049 What We’ll Do After AI Automates Accounting

Okay, hear me out. What if AI did automate accounting? Little thought, little thought experiment. Because what we're afraid of is the unknown. I don't know. We're not. A lot of us aren't afraid. But if there's anything that is spooky about AI, it is the unknown of what work looks like on the other side of some sort of disruptive change. Right? So, hey, fun little thought experiment today. Let's just think down that path. Like, what if it just could just snap, do all these things? How would our work change? Like, what would be the things we would be doing then? Because I think if we explore this timeline, it gets less spooky, like the more we flesh it out. So let's do it. Today we're doing the darkest timeline of what if AI takes over our work. Okay. Two, like two versions of this discussion that I want to have because I actually think this is really interesting. If we can paint a picture of what a displaced version of the profession looks like. I actually think it's not that scary and it's just something interesting to think about and be mindful of in the event that it does happen. Like, because then we're ultimately, you know, more prepared for whatever that next version of accounting looks like. Just like when, when personal computers became a thing. Like, how did you see into that, into what accounting would look like on the other side of something like that? Or QuickBooks or the first Turbo tax, like kind of these big profession shifting things. And man, I'm still reeling from the whole GPT4 announcement. I have PTSD from them ending that announcement by saying we're going to prepare somebody's taxes. Not because it was a super impressive demo, like it was an oppressive demo, but not, not that it was like a super relevant thing. Like, okay, now people can go and pay their, prepare their taxes with GPT4. But because of like how that really opened my eyes to the era that we now live in, the fact that we're getting these big AI model announcements that just like snap overnight, change what it's capable of. And all of this stuff is being developed in a much, I don't know, like, it seems like all this is happening much faster than any technological change that we've had in the past. And ultimately there's nothing to stop a company from coming out and saying, hey, we made this new thing and boom, like overnight it has this huge impact on the way that we work. So like, obviously preparing a tax return, that's a real big example of that. But also stuff like you know, accounting, like bookkeeping agents. We've talked a good amount about AI agents and how they can navigate software programs and stuff like that. I mean, a thing that can, like an agent that can go into QuickBooks and reconcile, you know, do a QuickBooks reconciliation, like that's close. Like that's like a next 12 months thing probably. And so it's interesting to think if that, say hypothetically that thing did drop, this is like a worst case scenario, right? Let's say it actually can do that work in a meaningful way. What does work look like on the other side of that? And I say worst case scenario in reality, it's even much more granular than that because, man, there's a lot of stuff right now that we would happily not do if we didn't have to do it right? Like the level of heavy lifting that goes into the work that we do. There's a lot of aspects of that that we would really happily give up. So this, the spookiness, like it really comes down to like what is the volume of the stuff that it takes, that it takes off of our plate. And that's what makes this a fun thought experiment to me. So today let's explore specifically what it would look like for AI to come in and take a bunch of tax work off our hands. I'm a US guy, so it's really going to be through the lens of that. And then what running a, you know, tax practice would look like after AI coming in and doing all this stuff. If you haven't seen it yet, the video that I did on the GPT4 tax announcement, it's probably my favorite video that I've ever done. I'll put a link to that in the show notes and that's kind of helpful context around like what something like this could look like. Like how somebody could genuinely steal some aspects of our lunch overnight with a very public release of this thing that at worst takes it off our plate. And again, maybe that's not a bad thing because we don't want to do it. But I'll link that video in the show notes because that's helpful context, I think. And another thing specifically around tax here. So I saw a poll recently and so there's AI talk I'm giving to some tax folks and they pulled the tax folks ahead of time and over 80% of them didn't think AI would, would meaningfully impact the way that they did their work, at least at the time that they were pulled. So over, over 80% didn't think it would meaningfully impact how they do their tax work. And that seems really high to me. I will say the more I have learned about AI, the more I do think it will be disruptive. So part of, like when, you know, when we're. And I'm doing a bunch of AI talks at conferences and all that and like the relevance. This stuff's really only relevant to people that see it actually impacting the way that we work. And a lot of people aren't, aren't there yet. But boy, look around at different industries and how there's already AI blamed layoffs happening all over the place. Whether that's just having to put. Wanting to put your finger on something to justify some layoffs or whether it's truly like, oh no, we're now just going to use ChatGPT for this rather than this little team of people that we had doing this thing. It's happening, you know, in very nuanced, like very specific places. But OpenAI put out a post in the last week around governance and, you know, the OpenAI PR and what all that looks like. Like it's going to be very centered on AI, like doing AI responsibly as it should be. Like there's a lot of problematic aspects of AI. But I will, I'll link this governance blog post in the show notes. And it's talking about like, how do. How on a, you know, like a civilization level do we govern something as ambiguous as AI? You know, in many ways it's virtually impossible. Like, what are you gonna have some sort of like, you know, global monitoring regime or something like that that will somehow like enforce all of this stuff. But the fact that it's hard doesn't mean that we shouldn't be thinking about, like, what is. What is kind of the right way to handle this stuff. Much like, you know, climate change, like, it's same issue. Like, how do you get everybody to uphold the same standard? The fact that that's really hard to do doesn't mean that it doesn't need to still be a conversation. But most notably was the, the statement they made to start off this little blog post. And so OpenAI, obviously they have their own interests, but they ought to be more attuned to what is around the corner than anybody else. They said they stopped development on GPT5. Somebody, a Microsoft person, recently said that they actually have dates for GPT5 and GPT6. So some conflicting information there, but ultimately, like, we don't know what exists in a lab right now and how much better it is than the stuff that we already have our hands on. We know there's amazingly powerful stuff like GPT4's image processing that they're just not giving us for safety reasons. GPT4 was done being trained last fall and wasn't released until almost six months later for them to do a bunch of safety and alignment training on it before it got into public hands. So right now in a lab, there's all sorts of cool stuff going on and nobody's more plugged into what's happening in a lab right now than OpenAI. So while we kind of have to, like, see the things they say through a filter of, you know, PR and, and a bit of how does this reflect on, on OpenAI and affect the business and all that stuff, this was an interesting statement that they started off this blog post with. It says, given the picture as we see it now, it's conceivable that within the next 10 years, AI systems will exceed expert level skill in most domains and carry out as much productive activity as one of today's largest corporations. Corporations. So they're saying that within the next. Within the next 10 years. This episode is sponsored in part by Finn Daily. Finn Daily. Ever heard of IT Entrepreneurs? Let me tell you, they don't have time to focus on their numbers. But Finn Daily solves this problem by sending business owners an automated daily email. You know what automated means? You don't have to do a thing. It just sends. It goes, goes right on out. It includes key metrics, bank balances, and accounting for about a buck a day, a dollar a day. You know what? That's less than the cost of a cup of almost anything these days. In addition, Finn Daily allows the emails to be white, labeled, that is Slap that firm stamp on there. Putting your accounting firm top of mind with clients. Because that email's sliding right into home every morning. Try my Brainchild. It is. I did, I did make it. I created Finn Daily. I don't own it anymore. Try my Brainchild for free. Link in the show notes. Go check it out. Pretty cool, hey? This episode is sponsored in part by dark horse CPAs. Ever heard of them? Hey. In an age of rapidly evolving AI, Dark Horse CPAs has decided to entirely abandon the computer and go back to doing everything by hand. Bold move. We'll see how it works out. They believe that public accounting is rooted and miserable toiling. So to deprive accountants of the pride and bragging rights of working over 3000 hours a year is a slap in the face. Sure, they could use computers to work half the hours for five times the what is this ad read. But what is an accountant if not burned out and underpaid? Sure, they could be known to their family and actually have relationships with them, but the truth is, accountants just don't like people. And if you're looking for a firm that will exercise an undue amount of control over your life and actively participate in the decline of your physical and mental health, you should get in touch with the overlords at Dark Horse. I feel like somebody that doesn't like Dark Horse just bought an ad on this show. This is tongue in cheek. They're passively seeking mediocre talent to fill roles that are obsolete to the modern accounting firm and would appreciate the opportunity to stunt your career growth for as long as you let them check out the link in the show Notes to Dark Horse CPAs to learn more. So if somebody told you, I mean, and this is, I mean, we can go back to the, the pipeline issues around this narrative too, but if somebody ultimately told you, like, hey, within 10 years, there's a pretty good chance that AI will exceed your level of expertise in the domain that you've spent your whole life working on. Like, you'd want to know about that. You'd be thinking about that. We could at least be thinking ahead to, like, okay, if it does get there, what's that mean for us? And what does our work look look like on the other side of that? So as I say this to kind of temper that, like, 80% of people still don't think it will impact them ultimately. None of us, like, know for sure until it happens. Right? And there's going to be a lot of nuance in there. There's going to be aspects of it that are impacted, other aspects of it that aren't. But I still think, like, for me, I want to think what's on the other side of that if it did happen, rather than it being this big sort of ambiguous thing. So let's follow this timeline. We'll focus on tax this time. Maybe next Friday. We do it on accounting. Like, what if AI truly, like, takes all the legwork out of bookkeeping? And what does running an accounting practice look like on the other side of that? I think that would be a fun one to do, too. For this exercise, I think we probably have to define exactly what it is that AI is capable of doing. I would say it's capable of calculating taxes correctly, but let's also say it's capable of filling out forms that can be submitted to taxing agencies and asks all of the necessary follow up questions to get complete information, which, I mean, just saying that out loud sounds wild, right? So like this is like a phenomenally powerful AI in this scenario. It is asking the follow up questions of your clients to ensure that it has all the information that it actually needs to do stuff correctly. Which is the biggest issue with any sort of DIY self led tax prep right now is you can check your W2 in there, but you don't know what else to ask and what else to provide it. And that's a huge part of the value of a tax Pro is not only their technical knowledge of what else could be out there, but, but all the other scenarios they've seen from other clients that plants that little seed in your head that, that's like, oh yeah, no, this other client had this other kind of weird thing that worked out for them. Does this apply to this other client? But we're just going to assume the AI can do all of that and ask all those follow up questions. In reality, what would running a tax practice look like on the other side of this? What would the client experience look like? How would we spend our time? A few thoughts I jotted down. The client still wouldn't want to answer all the questions. So like a lot of the narrative, some of which I don't agree with, a lot of the narrative recently around the tax prep process and designing your ideal client experience in a tax firm has been, buddy, if you're just going to make me complete this giant form, why wouldn't I just go do it myself on TurboTax? And I mean there's a hundred reasons why that breaks down. I mean the big one being TurboTax doesn't handle any degree of complexity and there's a lot of context there around. Yes, it is bad to give clients humongous generic organizers and questionnaires, but unfortunately there is still a lot of stuff that we have to ask that I sure wish we didn't have to ask. But like the reality is like we still have to ask a lot of questions. And so as much as I would like to do this entire thing for somebody and take all that work off the client's plate, unfortunately it still involves more client involvement than any of us would like, the clients included. That being said, there's definitely wrong ways to do it where you put too much of the onus on the client. But if an AI could do all of this stuff, the AI would still ask all the questions that need to be asked. Maybe it would be Better than the really bad firms who are giving just way too big and dumb of lists to clients. Maybe it would be better and more contextual than that, but it would still require the clients to answer a whole bunch of questions. As a first thought I had. Second thought I had, is your client going to answer the questions correctly? Even something that seems as binary as are you capable of answering a question about a thing when it comes to tax law, like is actually a really squishy thing. Like there are questions for which there aren't super straightforward answers and there's also answers for which you know this isn't right. Or I saw this other thing and I've worked with this client before and I'm pretty sure that this is what they actually mean when they give that answer. So to me there's still a lot of like room for fudging and like just the fiddly questions. Where would I. Let's say, I mean, and maybe there's a good analogy. Let's say there was a turbo tax that could handle like any tax return, right? That could handle just crazy levels of complexity. That's essentially what a killer tax AI is, I guess. Right? And maybe in a chat interface or a voice interface where it in a really dynamic way asks all the follow up questions that it needs to and it can handle any complexity of return. It's just basically a super impressive chat based TurboTax. If that tool was out there, would you trust your clients to complete their tax return correctly? Probably not, right? Like I don't like, even if that tool was out there and you put it in their hands and it was really easy to use, I still don't know that they would complete that correctly. That being said, if you don't wear into that like what part of that process is the tax pro then engaged in? And is that tool for the people who like, for the people beyond a certain complexity where it's like, no, I either don't trust them to do the work or they're just not willing to do that volume of work like unassisted, then what does that process looks like? Probably looks a lot like what the process is now. Now. But we're using tools that save us a whole ton of time because we can understand the output. Like we can validate that the output is correct and we can navigate that nuance for them. So it's like this big, incredibly power, powerful TurboTax for pros or something. Just like a much better version of traditional tax software today. Third issue, the client ultimately wouldn't understand the output if we were totally removed from the process. Not sure how big of an issue that is right now or not. Because I would argue the vast majority of clients already don't understand the output. And that's just how things are right now. Like the Rudy Giuliani letter that he would send in with his tax return every year, Was it Rudy? No, it wasn't Giuliani. It was Donald Rumsfeld where he would, he would mail in his tax returns and he would put a letter on top of it and sign it and say, you know, because of the vast complexity of the tax system, I've worked with my tax accountant to do my best here, but I fundamentally don't understand if any of this is right. As if somebody at the irs, you know, the person that processes, like in the warehouse, processes his paper returns that comes in, as if that person was going to stop and read it and then do this, like go to Washington and change the world or something. This episode is sponsored in part by Client Hub. This week and Tales from the Hub. You remember last week, Super Smart accounting firm, totally a real firm, fixed their client request process by implementing client tasks in Client Hub. The leveling things up, they discovered Client Hubs internal workflow features. This week they use it to track deliverables and they were wowed, wowed when they found Client Hubs magic, Magic workflow. It's GPT Power and Client Hub. Stop the music. Keep playing the music. It is still an ad read. GPT Power built right into Client Hub. Give Magic Workflow a few words that describes the work that needs to be done and it will generate the initial task list automatically. With AI now, Super Smart accounting is no longer putting off documenting their processes. Within a week, they're all set with their client work and detailed processes. It's all captured in Client Hub. Somebody want to take some pto? Remember that episode from last week? Unlimited pto? No problem. Because somebody else can pop on in, do that work. Because it's all documented in Client Hub. How's your documentation looking right now? Bubble? Hmm. That's what I thought. That's it for this week's installment of Tales from the Hub. Learn more about Client Hub. The link at the show notes so clients wouldn't understand the output, but that's also probably the case today. How about more strategic decisions? And maybe the AI could still do this. But sometimes when you're filing a return, there's like you could genuinely do A or B. And the right answer of whether to do A or B may depend on next year's tax situation, you know, do they need cash now versus do they need cash later? Maybe the AI could navigate them through that outside of just filing, I think that's an area where, filing aside, we're probably more than focused on like the kind of adjacent advisory stuff, which a lot of tax practices already do. But I think that's the stuff that would probably hang around for longer if you had a mega powerful AI to do all of this stuff. So like that can help you with filing, but what about, you know, proactive strategies that you're doing during the year that impact how you spend money and that sort of thing in order to optimize for a certain tax result? That's probably a on the spectrum of things for AI to do that's probably on the harder end of the spectrum from the deterministic given all of these inputs. Here's a tax filing and submission that proactive strategizing, that's probably a little bit harder of a thing for an AI to help to help with where your tax advisor may be more helpful. Also representation, I think that becomes a much bigger like part of running these tax firms. If the filing itself is vastly simplified, then what happens when there's issues with those filings? And I mean we're gonna like the next couple of years I think we're gonna see some shockingly bad AI enabled tax services. And we already see this to a degree in bookkeeping services where like in my opinion some of those productized services are getting more and more freewheeling and like telling people, oh this is all you need for it to be done correctly. And so the quality of what's coming out of those services isn't very good. We're gonna get that with tax stuff over the next few years for sure with these like AI enabled tax services. Now ultimately I think the reality is that oftentimes the type of people who are coming to use these services, it's very likely they're already not filing correctly as it is. So are these ultimately better than what those people are doing right now? I don't know, maybe. But the net impact is going to be a whole bunch of really crappy tax returns getting filed. There's already and I don't even want to send people to them. I've already seen a AI income tax prep service and I don't know if it's just a accounting firm that says we now have an AI model to do taxiderm, but it's really backed by a firm or I don't know if they've actually think that like they have designed something that can do this accurately. But we're going to see a whole ton of those. It's honestly an interesting marketing angle. Like we've seen this time and time again with R and D credits. People that build a practice around R and D credits by saying like, hey, here's a way to get free money. Oh yeah, by the way, we'll also do your tax and accounting. And that's just how you get people in the door. Saw a bunch with ERC credits. Like there were people that did ERC credits responsibly and they used it for long term client acquisition. I see no problem with that. So this kind of AI tax thing may become a marketing play. Specifically when it comes to stuff like being able to chat with your tax return or as part of working with that firm, having access to a tax chat bot that can answer questions. It may not even change the actual tax preparation service itself. But like I have, I have no doubt that AI enabled X will absolutely become more of a marketing thing for firms. And honestly, like I don't have a problem with that. Like if it's gonna appeal to a specific type of person, that's a great client for you. Like I don't, I don't see any harm in that. Like ultimately you're still responsible for the output and for the correctness and all that. So like that's always going to be the case. But like keeping that in mind, like I don't myself, like I'm not spooked by that. I think what's going to be problematic is kind of like these productized bookkeeping services, these volume based tax filing services where they will kind of sort of put together a tax return and file it. But it is far from a robust system. Like I think we're talking about something that is probably worse than like the TurboTax and other productized systems that we've seen out there. So on the other side of that representation is probably going to be much bigger, you know, tax resolution, like troubleshooting those issues that people have. But how would it change? You know, what, what would we do less of? I think this is all taking us to a place of doing less of the actual preparing of the return. You know, if you take a tax practice of all of the things that are happening in a tax practice, especially during busy season, like a huge chunk of that time is just doing the work. But if you drill down into what doing the work is, that's also a lot of like sending questions to clients and following up on stuff that you need from clients. And it's a lot of stuff that isn't actually creating work papers and putting information into the tax software. But that part of our process, I think is the most ripe to be improved and done away with. And ultimately is that a bad thing? Like, I know we associate King the numbers into the tax software as this is part of my job. But if that went away right, like, I think that would be fine. Because our expertise is not King numbers. Our expertise is like validating the output and making this suggestion that will save them money and saying like, I'm hearing that you have xyz, which probably means you also have abc, like, do we have complete information? They're like, that's the real value of the Tax Pro in my mind. So the stuff that it's going to, you know, and this is in a very impressive AI timeline, the stuff that it's going to be best at is probably the stuff that I would most love to stop doing. So like the, the bigger question there is probably, does that change, you know, price pressures on our services? Does that, does that mean that we can now only charge half as much for the work that we do? I don't know that it does. Because if you think of about like what that tax prep market would look like from a bigger picture, we would see an explosion in productized services. And I do, like, productized services have been growing a lot lately and I think AI is just going to create another wave of that. You'll have an AI enabled kind of new wave of productized services who are the first ones to say, like, we've solved income tax with AI. Whether they have or not, the net effect of all that is going to be that the complexity that these productize, the complexity of the return these services can do, like, will continue to increase. And that's totally fine because right now I would argue the turbo tax and the H and R block level of complexity where somebody can go out and get a return prepped for one to $500 through a productized service. I would argue the gap between that level of service and what it takes to get into a professional office is too big. There's a big void there. Like to get into a tax firm these days, like most mature firms, like, you're easily one to two grand a year just to get in the door. And in my mind that is too big of a gap between what we do and product services, which are wildly different, very different level of like robustness that goes into those things. So if the like, if the complexity of what those can handle continues to come up, I think that's fine. I think it's only problematic for you as a tax pro if you are setting up shop in the same place where those services are, which you just shouldn't be. You shouldn't be doing $300 tax turns, shouldn't be doing tax returns or just W2s. Like be a good advisor, send those folks out somewhere else. There will always be clients that just want a pro to do it, don't want to rely on those services. That's also fine. I did a lot of those. The people who are like, I get that I could probably use an H and R or TurboTax, but I would rather have you do it, that's fine. As long as they know that they're like paying a big premium for that. If they're happy to pay that premium, that's their decision. Right. So I do think the kind of the market as a whole would shift, but ultimately what it would be displacing would be the stuff that I think is least valuable for us right now anyway. Like a client doesn't care how long it takes you to key stuff into a tax return. They're more concerned with the outcomes. And maybe my biggest single takeaway here is if you think down that timeline of what if there was this phenomenal turbotax that could do anything if everybody went out and used that service. One, would the output end up being correct? I would say oftentimes a no, because the inputs, if you leave it up to the client to make the inputs, given the current state of the tax law. Right. If you leave it up to the client to, to make those inputs, I think there's probably gonna still be issues there. And then two, you have a lot of people that just aren't willing to do that work. And maybe the AI system can make that as easy as possible. But two, there's still gonna be a big desire for that concierge alternative who's not just going to file the stuff for you, but be more proactive in how we advise and help clients and all that stuff. And it all kind of brings me back to what was my thinking around when they originally did that GPT4 announcement stream, which is who has the most upside in this future timeline. I honestly think it's tax pros. So like the tax pros most at risk, like they're going to be the real entry level ones and everybody starts somewhere. Right. And so if, if there's anybody that loses out in this situation. Maybe it is like the first year tax pros that are maybe maybe adding this onto a bookkeeping business and went out and got their EA or a certification or something like that, those folks got to start somewhere. Man. We all cut our teeth on returns that were very simple and shockingly cheap, like in retrospect. And you wouldn't do returns at that level again. So if maybe anybody is left high and dry in this situation, maybe it's the really entry level tax practices. But otherwise this ultimately becomes a really good asset for tax pros. I think because we can't ultimately depend on our clients, given the current state of the tax system and all of that, we can't rely on them to do this correctly. And when there's as much dollars at stake as there is for high wealth, like people around tax filing, you know, having paying somebody a few grand to do their tax reporting, like, I don't see that going away. Even if you had AI that could do the tax filings phenomenally, if anything, it would probably just change our business model. It would probably force us to ensure that we are moving up market, not doing the same things the productized services are doing and lean more into tax advisory rather than just the transactional helping people with filings. And to be totally honest, that sounds pretty good to me. That doesn't sound too bad. And so the net effect there is like, I've got way better tech that can let us help way more people or go way deeper with the same number of people. And I think that's pretty cool. That's exciting to me. What do you think? Like any other kind of variables in that timeline that you see around, how it changes the larger marketplace, productized options, what we do, how that changes the irs, even like I'm always shocked to think, and this like dawned on me when I was doing my first audit is like, by and large the IRS doesn't have quote unquote tax software. The way that we think about tax software, if you ever been through an audit and almost all the ones I did, they were like, hey, we're going to make these changes. Could you run this through your tax software and send me an updated copy? Like they don't go back and use Drake or Ultra Tax or Profax or something like that. And so, I mean, the IRS would absolutely invest in this stuff too, right? For the same reasons that we all would. So maybe they're going to have better calculation engines. I mean, I think we're all in favor of a smarter, more Targeted irs, if, you know, correspondence, audits and all that stuff can be more intelligent and actually focused on like where that stuff needs to happen. Because there are a lot of places where it needs to happen and that like they need to go after the cheats. But interesting to think how that would impact the irs. Also I know they do some machine learning type stuff already on basically just finding outliers. Like so you know, if you've got a million dollar management fee and other expenses, that is an outlier. And that's the sort of thing they right now that they refer to as machine learning, like AI, you know, intelligent solutions. Like, I don't know, like that seems pretty obvious to me that that would be a, a little suss and a good one to take a look at. But I gotta imagine that their systems would get much better too, which, you know, there's the very political narrative of a big spooky irs, a smarter irs, a bigger IRS being bad. But a smarter IRS I think is, is great because it's less likely to waste the time of the people for whom it shouldn't be wasting and more efficiently goes after the people who are just the cheats. Right. So I don't know, like I have, I honestly have a really hard time coming up with negative outcomes here. Again, the folks at the super entry level, like that is probably just the no man's land, which it already is. Like if you're setting up camp where turbo taxes and all that. But I have a really hard time seeing this being, being a negative. I don't know any negative outcomes you can see here. I'm not feeling too spooked about it. So that's it for this week. Thanks for coming and hanging Engage. Next week, I'm pretty sure the next show I will be in Las Vegas. If you're gonna be at Engage, please come out and hang. I've got two talks on Monday, I've got the AI keynote at the end of the day on Tuesday. And please bring out your best heckling, inside jokes, hair questions in the Q and A's, all the usual stuff. You know how this goes. It'd be good to see you all if you're there to come and hang out. So that's it for today. Have a good weekend. I will be back on Monday.