really want to go to the extreme edge, check out Automation Town. Otherwise hang out on the podcast like and subscribe. We've got ChatGPT videos on my YouTube channel. And find other firms who are working on this too. Like see what they're learning. This is someone who's just starting their accounting business. So find some other folks that run small firms that are leaning into AI. Like that really is the fastest way to learn. Hey, real quick, I want to invite you to my super excellent weekly newsletter. My newsletter was like the stick start of my thought leadership journey. It's where I try to collect what I think are the most important things for you to see every single week, sent to you every Wednesday morning. Make sure you don't miss any of the good stuff. Right? Increasingly we're also building into that some like of the best evergreen thought leadership I've ever done. Kind of make that like the most action packed, highest quality email you get to your inbox each week. Because listen, we're all busy, all right? You're not gonna see every single little thing that I put out in a six. There's just too much. If you're not on my newsletter yet, check that out jasononfirms.com or I'll put a link to it down in the show notes. Hey, quick word from one of today's sponsors. Me. Me. Yes, I run a community Realize that is for accounting firm owners and Realize is kind of the secret behind how I've done all of my content and been able to like financially turn thought leadership into a business. About 100 firm owners in there realize is a systematic way to build connections with with other peer firm owners. It's not a framework for how to run a firm. It's a way to introduce you to other super cool firm owners. Right now we're putting together small groups so you can have a group of like 5, 6, 7 other firm owners just like you that you meet with every single month run very similar firms. Having this for me totally transformed firm running my North Star. Like my goal for you coming into this group is to make a new best friend, somebody that runs a firm because that was what I needed in my journey. The whole community is built around that. It's called Realize. Link down in the show notes if you want to learn more. Any experience with Perplexity and Comments so Perplexity was kind of the original really really good AI online search even before ChatGPT could do that well. And Perplexity still does like online research very well. I wouldn't say it does it enough better than Claude and Chad GPT to make me use Perplexity. I really don't use Perplexity that much, but they kind of made a splash what was it maybe six months ago when they launched Comet, which is like a sort of AI agentic browser. And so the whole idea is like in the future browsers will be like you just give it a thing to go do. Go find me a place to rent a vacation rental in Austin for this date to that date where I can fit six of my bros and it's got a movie theater. The thinking is like the future of Internet browsers are more like that where you give it the job to do rather than you doing the job yourself. So comment the browser was kind of an example of this where you kind of give it the task to do and it was okay. It just wasn't that amazing. Like it wasn't really useful enough to me to get excited about. Since then ChatGPT OpenAI has their own browser called Atlas. Since then Google launched these sort of Gemini powered tabs where you can open up a tab for it to go and try to do a task for you. The reason that none of these things have like caught on like wildfire yet is they're pretty janky and slow and not super high performing. What is kind of changing now is in a few cases you have versions of this that actually do work really good. And like the closest to today ready version of this that we have right now is probably Claude Cowork. It's still in Research Preview and Anthropic says to not use it for regulated workflows in the in the US Especially if you do tax work that's regulated workflows. And it doesn't come with some of the security and privacy stuff that you get with. What is it called? I think it's called Cloud for teams, which is just their normal business plan. So I don't recommend using Claude Cowork for client sensitive stuff yet. But Cloud Cowork can work a Chrome browser and in addition to having local file access, like access to the files on your computer and the other things that Claude Cowork does. Well, to me it feels kind of like the first AI agent that I could actually trust to semi autonomously do some work for me across several different apps and systems, which is really exciting. This question, this is an interesting angle on sort of the same thing. They said they're leading an acquisition platform focused on buying and integrating accounting and tax firms within a specific niche. They're trying to decide how far they should put push tech unification in a multi firm rollup, whether it makes sense to build their own sort of orchestration layer over the top of legacy systems, or just standardize the tech stack across all the firms they're rolling up. It's interesting the roll up folks of whom there are many these days, right? There's all sorts of different accounting platforms and folks trying to figure that out. And the fact that they haven't really yet doesn't necessarily mean that they never will. But a lot of them will say things like, well, once AI gets in here, it'll be way easier to roll these up. Because right now the biggest blocker is the human change issue. Like how do I get a hundred humans to start doing this thing in a totally different way? It doesn't happen overnight. But the hypothesis is, well, when AI is doing that stuff, it'll be much easier to like, you know, do all that in a more standardized way. I would argue that the whole model has bigger problems when AI is doing all that stuff just because it cheapens the work, you still need the verification step and all of that. But if anybody can just AI their way through a project, like you've got some more fundamental potential issues with the business model. I think historically I have been against firms building their own software layer. I've seen loads of both small firms and big roll up groups try it and it's a black hole of money that usually doesn't get you all that much further than what you can get with the right off the shelf tech stack. I mean, if you want proof of this, look no further than like the Big four and a lot of regionals who will earmark tens of millions of dollars for AI investment and they just pour money down the drain on their own tooling and it ends up being like marginally different than what you can buy off the shelf. And honestly, everybody inside these firms seems to kind of know that and they make these investments for the press release of it all and to look impressive. But in firms both big and small, the better answer is almost always off the shelf tech. And I get that it's not perfect, but it's really hard to be a good like software development company and accounting firm especially. And another variable here is how quickly is tech changing in an era where tech stacks have kind of stagnated and tech isn't changing all that much, then building something to bring those things together that are kind of stale and they are what they are, not that big of a thing. But right now, when the reality is tech is changing so fast, you have a lot of really cool automation happening around. Tax workflows for the first time around month end close workflows, all that stuff is evolving so fast that even knowing what to build over the top of it is kind of impossible. Because six months, 18 months down the road, everything's probably going to look so different, even if it's not new AI tools that you have in your hands. The crazy thing right now is all the people at these software companies are using AI to write code in a way that's never before been possible. And so what we will see from software companies over the next six to 18 months is like this unbelievable level of productive output that they've never been able to achieve before because AI has gotten so good at writing code and so inside of your accounting firm to be able to guess what's the right thing to build, to like weave all this together. Finally, like the idea of it makes a ton of sense in a boardroom. You can totally understand how it's like, oh yeah, oh, we can see that we're spending all this money on tech here and there and we just make a few hires here. We could totally build something to bring all this together. Makes sense on paper. Reality is I've just never seen it work in practice, but I've seen a lot of money spent on it. So I probably wouldn't steer you towards building your own stuff. But on the question of how aggressively do you push firms into tech standardization when you are acquiring firms, I've seen extreme versions of this that were too much, where everybody had to change all at the same time and it was an absolute dumpster fire. And the resources that you've set aside, that is people that will help like navigate these firms through these changes. You spread those people too thin and you can't do this the right way. And so my advice here for rollups and is, is actually this and it may be a little counterintuitive. It is to start with that standardization in the firms that are going to be easiest to standardize. So order them, like, rank them on like, well, these ones are going to be brutal. These ones are going to be easiest. Start with the easiest stuff and then work your way down. Because with each new firm you will be better equipped to help the next firm and you actually know what you're getting into rather than doing it all at once. And you don't really know what you're getting into. Start with the easy ones. Every single time you do it, your team will get better at it and increase the likelihood that it will go well when they then are working with the more difficult firms. But if you just bite off a big old chunk of it and try to do it all at once, I've seen too many examples of that kind of being a mess. Okay, last question and it is honestly a really important question right now. Should you replace your Chat GPT and GROK paid subscriptions with cloud cowork? It is a good question. Right now I don't, I don't have a paid GROK subscription. Grok, like the tools, the models for Grok have been very good, but the tool set around it have never been good enough to me to be worthwhile in an accounting firm. Like working with PDFs and like kind of custom GPTs and all those things. Projects were all really, really valuable in a team based setting. And GROK just hasn't delivered those features. Crazy. The quality of models that they've delivered. But it's, those features are very important. So it's really a question of should you ditch Chat GPT to go to Claude and kind of a different question. Should you use Cloud Cowork right now? You shouldn't use Cloud Cowork with client sensitive stuff. You just shouldn't, unfortunately. And that's not hard to do. Like when you use Cloud Cowork, the only local resources it's going to use are the resources you've given it access to, which is usually a folder, a set of files, something like that. And so if you're doing that with non client stuff, awesome, go wild. Maybe the bigger question here is simply, should I ditch my ChatGPT subscription to go to Claude? Because Claude has basically for the most part everything Chat GPT has just in the browser as an AI assistant. Honestly, right now I would probably say no. If I was able to use Claude Cowork on client sensitive stuff, it would be a resounding yes. But because Claude Cowork is still in Research Preview and that's the thing that I'm most excited about. While Claude is a bit better than Chat GPT today, I wouldn't say it's so much better that I should burn down all the history that I've got with Chat GPT and make the jump. Now ChatGPT has something kind of like a Claude cowork called Codex. Originally it was to be like Claude code, but a recent update to Chachi BTS Codex has made it more normie friendly where like you don't really need to be a programmer to use Codex. It'll be interesting to see how Codex evolves. Right now it's not as good as Cloud coworkers, but the other big headline as I'm recording this is this week the guy behind Open Cloud, which is kind of this huge thing that will we did another episode on it, but it's basically an agent that will take over your entire machine. Like I can literally just message with it from away from my computer and it can navigate everything on my computer and use my password manager, all that stuff. Like pretty spooky. Obviously some big security concerns there, but the guy that built this just went to work for OpenAI. So my concern is as soon as I tell you to burn the ships on Chat GBT and go to Claude is then OpenAI ships a really cool version of, you know, Cloud Cowork, this, this agent that will work on your machine. And then I'm like, oh no, let's all go back to Chat GPT now. So I don't think I'm quite far enough to say you should ditch Chat GPT. That would feel a little shortsighted. But I cannot wait to show you the cool stuff you can do with Claude Coworkers. Super, super powerful. If you've been playing with any of this stuff yet, drop a comment, man, like tell me what you're doing, what you're seeing. I can tell you a huge breakthrough for me in testing all these AI tools. This is going to sound really weird is I'm using them to create test files like that is like actual client data. It's not actual client data, but I'm using it now to first generate a set of transactions for a company. From 100 transactions to 10,000 transactions. I can give it an industry. It will pick a geographic location for where that business is and all the transactions will map around it geographically if it should for that type of business. It has like monthly cycles as you would expect from a company like that. Like these fake transactions are really, really good. So I can have the drop of a hat, generate a fake set of transactions that look just like the real merchant descriptions. In fact they are real merchant descriptions. And then I've got a separate service or skill actually that will turn those transactions into PDF bank statements. I actually created there's like three different style formats for bank statements that I had it steal from like real bank statements. On the tax side, I have a skill where it can create 16 different types of tax documents that are like the exact IRS tax documents. Even like 60 page brokerage statements that roll up to the 2 page summary. Talking like 1099 B's. I've got another skill that will create basically a whole tax file where it imagines this client and like is it a taxpayer, taxpayer spouse, do they have kids? What are their ages? Here are all the files that you would have for a simple case in a complex case and you can generate like an 80 page pile of PDFs that is like a very realistic client scenario. Even with like printed Outlook emails of like additional details from the client. I have AI generating all this stuff. I told you I burnt a lot of money on AI tokens the last few days. But now that I have these examples, I can then build stuff that tries to complete the project as if it was actual client work. And this was the unlock for me because I've kind of been paralyzed from like we can't use it for actual client stuff. So what am I going to demo in it? And the unlock was like I worked out how it could actually create really true to life stuff that I can then use as if it were actual client files to see if it can actually get the work done. And what we've worked out so far has been super exciting. Now for me, when I see this stuff, I'm still excited. There's a lot of folks, when they see this stuff, they are scared. But as far as I can work out, there is still such a tremendous demand for what we do. Like so much shortage and so many dumb things that I wish we didn't have to do. I don't want to key another tax return into tax software. Like, that's not us adding value to the work output. It really isn't. That's administrative on just a crazy scale because there's tens of thousands of potential inputs. Right. And so I just see such a gap still from the actual things that we have to do every day to like what we could be doing that I, I feel a long ways from scared and like there's just so much entering data from one thing to another thing in our profession still that I'm like, I think we got a long ways to go of automating things before we're not going to be like, yeah, no, actually that's great that we don't have to enter bills anymore because we kind of hated that and we have better things to do. But we'll see. It's going to be a wild next 12 months. 2026 will be the year of the AI agent. Back in 2024, everybody was excited about it. And in my 25 predictions at the beginning of the year, I'm like, 25 is not the year of the AI agent. It just isn't. But 26 probably is. I think that's checking out because man, is 2026 off to an interesting start of the year. And you know what? You and I are going to be here geeking about it, sharing good ideas, getting smarter together every week, gang. Thanks for being here. Appreciate you. Any questions, I thoughts, anything like that? Just thoughts around how to run a firm. We went way deeper on AI in this one than we normally do. Pop that stuff down in the comments. I'll catch you in the next one.