044 PTO In Accounting Firms

It's Friday, so let's talk about time off vacation policies.

What should your vacation policy in an.

Accounting firm look like? Like, what is a modern approach to this? I think there's some traps here and it is not as simple as sometimes it's made to sound. But even if you're a solo operator and you're thinking about what's a reasonable framework with my employees, with my contractors, if I'm going to make that first hire, how do I set that up the right way from day one, let's talk about it. We're going to talk about the right solution for different types of firms because it is very firm dependent. So come on in, let's talk about time off policies. Okay, little under the weather today. If I sound different, that's why this was almost the first time since starting the Daily show that we didn't have a Daily show. And part of me wanted to do it just to give my, myself more permission to do it down the road. And it won't be a daily thing forever. But I also got excited about this subject, so let's just do it. And it gave me like a, it forced me to go take a shower and just some other basic human hygiene things that probably should happen today, regardless. Okay, this discussion kicked off on Twitter. Okay, so Alison McLeod started this discussion on Twitter. She said, what's standard PTO these days for accounting firm employees and owners? And you know what? Turned into a big old fist fight, as most things on Twitter do. A lot of people were disappointed by the replies that came through there. And it was totally across the board from three or four weeks to various flex plans to unlimited PTO to calling them different things from sick days, vacation days, pto, sad days, rolling that stuff all into one. And a lot of people were pretty opinionated on the right and the wrong ways to do that. And I think that often stems from when we have a really negative experience, maybe with a past employer or when we have a very clear vision of, well, hey, here's the way to not do it. And that can kind of skew us kind of to the opposite end of the spectrum. So today I want to run through like what my experience was with various different plans and why I think the way that work has changed in the last three or four years actually impacts what the right answer here. What the right answer is here. Because my initial, like my initial thoughts here are unlimited PTO can actually be a terrible thing for your business. And many of the original justifications for unlimited PTO where there are some Studies early days that said if you use an unlimited PTO policy, people will take less time off. And that was a justification for doing unlimited pto. And that's not why most people do unlimited PTO plans. They're not evil in that way. But there is a very real risk of that like that. I mean the reality is like we all look at our colleagues and we see how much everybody else is working. And I think oftentimes what we do is we just don't want to be the one. We don't want to be the caboose. We don't want to be like in the laggard group. So you look at all the people around you and you're like, I just got to make sure that I'm doing more than the people around me. And you end up turning what ought to be a positive thing into like a negative kind of cultural thing. But I do think there's ways to do unlimited. Right, and that's some of what we're going to talk about. So like, to start out here, let's zoom out and kind of look at like the progression of work. And because I think the, the you have firms that are at different stages in this progression and the right answer is going to be very dependent on what stage you're at. So starting off just like traditional on site firm, people go into an office every single day, they work out of the office and those firms are generally built around like hourly expectations of people to work, maybe expectations around billable hours and that sort of thing. And for this type of firm, because it's not generally going to be oriented around an expectation of output and because the work is happening while you're in a firm, some things work for this firm that don't work for remote firms. So for example, in my opinion like hourly based PTO generally is going to work best for on site businesses where you working is measured by you being in the building. Like, and obviously that's not work and that's not a, that is far from a perfect yardstick of how much work somebody is doing. But for many on site businesses, like that is what it is. And like that was, you know, kind of the old, old fashioned approach. Like you walk through the door and you punch your time card and you go in to work. And for those types of businesses, maybe hourly PTO plans work fine. But I don't think for the next type of business, remote firms, hourly PTO policies really work. So if the kind of the baseline of the traditional work environment is on site, then the second I have here is remote. And that is simply a distributed team. Doesn't matter where you are. It's people working, not necessarily from home, but just not working together in the same place. That could be, you know, co working spaces could be absolutely working from home. And one of the big things remote work has done, I think, is like showing a light on how arbitrary those hourly PTO policies are and what a poor yardstick. Just hours of butt in a chair is for work and by extension time off. And so when you're, if you're running a remote practice, to me there's a little bit more of like an inherent. There has to be. To do this successfully, there has to be a little bit more of a focus on output rather than like hours spent in a chair. Because if you're having people come into a building and like the traditional work environment of like, oh, the boss is walking around and making sure people are working, which is like, you know, if they're gonna steal time from you at home, they're also stealing time from you in an office. But this gets like, this whole like old timey notion of hours in a chair I think just looks a lot more absurd in the framing of remote work because people work at different times and there's not that same level of transparency in how much everybody is working. And like, you know, I think we all probably had times in our careers where it was like, boy, I can't leave the office yet because nobody else has left, right? And obviously that's a silly, arbitrary thing, but that like, that sort of cultural kind of, I don't know, vibe doesn't really translate to remote work. And so to me, like, you can do the traditional hourly PTO model with remote work, but I think it feels kind of silly, like for firms that are going remote for the first time, it's kind of like firms when they very, like when they go paperless for the very first time. The version one of that is all of the exact same things that you did on paper but done in virtual form. So like, when you first went paperless, you were like, here are all the things that I do on paper. And I want to replicate that as closely as possible in a digital fashion. And it's not until you spend more time with digital work papers that you're like, oh, like all this stuff actually isn't necessary because they're digital work papers. Or here's the workflow. That's much better in a digital setting. I think firms that are going remote for the first time kind of work through Some of the same things is they apply the same policies that they did to in person work, to remote work. And then they find that like, oh no, this is actually like pulling our attention to what different approach makes sense kind of in this new work paradigm. And you know, most firms, I would honestly say most firms probably had to grapple with this for the first time during the pandemic, especially small traditional firms who I don't think were really leaning into remote work that much. This is something where a bunch of people really had to like get up to speed on this quickly. And then now coming out of the pandemic, we're seeing kind of that counter reaction to like pulling people back into the office. And now we're kind of something in the middle. But I think what remote work is a precursor to is like the ultimate work paradigm. And so this is like the third step in the progression to me, which is asynchronous work. So asynchronous is like a word that I feel like we've been seeing more and more in different contexts. And one of the examples we've been seeing a lot of it around has been communication. So asynchronous communication tools and leaning into asynchronous communication rather than synchronous communication. And so an example of that is, you know, a meeting where you're sitting down with a client or with a bunch of team members and talking live, that's an example of a synchronous meeting. Whereas me recording a loom and sending that to a team member or something like that, that's an example of asynchronous communication. And there are many times that we do things synchronously, that is live together that could otherwise be done asynchronously. And this is kind of an extension of, you know, this meeting probably could have been an email. But when you kind of pull away from synchronous work and lean more into asynchronous work, to me it is like the next evolution of work because most of the things that we do do not have to be done at a certain time or sitting down alongside a colleague. And there's a lot of exceptions to this from getting new people up to speed and just like times where that collaboration is necessary. But the 8 to 5 work day, like to me is a, a kind of blue collar leftover, you know, vestige of like what factory work used to be. And, you know, that whole paradigm of 8 to 5 work in my mind doesn't really apply to like white collar work. And Knowledge, work. And, you know, most people are not capable of doing deep thinking for that long even. And, you know, life circumstances and even just personal preference. Given the flexibility, not everybody's going to be at their best from 8am to 5pm and so if you can give people the flexibility to work when they can work and work when they're at their best, it leans a little further into like meeting people where they are, in my opinion, and giving them the space to do their best work. And like a firm that first transitions to remote work where they're like, okay, now we're just gonna all work online from 8 to 5. They're not really doing that. And I think you still end up wasting a lot of time on synchronous activities, meeting with staff, that sort of thing, where if you lean into an asynchronous style of work and you don't always have that expectation that the person needs to reply immediately, it actually unlocks a totally different level of freedom in the way that people work. That then leans into what I think the ultimate goal in accounting firm. In an accounting firm that is to incentivize people based on output. So not based on the number of hours that they have a butt in the chair, but some sort of yardstick of output in their domain of what they're working on. And so that may be, you know, some sort of revenue threshold in terms of the tax work that they're getting through or a revenue threshold around the monthly closes they're responsible for, like their book of business that they're managing. Ultimately, the solution to all of these problems, in my mind, somebody's leaf blowing outside, pardon the leaf blower is having a yardstick by which you can assess somebody's performance that is not time based. And this solves for just a mountain of issues from how to set standardized expectations across team members and how to keep from disincentivizing your high performers when they go the extra mile and get more work done than other people.

This episode is sponsored in part by Client Hub. That's right. Hey, this week on Tails, we from the Hub. Remember last week when we did this? Super smart accounting firm figured out that getting answers from clients was the key to unlocking the profitability of their firm. So they chose Client Hub, a practice management system with a client portal at its core. When they rolled it out, the clients were like, omg, thank you.

Beautiful and modern.

Modern, simple experience.

They said their exact words for this hypothetical firm. And a killer mobile app.

How many of our like, accounting platforms right now have a helpful mobile app. Not many. Now the firms and the clients are on the same page about about what's required to do the work. The staff at Super Smart Accounting Solutions can assign clients tasks for the clients to tackle. They can be like a yes, no answer. They can be a request for files, even requests for categorization. It'll automatically sync back to QuickBooks or zero. That's handy, right? Whatever the client task is. They discovered that their clients on Client Hub now respond right away and have overcome some of the like blocking that happens with getting the work done waiting for clients. Nobody likes that. Hey, to learn more about Client Hub and how you can unblock your life, check out the link in the show notes. This episode is sponsored in part by the fine folks at Cloud Accountant Staffing. Do you hire accountants? Bless your little heart. Not the best part of the job in my opinion. Not something I ever enjoyed. Well listen, you can build your accounting dream team with talented offshore accountants in the Philippines that work 100% full time for your firm. Their accountants aren't freelancing or contracting for multiple firms.

They're all yours.

They work exclusively for you and are incentivized to stay with you and your team long term. They're not going to get swiped. Cloud account staffing is 100% dedicated to the accounting industry and founded by a former accounting firm owner that understands your business, knows your pain points. They had to hire some accountants and they said, you know what, we're going to build our own pipeline in the Philippines, going to pull in some super talented people and then open that up to other firms. Basically that's the story. I've been talking about a lot about staffing, building more resilient staffing pipelines for your firms. I had staff in the Philippines like totally red pilled me to like oh.

Geez, like we need to globalize the way that we get our work done. Check these folks out link in the.

Show description cloudaccountantstaffing.com you look at issues.

Like people working multiple jobs. Right now you have really high output people. I'm sure accountants are doing this. Engineers definitely do it, who are working multiple full time jobs at a time because they're very high performers and they can actually swing that. And like the initial kind of reaction to all that was like oh my gosh, that's wrong and that's dishonest. When I think it really just shines a light on the fact that high performers like are chronically undervalued in the traditional work paradigm because we're not measuring the difference in output between what they do and what the colleagues do. And if you've ever been a high performer in a, in a firm environment, you know exactly what that looks like and how much of the weight you carry, but your salary and all that like that is generally going to be dictated by your role and what your colleagues make and all these arbitrary things that shouldn't have an impact on how much you're contributing to the company. And so ultimately the goal in an accounting firm is how do we get closer to output being the yardstick rather than time. And there's so much nuance in that, you know, there's downsides to it where when everyone is compensated based on that output yardstick, then the game instead of being ours, the game becomes the calculation metric, you know, and how you fiddle with that and the reality that every time you change that you're impacting everybody's pay. Like, there's a lot of really hard things when it comes to output based pay. How do you reasonably split that stuff up amongst teams? You know, if teams, if you have a pod of people that are responsible for a certain amount of output, how do you divvy up that output based incentive when a number of different people have their fingers in the pot, like, and are responsible for that output? It's hard, but like, it's ultimately what we all need to be striving for because it solves for all of these things, from people who are going out and taking multiple jobs to setting fair expectations for, you know, your entire team and upholding everybody to the same standard. And to me, how you approach, you know, what that yardstick is for your firm and where your firm is at in that progression of how you work, I think dictates ultimately what the right PTO solution is for you. So in a traditional work environment where people are, you know, coming into an office and all that stuff, I genuinely think if you roll out an unlimited PTO policy, people are just going to work. More people are still going to pay attention to what time everybody else got into the office, what time people are leaving the office. Like, I don't think you've actually escaped that sort of kind of political gamesmanship of looking like you're working as hard as everybody else, which is just a, I don't know, something that takes away from like the true reality of what's happening and people's output. So I'd honestly say, like, I think the worst version of a PTO policy in an accounting firm is probably an unlimited policy in a traditional Work environment. The flip side of that, I think the best PTO policy, and obviously that's glossing over like, you know, somebody that doesn't give any time off or two weeks off here, just something absurd. On the flip side of that, I think the best version of a PTO policy is unlimited. In a asynchronous work environment where folks are incentivized based on their output. So if they make their money based on the work that their team ships out the door or the work that they get out the door, it makes that whole unlimited PTO conversation really easy. And honestly, day to day, like, this is the beauty of asynchronous work to me is you don't care how or when they work. There's gonna be exceptions to that. And you do have to meet from time to time and all of that. And that's just a matter of getting stuff on people's calendars. But I was first like unlocked to asynchronous work when I pulled my first video editor into my little production company. And we did that completely asynchronously. And I had a whole kind of like getting started page for her on day one in notion that ran her through everything that she needed to know. And then that page would be updated week over week for like, hey, here's what we're working on this week and here's what we got coming on, coming up around the corner. And I realized there's so much more of what we do that can be done asynchronously. I think, like doing a lot of that stuff live, there is value in it. And you do need to do that at times to like, still build that human connection. But we do a lot of it because of how work used to be still, I think. And asynchronous work, on the other hand, like, forces you to very clearly think through and document all what your expectation is for that person. And oftentimes we don't do that. We like, that's why we put the meeting on the calendar is so that while sitting in front of that person, we figure out, okay, what are you going to do? And what do I need to show you? And all of that stuff when the way better use of everybody's time there is for you to think about that ahead of time, to document it and to cut that person loose and say, hey, feel free to reach out if you need and any help. And so an asynchronous firm is people work when they want to work. Honestly, if you don't work today, it doesn't matter to me. All I'm worried about is your output. All I'm worried about is the fact that you're managing like this volume of business and the clients are happy. And beyond that, I don't have to worry about when you're working, how you're working, if you're, quote, unquote, stealing time from me, and if you're even working another job. Like, none of that stuff ultimately matters when you have got to the point where people are incentivized based on their output and you've stripped away all of the old timey notions around how and when people should work. Because ultimately all that matters at the end of the day, I think, is that you have set a clear expectation for what needs to be done, and people are meeting that expectation. And if they're not, you come in and you help them. Like, it's not. Like, it's not a. It's not a, I don't know, pass, fail sort of thing. Like, you don't want to culturally make that a scary thing. But I do think everybody, the more explicit those expectations are, I think everybody appreciates that, and that is a more fun work environment. I'd done stuff, you know, at my firm in the past. You know, historically we had the more traditional. I think everybody got five weeks by default, but then they could either bank overtime and take that time off also, or they could cash it out. And we did that for. I mean, I was at the firm for 15 years and we did that all but the last couple few years. And I remember before, I think it was 2019, like pre pandemic, on average, our tax staff took like nine or 10 weeks off a year or something like that. But it was something where I gave people the flexibility to either take extra time off if they wanted it, or cash that money out if they were at a stage in life where, you know, you got young kids, so you're like, where am I gonna go? Like, what am I gonna do? Or we're buying a house, so we want to cash that out. Like, I do think there are perfectly fine ways to do the more traditional PTO approach. But when I see this conversation online and in conferences and stuff like that, I do think it can get very oversimplified because it depends on at what stage your firm is at, kind of in the ways that you work and getting to output based work, it is hard. And there's a lot of aspects, like there's a lot of challenges that introduces that you don't have in the traditional model of work. But it's absolutely, I think, where we need to be striving to get to from an aspirational standpoint. Like, if you drew up your dream accounting firm from scratch tomorrow, it is, everybody's working to this specific expectation. I don't have to like, micromanage and hop in and, and, you know, put monitoring things on their computer to make sure that they're working from time A to time B. Like, it totally sidesteps all of those complicated issues, leans into the very best parts of remote work, of asynchronous work, and I think is ultimately the most fair thing to your team members because how they're compensated. Success is like the yardstick is output. And ultimately that's probably not the only output. You know, the, the only yardstick. There is more to working for somebody and being part of a team than like, did you deliver 100 widgets this month? But I'll tell you what, it's a whole lot better than did you sit in your chair for eight hours today? Right, Enough on that. Following up on actually what we talked about yesterday. So yesterday we ran through the. Was that yesterday who this dick Will, man. We ran through the new study where GPT4, like, aced every single accounting certification test that there was. But the way that they did it was with, quote, unquote, chain of thought prompting. So they basically wrote the prompts in a specific way and it performed 16% better. So switching from model 3.5 to model 4 yielded a 16% improvement. But then just wording the GPT4 prompts a certain way yielded another 16% improvement. And why I think this matters is, I mean, look at the future right now. So much of our future today looks like, who are going to be the best people at working with language models? Like, everything we're doing now, like all the AI stuff, it is all like prompt engineering around language models. The cool new AI features that companies are launching. It is prompt engineering. They're doing the exact same things that we are via ChatGPT. They're just prompting it to give a specific output. Like, that's all there is to it, is prompt engineering. So I was thinking more about this chain of thought prompting and I, you know, I mentioned the other day, when we're looking at things like tax calculations, if somebody builds out just this ridiculous chain of prompts that is much more step by step, you're going to have, you know, a solution that is much better for calculating taxes and stuff like that. But thinking through this, I've had this question that, like, has been my test that I've thrown out all of these quote unquote research tools, which is, and it's a tax research question. So for us Tax, when it comes to deducting the business use of your home, there's two exceptions to the quote unquote exclusive use rule. That is, you don't have to use that space exclusively for business if you use that space for storage of inventory or for daycare. And those two things are clearly outlined in an IRS publication. And, and it's a item that's particularly hard for a search engine to find or something like that because they're exceptions to the rule. And so all of this documentation says, oh no, no, no, no, this is the rule. But then here's these couple exceptions. And that's what makes tax research really hard, is you can, you can read the rule and say, okay, here's the rule. And then 70 pages later, there's an exception to that thing that you just read. That's what makes that stuff so hard. So I've used this question over and over again with the original Bing with, you know, GPT3 back in the day. And it's kind of the one I always come back to whenever there's a new model or new methodology or new app or something like that. Most of them bomb it generally. Well, thinking about this chain of thought prompting, I started thinking through, like, what is a more intelligent way of asking a language model? This same question. And this week, GPT's browsing model, that is the version of ChatGPT that will go out and pull web search results. There was a change this week where now that model exclusively relies on Bing search. So when you type into ChatGPT's browsing model something, it will go out and pull the top results via Bing and merge those into the answer. But you can also, I learned through testing, just give it an explicit URL and say, according to this URL, answer my question. And for me, especially when you're doing technical research around tax and stuff like that, I want control over what types of sources it pulls from. I don't want it to go and grab like a nerd wallet article and say, this is the answer. Like, ultimately I need something more authoritative than that. So for this business use of home question, I use the explicit URL of the publication, which is like this big hundreds of pages publication with anything and everything about that subject matter. And I fed that into the prompt. And so I'll run you through the prompt just to like show an example of chain of thought prompting that got that actually is starting to me to look like a very viable tax research tool, much more so than it did initially. So what I have it do is I have it return three sections. First, a section called Excerpts, where it is 10 bullet points of the most relevant excerpts from the document or the URL I give it. So I give it a URL. And that's the only place that it can pull information from, right? I don't want to pull in stuff from other sources. I don't want it using its general knowledge in the language model. We don't want it to use that at all. We just want it to pull from the place we point it to. So the first thing it does is it bullet point lists the 10 most relevant excerpts to the question that I ask. Second, it answers the question using those 10 excerpts. And this is the chain of thought approach is we've now chunked what is the correct answer to that problem into. Here's the most relevant material, word for word from the source. Now make an answer according to these 10 bullet points. And it's the same way that we research as humans as we pull in these relevant bits and, and we kind of synthesize these to form an answer for a specific scenario. And then third, I have a generate a third section called Other Considerations. That's a bullet list of any other factors that could have impacted the correct answer. For example, if it didn't have complete information. So oftentimes the right answer is going to be very dependent on something that you didn't give it in the process of asking the question. Maybe it's the right answer for person A, but not for person B. And so then the model spits out any other considerations, and it does it really, really well. And it absolutely nails the answer to this question. And it's like a great advertisement for chain of thought prompting and how learning prompt engineering is like a really big unlock when it comes to getting the most out of our AI overlords. Hey.

This episode is sponsored in part by Firm 360. Firm 360 is a practice management system. It's just gonna help you get more done, help you run a more organized accounting firm. If you're out there running an accounting firm on a spreadsheet or on that legacy tool that your tax vendor said, oh, no, we're gonna bundle this one with you and it's gonna be like.

Free for three years, okay?

If you're that guy, you already know you made a mistake. Listen, Climb Cloud practice management systems, they're here to stay.

This is the future.

You just gotta get on board. Okay, let me tell you a bit about firm360. Nice thing about firm360 is it's trying to do all that stuff for you. It is trying to do project management, file management, time and billing. You're getting all that stuff into a single place in the cloud that you can work with anywhere. You can associate your documents with the projects that they related to your time and billing, all that stuff in one place. Just like that crappy old tax vendor told you their tool would do, right? How's that going for you? Mm, check out firm360. Developed by actually an accounting firm that.

Was like, none of these things do.

The things that I want them to do. Let's build our own solution, right? If that's feeling like you right now, check out firm 360.

I'll put a link in the show notes, but I set up this prompt to be this reusable research prompt really for anything. Not for tax research, for. I mean, think about, you know, the next time you have an issue with QuickBooks Online, say like, you know, for the URL, use anything from like QuickBooks Online help and like you provided that URL there or using Xero, same thing. Like you've got this, this kind of fiddly thing that you're struggling with. I see a lot of value in pointing it to the parent URL to say only pull answers from this site because the less specificity you give a language model oftentimes, the more likely it pulls in crap that just creates noise and decreases the likelihood you're going to get a correct answer. But buddy, this is like another thing I love about language models is these prompts are copy paste. You know, like one person makes a cool prompt. And so I'll put a link in the show notes. You can swipe this prompt. All you got to put in is the URL and the question and it does everything else for you. And so my experience with ChatGPT takes like 30 to 90 seconds for it to go out and pull this stuff. One limitation I ran into is that browsing model won't go out and read PDFs right now. But you know what? I did. And so like for an IRS publication, as an example, there's both HTML versions and PDF versions. So I just linked it to the HTML web page and I could see all that. But let's say it was only a PDF.

So I copy pasted all the text.

From a PDF into a notion page and then I Toggled sharing for the Notion page to put it at a public link. And this for a situation where that information would only live in a PDF, I copied the public link to that stuff that I pasted into the Notion page and gave that link to the prompt and it just went out and fetched all the stuff from Notion and it totally worked. So like, even if it only lives in a PDF, just copy, paste all that stuff into any public document and it will go out and fetch it and answer your question. And gang, that's really flipping cool because there's theoretically no size limits on this. And how many times have you had to navigate this big old piece of legislation or several different potential sources for information where you could probably build this out a little further to be like, I'm going to give you three URLs and you go pull these things from it. Really darn cool. Like, and way better, which is the important thing here. Way better than just asking the language model. And so the people that are just getting started with language models, like, that's what they're going to do, they're going to go to it and they're going to say, are there any exceptions to the exclusive use rule? And it'll give them a garbage answer and they'll be like, pah, my job's safe. When like, the better version of this is, as we're learning, chain of thought prompting will get you a much higher quality response. But I don't know about you, I'm starting to like, save up my own little personal database of cool chat GPT prompts and build on those over time. And they're kind of like a cool, like, I don't know, like, little personal advantage kind of library. Like, and I get like the human reaction, and I can completely understand this is I don't want to have to go to that level of work to get a machine to do the work that I want it to do. And I think that's something that as a society we're all grappling with right now and maybe trying to get over. And some like, the real nerds like me, like, that's easy to get over. But if you put this, if you frame this like through the amount of work we do with fellow humans to get on the same page about things and to train people and all of that, you know, Is training and upskilling other humans a higher calling than becoming a robot whisperer? Yes, it probably is. And so I don't want to like, in no way don't hear this as Devaluing the value of sitting down with a fellow human being and showing them things. We should all be doing that and, and pouring our, our brain juices into other people. But that doesn't mean you can't still.

Be a robot whisperer too, right?

So like, this is, this is really cool. And if you, if something like this makes a huge difference in the quality of the output that you can give that you can get from it. My gosh, learn it, right? This thing is effectively free. I mean I did this. You have to have GPT plus to do the, use the browsing plugin right now. But we're talking $20 a month for something that you basically have unlimited use of. You can use 24 hours a day to do this work for you. I'm really excited about this and I'm, I'm, I'm thinking more about like, what are other chain of thought examples that could get really powerful. And like tax calculation is probably the example I'm most excited about because if you think about the different level of granularity and asking for the outcome, you know, at 100k in wages, blah, blah, blah. How much tax do I owe? You are asking the model to do so many things in between there, which if it can do it, cool, great, awesome, right? But it's going to be capable of so much more when somebody takes the time to do chain of thought to an absurd degree with a tax filing. And that's probably going down to every single line, every single box and that's going to be so much higher performing.

When you think of everything that goes.

Into a tax return, then what is the taxable liability? And gang, like, I think ultimately that's how we get AI to prepare tax returns for us probably is it gets that granular. And if you, and it seems absurd, but if you think about our tax software and the amount of technical things that are happening for every single box ever return and all of the logic that's happening there, like, yeah, like that's what AI based tax prep probably looks like is like chain of thought prompting like to an unbelievable degree and like people are split right now on what the future of AI looks like. If it looks like bigger and bigger language models, like if it's just like we get another GPT every year and it's that much better. Like so we just go 5, 6 of like, is that the future? Sam Altman, CEO of OpenAI says it isn't. He said this is about as big as language models will get. And like you get diminishing returns beyond this size. He thinks actually that the future of language models is just applying stuff like we have today in a more, like, specific application. He said this. He thinks this is ultimately how we get the most out of AI, is just putting what we have today on rails a lot more and giving it very specific tasks like we're talking about doing here. There's some other groups, like anthropic, that think, you know, big gamungus, mega models are actually the answer. And we'll see. But I'm hot on this chain of thought prompting right now, specifically on what. What are other examples of how we could get the model to do things in a sequential way to get us to a really powerful output. Like, I don't care if I have to sit there for 90 seconds as it tells itself all of this logic. If the answer is way better, like, cool, I can get behind that. So I'm excited about that. That seems like a. Like a really cool development, like a really big unlock. Hey, thanks for coming and hanging today. Have a great weekend. We'll be back on Monday, of course, and I'll see you then.