623 Why Claude in Slack is DANGEROUS To Accounting Firms [The beginning of a troubling AI era]

Okay, today on the pop, this whole like, Claude being built into your slack or where your team communicates. This feature they recently launched is a big time Trojan horse. And I actually don't know that I'm into it or not. The powers that be at Anthropic are really pushing this as like this huge new step forward for how we work with AI every single day. But I would argue it's going under the radar in ways that are a little bit concerning. Like I'm starting to feel like this is Anthropics way into like just running entire companies. We're gonna talk about this because this may actually become a really, really big deal and impact whether it's something you may go running to and say yes, I need this today or the first time you hit pause and you're like, okay, Mr. AI, not actually sure that I want you. This embedded in what I do. Also, I'm gonna show you three wild AI use cases from accounting firms, really cool stuff people are doing in the wild. Plus, would be a day when QuickBooks in Zero aren't the default accounting ledger choice for SMBs. How do you price your new client, clean up work and we're gonna talk about how to rebuild your work statuses. You know, like all the little steps your projects progress through, how to rebuild work statuses in a way that makes more sense for an era of AI, actually in a way that will make it easier to pull AI in to help with that process. My friends, we've got plenty to cover today. Come on in, let's have some fun. (upbeat music) (upbeat music) How do you do fellow accountants? My name is Jason Stats. Welcome to the Jason on firms podcast where we talk all things accounting firms. It's super cool. And lately that is meant talking a lot of AI, which frankly, I get kind of sick of. Actually, I've been going really hard on Claude since we got fable back ever so briefly. If you're listening to this episode, the day it comes out, we lose the really, really good Claude model as part of our subscription today. And we've just shipped some really cool stuff internally with AI in the past week, but I don't know about you. As much as I love it, there are times where I just need to just give the brain a good rinse. Go out in the woods and break a sticker too, you know? Run my fingers through some moss, climate tree. And the more I think about this story, the more it makes me feel like I need a break. So just to set some context here, couple really big developments and sort of best practices around AI usage that have really come in the last month or two. One is the use of loops and goals. And this is sort of a completely different paradigm of how you get AI to do something for you than, you know, prompting that we've kind of done in the past. Right now, everybody's got their little prompt libraries. And for the longest time, people had such pride in these like mega prompts. And we were all like passing them around like Pokemon cards, right? Loops are really goals. I'm just talking about this a little bit on the podcast, but it very much connects to the whole clot and slack thing. So I'm bringing it back up. But goals are basically an entirely different type of prompting an AI where you say, I just want you to be able to do this. I want you to get to this destination, port back once you've done it. And it is, as you may suspect, a way to get them to work longer on a task that has some sort of like a measurable definition of success. It is a measurable definition. And not everything has that, but many things that on the surface, you know, may not have a clear, is it done or is it not? Can sort of be re-engineered, re-manufactured, not re-manufactured, refactored to make this more clear. And this is kind of becoming an art form. How do I take a process and turn it into a goal? And behind the scenes, how these goals work, by the way, right now you can only use goals in clod, under clod code. So you just say forward slash goal. And you say what the goal is. And it will spin up all the agents it needs to try to get this thing done. And then it has one agent that basically with each turn will look and see, has the goal been completed? If so, it can stop work. If not, it needs to basically do the whole thing again and keep improving, keep working until it reaches that goal. And AI has gotten much better in the last six months in sort of long time horizon tasks and not doing the same things over and over again. But we now have kind of this emergent way of using AI that's much more powerful and much more suited for doing big sort of meaty tasks. And in an accounting firm, this makes vastly more sense than sort of chucking a task at an AI with a prompt. And it's sometimes doing it and sometimes not doing it because anytime you give AI a task to do, like if you are just giving clod some sort of task and chat or in co-work. Behind the scenes, there is sort of this allowance of compute that it is given to get that job done, like it won't work until the end of time. And you may have experiences before, where you give it a task that takes 10 minutes and it looks amazing and the results are phenomenal. But you give it a task that takes 30 minutes and the results are disappointing. It seems kind of dumb and you're like, how can this same thing be incredible over here? But just fall on its face with this longer project? It's oftentimes because it is not given, like allocated enough time to actually do that work properly. And so the whole allocation of how long is an AI allowed to work on something? That's a hard thing for it to just know. And the answer to that is complex because on anthropic side, like obviously more compute means more cost to them. But this is just a subscription that I'm paying is a fixed cost. And so sometimes it will kind of withhold some of the compute that it actually needs to get that job done. This is part of the appeal of goals. When you can very clearly define here's what done looks like and it doesn't stop working until it's done. Now you combine goals with their recent release of Cloud and Slack, which honestly when I saw this come over the timeline, I'm like, isn't that already a thing? What is, why is this really a big deal? Whether you're a Slack user or not, the more I have stood in this and the more I've kind of seen some of the commentary on this, it is a sneaky, sneaky humongous deal. So how this works is basically you connect Cloud to your Slack and any team member can be having conversations with it throughout the day. It can see into all the channels. So you can pull it into a contextual channel where you're talking about certain types of work. And over the course of doing all that, it can see all the messages, all the interactions that are happening, but still on the back end of Cloud, it can be connected to all the other places that you work. And this can make it a tremendously useful assistant, especially when it's pulled into the context of the interactions you're having with your team about client projects. I mean, imagine this thing being able to see not only your emails, but your teammates' emails as you're talking about what to do with this client that has been out of shape. And but it can also see into like the status of that client's work and things that were discussed three years ago when it has access to all that context, but also access to what are the humans and my company trying to do, what are they talking about? You could see how that becomes a very, very powerful thing, right? Anthropic actually said it was something like 65% of the code that they write within the company is now written from Claude Inslaque, which the first time I heard that, I'm like, what does that even mean? Like, nobody's writing code in Slack. It is people having conversations inside of Slack and then telling Claude to go out and do stuff. And that is how it ultimately makes the decisions of what to go out and do is the humans interacting inside of Slack and coming to a decision. And so if you can imagine Claude being embedded inside the messaging app of every company, you start thinking about this a little more and you're like, okay, so it'd be one thing if it was just messaging, but for it to be useful, it's got to be connected on the back and to our other services, right? So you can see all the work that is happening, but also see all the humans communicating and coordinating things. You think about this and you're like, okay, this is a lot of trust to put in an AI company for it to just be able to see. I mean, there are hundreds if not thousands of accounting firms that already have this installed. I have no doubt. And Claude is seeing into how is all of the work getting done across all these different firms. And to be clear, again, it's still a business plan. The prompts are not trained into the model, but what this Claude is building, like within that, within your account is memories, is receipts. Have we done this stuff in the past? What are people talking about? So that over time, it continues to get better and better and can work even more autonomously when you're like, hey, we're having this conversation for the 30th time, can you go do that thing again? It becomes this tremendously valuable thing, right? That feels much more like a hyperproductive human that is working to support your team. But let's follow this through like just a couple of beats more. Because we say so often on this podcast that like the tools that we use in the future, they need to have really good MCP connections. Like everybody, I mean, there's a lot of people that are like, well, Claude, AI, it's just gonna replace the software that I use. And it's like some apps for sure. Like, I mean, it's already replaced some apps in my life. But the core systems, that's maybe harder to get your head around. Your practice management system, your accounting ledger. But in those cases, you still want a really good MCP connector so that your AI assistant can see into all of those systems and the very best type of MCP connectors are read, write, connect. So it can not only read information from the platform, but write information to the platform. Modify the status of a project, classify an accounting transaction. We're talking about a Claude that in the very near future ought to have all these capabilities across all the different tools that we use. And then we now embed it and where all of our humans are having these conversations. And to me, that brings it way closer to this thing that kind of has its arms around the whole company for good or for bad. And I guess this is the downside of having something that's tremendously, tremendously useful, right? I would say many of us like, we're already, We've already waited into like a a a degree of reliance on Claude. There's a lot of people now, I mean, Claude, there's a couple of weeks ago, we had a Claude was down for like eight hours, and for a lot of folks, it's like a snow day. It's like, if I don't have Claude, like how am I gonna get anything done? And if you're a software engineer, like you can totally understand that. I guess I'll go check the mailbox. I guess I'll go scoop the litter box, Claude's down, so I will go take care of whatever admin I have, because surely I can't work with it with Claude down, right? And we already have versions of this with, you know, QuickBooks having outages or just humongous outage around tax season. But this is probably the first time with like a launch from OpenAI or Anthropic that I sort of was like, all right, I could see this feature actually being like, Anthropics backdoor into like pretty much every company. How do you feel about that? I don't know if it is the level of access we're giving it that makes me uncomfortable, or the fact that I can see the path to it being insanely useful. Like, remember, these folks talk about these AI tools as things that will in the near future not just be assistance to humans, but will literally be entire organizations. Like imagine an organization with an org chart where there's not a single human inside of that organization, like completely self-managing. And we're now open in the back door and saying, "Okay, come on into Slack. You're gonna be here to support all of our humans, but we're also gonna give you access to all the other tools that we use, at least anyone that will give us an MCP so that you can go and do our bidding and be incredibly helpful on the back end." I think what probably spooks me about that, more so than like the privacy thing is just the fact that I can see that being really, really, really good. And when I say good, I mean useful. Like at getting work done and being helpful to people in the team, which isn't this what we've always wanted. Maybe I'm just coming to terms with like I can see the path to it being the sort of thing that's like, this is kind of running the firm at this point. And so it is like, I don't know, my kids are young, but it's like taking the card of the parking lot, when your kid gets a permit and you're gonna sit next to him and show him how to drive now. And there's a lot of you that doesn't like that, but there's a certain inevitability to it as well. And I think that's maybe what I'm processing in real time. Now when it comes to like the software in our domain, what we often do next is we always say, oh, that, no, that is interesting. I do understand that, but I'm gonna wait until X platform in my space launches that, or does the same thing, right? Like we very quickly, there's some folks that never jump there that are like, no, chat cheap until the end, clawed till the end, try to look to the end, click up till the end. But most of us, like we see a cool new feature and we're like, all right, this is great. I see the value of it, but I'm gonna wait until, you know, my practice management system launches this or something like that. And so does it really make sense that Slack, or Microsoft Teams or wherever it ends up being connected? Does it really make sense for that to be the home? For what governs how all the AI stuff in your company is happening? My initial reaction was like, no, because Slack and team communication is kind of this side thing to the main work that's happening where the work is managed, right? Thing is, we're talking about a world where the doing of the work is reduced to very little human effort. And in that timeline, like what does that leave? Well, all the human communication, all the messaging. And so right now, if you look at a practice management system that doesn't have like built in messaging, and you say, I'm gonna wait until my PM rolls out this assistant. I don't know that it can, because it doesn't have all the context to water all the humans on your team talking about and working on. Some PMs do more than others. But where my initial reaction was, if anything, like I've always thought that Slack connections were a little bit distracting, 'cause it's like, no, I don't want you talking about that and Slack, I want you logging all that information in the actual system, like in the source of truth, because Slack is not the source of truth. Now, the more I think about it, I'm like, well, if I could do the work, Slack is what remains. It is the human conversations that actually remain. And so why wouldn't it live there? I've seen more commentary around this in the last week, where people for the first time are like, this might be too far. I don't know that I'm ready for this. Whereas there's other people like throwing their doors open that are like, this is amazing. This is gonna be incredibly helpful for us. I totally see it. Now, the next time we're working on something, you can literally just take a clot and it can come and help and get that thing done. What, why should we be scared when it's sort of the most capable, most helpful version of this thing that we've had yet, right? So spooky. It's spooky from like a level of access standpoint and just how powerful that could become standpoint. Especially when you consider a company like Anthropic, a huge percentage of what the company does is simply managed from Slack now. The interactions you're having with your teammates, telling this Slack agent to go off and do your bidding and that stuff just happening. Powerful, but I think there is a part of me that might be grieving. That is like, well, I don't know that I want that. What about maybe I want to bang those rocks together myself and not just let Cloud go off and do it. Strange times. Now, writing software, which is, you know, what Anthropic primarily does, it is the domain that has been most eroded, maybe not the right word to use, but most impacted by AI. Like virtually nobody writes code anymore and does that translate to accounting and what accounting firms do? Not really, but it is maybe an early indicator, an industry that is out ahead of most industries in terms of how AI has disrupted it. So it's oftentimes a good place to look for like, what are the things that we ought to be watching out for ourselves? (upbeat music) Are you still running your accounting for a monospread sheet? (grunts) Don't do that, man. That's gross. Financial sense, it is the practice management system for small and growing teams. Think 20 employees and under. It is everything you need to get your work done all in one place and they've been chipped into some cool stuff lately, like built-in month and close automation, this deep integration into the accounting ledger to actually pull all the transactions in platform and do that sort of quality assurance pass from inside up financial sense. Now they also recently launched a campaign called proudly small. Because being small isn't something to apologize for, it's something to be proud of. And just to be clear, we are talking about your accounting firm. Wouldn't it be nice to have a tool that's actually built for small firms? That's actually a very common complaint right now, is tools are increasing in cost. They won't come down to support firms beyond like a certain number of clients. Financial sense, they are built for you. Plus, if you use code Jason10, it will make the price of financial sense even smaller for your small firm. You'll get 10% off the annual team or scale plans with two or more users. I'll put the link to claim that down in the show notes. Check that out, see financial sense might be for you. Hey, if you're done licking your wounds from tax season, if you run more than a few hundred 1040s a year, you gotta look at this episode sponsor, Sorban. One of my leaders in tax workflow automation can save you a tremendous amount of time right now, specifically on the intake part of 1040 workflows, but they're also working on connecting to your tax software automatically, like pushing data in there. They're working on automating the delivery of the tax return. But this is like the fastest growing category in tax tech. If you're a US tax firm, you absolutely need to have a good tax workflow solution nailed down. We did about 1,840s a year in my accounting firm. This would have saved us so much chasing clients for documents and just tedious organizing of all these PDFs. Oh my gosh, can't say enough about Sorban. We actually did a full demo of their platform on my YouTube channel not so long ago if you wanna see it. But to learn more about Sorban, take out the link down in the show notes. In practice right now, we are blocked by the quality of the connections that we have to the tools in our space to our practice management system to the accounting ledger. Don't give me started on tax software. And that is ultimately what is gonna hold us up from getting a really, really useful version of this. But I'll say this again, as I've said it many times. Right now, if you put two tools in front of an accountant that look like they do the same thing and one of those tools comes with a really, really good MCP connection, you're gonna go with the MCP enabled tool. Because I can go to cloud, I mean, we talk about kick the accounting ledger that has a really good MCP. I can literally go to cloud with 12 months of bank statements and say, go do the accounting and kick. And that's a really, really compelling way to just get work done right now. And if every single one of my clients was on a ledger that was well supported by the MCP and I can just have an interaction with cloud to move all these different projects along, what a way to work, man. Like you are the puppet master. I had a moment over the weekend doing something with cloud and it spun loose like 63 agents simultaneously that we're working on this thing. And it was a little bit of a yikers. And it was actually amazing. It like confirmed that I really wanted it to do this before it did and then consumed all of my five hours of usage in like three minutes. I could have kept going and like paid for additional usage. That would have gotten so expensive so fast. But it was basically this huge conversion of this icon set that we use for all of our videos and stuff like that is a big lift project. And I watched it just knock this out in just a matter of minutes and it was like, holy shimole. These armies of agents, doing accounting work, doing tax work, gonna be wild, man. It's gonna be a weird time. But we're cranking really hard on AI use case videos for the main YouTube channel right now 'cause I think more people just need to see like, how do I do this? How do I do that? And just try to keep the conversation on the doing and what do I do right now? The more existential you get the harder it is and the easier it is to be paralyzed by like, I don't even know what to do next right now. And so on the videos, we're pushing really hard into just how to make really useful things and keeping the videos focused on that. Excited to get more of those out into the wild. And to that end, we recently put together a list of how many was it? 127 AI use cases for accounting firms. These are all use cases that were sourced from accounting firms that join me on on my tour that just wrapped up. We had over 1500 accounting firms out hanging out. And I wanna show you three really, really cool. Like some of my favorites here use cases that I think are no brainers. for firms. So three really cool use cases I want to show you today, but you will get the full list of 127 use cases if you are on my newsletter this week. If you're not jsononfirms.com/newsletter, totally free to sign up. Some really really cool stuff here. Use case number one. Something proposals, this is like new client proposals from meeting transcripts. Maybe you have all these conversations with the client, right? And it is for whatever reason such a project to get to a proposal that you will send to them. I think it's a lot of it's because it's kind of an emotionally taxing thing, right? And you're thinking, what are they thinking? And what if they say no? And how badly do I really want this client? A lot of things that maybe aren't captured in a meeting transcript that you will still have to lose sleep over. Imagine a world where your firm has a clear set of things that you sell and past examples of onboarding meetings with clients, sales meetings with clients where you ultimately sold them on X, Y and Z and all that context couldn't form. Hey, I just got off a meeting with a client. Can you draft a proposal for me? That is a huge, huge time saver. How do you do that in practice? What is the AI need? Obviously it needs the meeting transcripts. Remember if you're meeting in person, most of these meeting transcript tools have mobile apps that you can use to record the conversation in person. Many of them also have voice over IP integration so that you can record the phone call as well. Only thing is with phone calls specifically, at least in the US state by state, you got to watch out for rules specific to recordings. But the AI would need all that context, the communications themselves. It would also need the context of whatever the things that you sell in your accounting firm. And this is probably going to be the biggest blocker for accounting firms, getting to a really cool version of this sort of transcript to proposal. Now two years ago, I put out maybe the most useful resource we've ever developed. It was called our service library template. And we actually spent a whole week on the podcast on this while I was driving across the country. If you go to the YouTube version of this podcast and go to playlists, you will see a whole like four or five episodes, episode playlist going through building a service library. But the idea behind this all is basically to take all the myriad of different things that you do for people in your accounting firm that you just started doing because they asked you to do it. And corral all of that into a sort of explicit list of the things that we do for people and what level we do those things at. So a proposal to a client will look like a collection of services like bookkeeping to tax prep three payroll processing one. And all of these different services have different kind of numbers that indicate a service level. And with those different things, there is an internal understanding of exactly what that means and what you will do for that client. This is not a language that we share externally with our clients. It is a language that we only use internally with the team so that everybody has a clear understanding of of the client's expectation of what we sold to them and what work we're going to do for them. The goal is to take all these different disparate sort of projects and in a service business, these engagements kind of the natural state of things as they will all get more different over time because clients will ask for all these different things. The goal is to create an internal language around here's the stuff that we do because once you have that, your projects will actually start trending to be more similar over time because the team internally has agreed on here's what X, Y, or Z means. And so the next time the engagement is changing, you can kind of steer it in a direction that brings it back to where you want it to be. But because the reality is clients still want weird stuff, you will still then have this sort of catch all around that of any unique to that client things that you're doing. But the goal is to make 80% of the engagement standardized and 20% bespoke, right? Rather than 100% being bespoke, every engagement being unique to that client, which is how most of us run our accounting firms because we have no like internal agreed upon language for what work we do for different people. I'll put a link to the YouTube video, this kind of the gateway into this concept and the show notes here. What we ended up building is like a 16 tab Excel spreadsheet of the most common services accounting firm cell and how I would recommend building that service library that you can swipe totally for free. But the beauty is once you have that service library internally, that language for what are the different things that we sell to people. It makes standardizing how you sell work so much easier and using AI for this, creating a proposal that sits on top of all those things that you know that you do for other people. I mean, this could literally look into your other clients and like what level of service am I giving to other clients? Who is this client most similar to and even build a proposal that way? But prerequisite here where I think where this falls down for most firms is do we like have any sort of standardization on what it is that we're actually delivering to people so that the AI even knows like, well, here's what we're in the business of doing. Second really cool use case that we got from a firm, we use cloud coer to build tax return summaries that are presented to clients to help give them a better understanding of their tax return. So we're talking tax here, this could be a set of financial statements could be literally anything using AI to build what this is is basically custom reporting reporting that your client will actually understand the problem with US tax platforms as they produce this very arcane output that is more made for tax professionals. And when you just print that out in the package that goes to your client, your client will never have any understanding of that because there's just so much vernacular that they will never understand baked into it. But imagine having your own version of this and we've got an entire industry of tools that are quote unquote reporting tools that connect to an accounting ledger and they create this visually kind of fanciful way of presenting information to a client. And I think the best versions of those are when they're distilling a whole bunch of data down into like a smaller set of KPIs that can actually feel actionable to the business owner. And you could certainly build your own version of one of these if you wanted. But like the most interesting version of this incorporates like the value of the engagement itself into the conversation. So to use a tax example, if you had a planning meeting with the client a few months earlier and you were talking about one specific decision, I want to see in that report not only the tax summary, like a high level of what happened in the return, but also what was the result of all the things that we talked about. So can we look past across past like meeting transcripts and any strategic stuff that we did to kind of see the through line of yes, we're still doing this. And here's the overall impact of it to communicate like yes, this work product is done. But here's how we're delivering value to you. That's a tricky thing for us to do the whole communication of why what we are doing is valuable. It's the difference between selling something that's commoditized like, oh, you do bookkeeping. I do bookkeeping too. How much do you pay your guy for bookkeeping? I pay this. Wait, what? It's the difference between getting sucked into that game versus like building an actual plan for people and them understanding like, hey, here's here's where we are driving profitability. Here's where we saved money. Here's where we deferred tax like actually being able to confidently communicate that not only at a snapshot in time, but the cumulative like over time difference that you've been able to make. I think some of us struggle with like, well, I they should just know. I don't want to have to tell them like that feels there's like this sort of like virtue signaling like we don't want to advocate for ourselves or what we're doing or something like that. But it's like, if we don't, man, like who will the client doesn't have the same understanding of this stuff that we do. And you now have this magical tool that will let you present that information in any format that you want to. How cool is that? Now in practice, what is the best way to build this? If you haven't used cloud design yet, it is so, so cool. Place to start. I did a pretty short video on cloud design. Check that out. I re-built like a whole accounting firm website. Just you just YouTube search my name and cloud design. But if you're building say a reporting package, just could even just be a one-pager, the place to start is probably cloud design to talk through like what ought to be visualized there. And cloud design is definitely the best at making something that just visually looks good. But when you finish the mock up there, you build the template there. You can then export that over to cloud and probably build that into a skill so that you will come to then cloud, not cloud design, but come to cloud with, you know, copy of the tax return summary or set of financial statements. And it then takes the data from that report that you gave it and pulls it into the reporting template. So, build the initial design, the template, in-clawed design, but ultimately pull that template over to cloud. And then you can create a skill that not only you can use, but your entire team can use. And for them, it's then just a matter of like tossing an tax return summary and saying, use this skill to make a report, double check the output, make sure it looks correct. Bob Drunkle got a really cool looking report. Third and final use case here. This one is pretty wild. We use AI to create an importable trial balance for Drake by pulling in the QuickBooks trial balance and then dropping our AGEs and FGEs directly into cloud. That was so much accounting speak. But if you're on the inside, you're excited right now. Now, if you don't do work in the land of the free, the land of the brave, the Lord's country, if you don't work in the US, you may not understand why do you need to stand alone trial balance tool. This is a thing that is long been in the tool gets of most tax firms because the financials that are ultimately like in the tax filing are not the same as the book financials. And so there's a whole process of converting the book financials to the tax figures. And so you somehow need this tool that sits on top of the accounting ledger that will let you do those additional adjustments. And so you have these dedicated trial balance tools, the OG, the arguably probably next to QuickBooks desktop. The second app on that Mount Rushmore, it is ATB, Accountants, trial balance. Think about that. What else should be on that Mount Rushmore? Yeah, QuickBooks desktop. You get ATB. Some people will argue quick and but I feel like quick and walked so that QuickBooks desktop could fly. That's not the expression. I don't remember what the expression is. Tell me what other two faces should be on that Mount Rushmore. And again, if you're not in the land of the free, the Lord's country, Mount Rushmore is a surprisingly small monument that is built in North Dakota and the grand scheme that is not small. But when you see it in person, you're like, "Huh, I don't know why, but I always thought that'd be bigger." And it has four old presidents faces on it. Just a matter of time before Trump puts the fifth face on that. Oh, would that be a Trump thing to do? But what should accounting apps three and four be on that Mount Rushmore? AI hasn't quite replaced accountants yet. But if there is a part of accounting that is in a endangered state, it is accounts payable. It is the AP department and the company that I think is taking AI the farthest right now, specifically inside of AP, it is makers hub. They have taken automated extraction from invoices and receipts farther than I've ever seen anybody take it. But have followed that through all the way to like complex approval processes. You know, like when you need a delineator that is more than is bill over $X dollars or under $Y dollars. When there's a little more complexity to it than that, makersubs got to cover the handle payments as well. They're doing for an exchange. Stuff they are doing. Canada, stuff. It is an elegant solution to a problem that I think we all wish didn't exist. In the first place, when you have ugly bill pay needs be it from project tracking inventory management, whatever that is. And the normal tools, they're just not cutting it, not getting the job done. Maker sub worth checking out. Put an ad work on the most hairy accounts payable workflows. Learn more about Maker sub, check out the link down in the show notes. Text selection season is upon us, accounting friends. This is the time of the year. If you're going to do any big pulling up of the carpet, big fundamental changes in core tech in your accounting firm, you want to do it right now in these next couple of months. And the most important carpet you pick in your accounting firm, it is your practice management system. Today's sponsor, Canopy, they've been sitting at the tipi top of my PM recommendations ever since I have been recommending apps. Always kind of been a feature nerd's dream, Canopy. But if you haven't seen it in the last 12 months, they've launched a bunch of new stuff from AI Taxant Take to built-in, month and clothes, automation. To now, recently, they launched Canopy Co-Work Urr, a built-in AI assistant to make your wildest dreams come true. And they're going to give you $75. Just for pop them by for a demo to see what's changed. Head over to getcanopy.com/json. Book a demo just to see the new stuff. But send you $75. Get card afterwards. Okay, that's all you got to do. You just got paid for listening to an ad read, man. Go do it. I'll put that link down in the show notes. Anyways, in the US, we have to use these trial balance apps. And some of us don't we like budget together different ways of spreadsheet, something like that. What this firm is saying is that they use Claude as they're not only the trial balance app, but they use it to then create a file format that can be imported directly into the tax software. This is the triple-saukel of business tax prep AI use cases. This is so cool. But you hear that and you just you have more questions. I don't know how they're doing it exactly. I can tell you how I would do it. You might have heard of artifacts. Artifacts are like little apps sort of built into Claude. You can go to Claude and say build an artifact to do XYZ. And then that app that it builds will live up in the top left of your Claude. And the most common place where I see people using this is building kind of like a personal dashboard that connects your email and maybe some other things. And you've got sort of all this stuff in one view inside of this artifact. But where artifacts are limited is artifacts don't really have much like of a back end and a little environment that they live within. They can pull like live data via APIs, but that's kind of it. And so when it comes to doing like heavy lifting work, artifacts are not really the place for that. But did you know that skills Claude skills regular old Claude skills can have entire applications packaged inside of them so that the first time somebody uses the skill a handy dandy little app slides out from the right hand side. No artifacts required. You can have entire apps inside of a skill that just work alongside a Claude conversation for what they're talking about here with this sort of trial balancing. That is what I would expect or that is probably how I would build this is I would build it into a skill. I want you to be this trial balance utility for me. I'll give you the opening trial balance. I'll give you any journal entries. You can define the journal entry types like a J E's FJ E's RGE's. And then as part of the skill, it sets how that trial balance is visualized over on the side of the conversation. And in this case, also how that file is formatted when it's exported to be then taken over to the tax software. Really cool. If that sounds intimidating, like I can't even imagine how I'd begin to build that, you'd be surprised how far you can get just going to Claude and like rambling at length about here's kind of what I need and what it should look like. Wherever possible, reference versions of it that already exist. And so if you can see an app online that does this or screenshots of what you expect it to look like, give it all of that stuff. That's like the most efficient way to kind of align it with your expectations. Wasn't that long ago? We used T value in our, ooh, should this go around rest more? T value? How about mutual fund tax guide? Anyways, we used T value in our accounting firm. It was this app that literally just calculated amortization schedules. What a time to be alive. I wanted to just build a super basic amortization schedule builder with Claude and I pointed it to T value. I'm like, go look at T value, look at all the screenshots, see what it does. That's more or less what I'm looking for, but like a more paired back version of that. Anytime you're trying to get Claude to build some sort of app for you, if you can point it to a reference or a combination of this and that, it'll make it much easier for it to kind of guess what exactly it is that you want out of it. Super cool stuff. I'm like constantly blown away with creativity of accountants, no less. And the ways that they're using AI super cool. And honestly, it's great ammunition for me to just go out and make a whole ton of AI use case videos that people will come and watch and comment and be like, wow Jason, you're so creative. How did you think of this cool use case? What a talented creator you are. That's my business model, man. Honestly, I have all these conversations with accountants that you don't have time to have because you have a job. And then I take all those best ideas and I regurgitate them and people are like, wow, that guy. How does he do it? It's like a new idea every day. I know, but that's the business I've built for myself now and it's mine. All right, we got some questions to get to. Let's get the old mailbag out. Mailbag! Mailbag! The segment is called Mailbag! Drive my daughter to tennis camp this morning and we had a big storm last night and a whole bunch of people had at the curb these big brown paper bags full of garbage. I did not know that garbage bags could not be plastic. I mean, it makes sense, right? Plastics is not the best thing. Like humongous look kind of like waxed papery garbage bags almost the size of a garbage can. I mean, but they'd been blown just all over the neighborhood. Who knew like whose bag was whose? At this point, we're all just trying to crawl them back out of the road. But if you use a paper garbage bag and I'm not talking about like paper grocery store bags, did I say bags? I meant bags. I don't need people measuring how I say bag, okay? But if you use these giant garbage paper bags, leave a comment. What are the pros and cons? Obviously, environmentally they make a lot of sense, but I saw them and I was like, "Mmm." That actually makes a ton of sense and they got a little bit of heft to them, a little fortitude. Like it can free stand. I'm going to talk about a weakness of a plastic garbage bag. Get one of those things to stand up. I'll wait, because it's not going to happen. Mailbag questions are sourced from social media DMs, from email from my cellular telephone. Text me 1516-980-4968. First question. Do you foresee a time when QBO or zero isn't the default choice for new businesses? Do I foresee a time when this will happen for sure? Obviously, this will happen eventually, right? We're going to go to space. We're going to go to other plants. The sun's supposed to like flame out at some point. So of course, there will be a day where there is no QuickBooks or zero. I think what you're asking is, "Can this happen sooner, please?" Honestly, I think it's happening. I think it already is happening in a number of different ways. Number one, you've got new AI alternative ledgers of which the leaders are digits and kick. And if you're not in the land of the free, then I'm sorry that you can't get access to kick and digits yet. But we'll give it to you at some point. These are being picked up, I think, by the folks that are like, "I want a tool that'll work better alongside of my AI assistant." That's who these ledgers are really attractive to. But obviously, that's not trickling down to the normy business owner. So you will see more change in what tools are being used among pro users long before you see it in normy users. I mean, there's a lot of like avocado toast US accountants that went to zero years ago. I was one of them and they're like, "Oh, this is the future. This is way better." And it wasn't. It was just something that was different than QuickBooks. There was a time when it was better than QuickBooks online, but it's not anymore. It's just where you go when you don't want to give into it your money. Into it. Really like the plastic garbage bag of the industry. Being untold damage to the environment, you wouldn't trust them around small children. But there's a subset of normies that are AI normies. And when I say normies, I mean non-accounts. And honestly, for AI normies, I mean, I think your AI tool of choice, Chatchy PT Claude, is probably the leading alternative soon if it's not already. In fact, I was talking with someone today who was like, "Man, AI's gotten so good. My dad can ask for help with something and I'll just send them the stuff out of Chatchy PT and it does what he was paying Bonnie $62,000 a year for." And I'm like, "I feel like we've maybe slipped past a little bit of nuance there." But I mean, people already trust AI to be capable of. of more things than maybe it is oftentimes. And so for a non-technical user to go to chat, GPT and be like, huh, can you be my accounting ledger? What's chat, GPT gonna say? No, no, sorry, keep scrolling. Of course it's gonna be like, I would love to. Can I go download this in that library and then I will be your everything? So I think the biggest competition to QuickBooks in zero right now is honestly, probably not other ledger platforms. It's probably just generalist AI tools getting really, really good. And there's a world where this actually hurts the people building the AI ledgers because those are kind of the same people, right? Like we did a YouTube video not that long ago about how to replace QuickBooks with Cloud because Cloud can do simple books very well. And there's a world where the people who will buy into that and actually use Cloud 2 do, that would have been the same people that would have bought the AI ledgers. Now right now, it's still a pretty solid case for like more complex books. It is nice to have a ledger, but I do still want my AI assistant to be able to talk with the ledger. So do I foresee a time when this happens? I do. I do not think QBO is analogous to a plastic trash bag which has a half life of, I don't know, millions of years. I'm pretty sure. Cloud coworker has got a lot of attention versus Cloud chat. I'm finding I like chat more in terms of creating new skills, formatting documents. I can give it the instructions on what I wanted to get done and then go do other things like Shuffle and Kids to Sports. Oh, don't give me started on kid sports. But then when I come back, I got a new skill that I will use over and over. If I were to do it in cowork, I'd be changed to my laptop. What am I missing? Okay, so Cloud coworker, that stuff only lives on your local computer, but Cloud chat, that stuff syncs across your computer to your phone to anywhere that you work. Those chat conversations live. This person says, if I were to do it in cowork, I'd be chained to my laptop. What am I missing? First things first, I'm concerned for your laptop because the whole idea of a laptop is its mobile. My mind goes to like a weird sort of dungeon where you use your laptop. When's the last time you left your laptop, touch the little grass, huh? But this is true. Chat doesn't come with that same limitation. I will say the limitation of chat mainly is that it can't work on your local files at your computer, at the battle station where you're doing most of your heavy lifting. And actually chat doesn't actually get as much compute allowance to do big tasks. So chat generally is not gonna spend 30 minutes working on something for you. Whereas in cowork, like that's the whole idea behind cowork is like run with some big meaty things that we need to sort of progress. So I am, and maybe I'm a bad person to ask about this because I've always been very particular about my battle station. There is nothing like being at the command center and it's very hard for me to go and work other places and be productive on a laptop. And part of that stems from being an accountant you're used to having multiple monitors and all that. But I would say there are two different things. Also, the skills that you build with chat can still be brought over into cowork. But for most people, we're doing the bulk of our production work, I would say, from our main machines, not from our phone, not like hopefully not waiting for soccer practice to end. But even then, if I'm like idle, like if I am at, I had a phase of life where I was hanging out at gymnastics a couple hours a week. You wanna talk about smells, man? Ooh, chalk in the air. You ever accidentally huff a little bit of powder sugar off your French toast? Gymnastics, standing, you're walking around. It's like, you're constantly, there's so much chalk around. You're just like, you're right on the edge of having to do that cough. But I would bring my laptop. And so if you are stuck for like a more prolonged period doing this stuff, like you're back to the laptop then, right? And then you can use cowork. (upbeat music) Have you heard about Chi? Chi is the new AI coworker built into this episode sponsor, Carbon, Carbon's a practice management system, a brain, the backbone of your firm, but Chi, let me tell you about Chi. Chi's like the cute new guy in the office. Chi, about Chi, oh my gosh. What can Chi do? Chi can pull from client history, from your actual workflows, from client communications. Chi will handle repetitive tasks, surface client insights, help you plan a priori-touch your day, what a multi-threat. Chi will include a mobile app, what? And with Carbon, there's even more cool AI functionality coming just around the corner, including an AI meeting note taker, an MCP, an MCP, rejoice. Carbon is cranking on AI stuff. You love to see it, man. Can't wait to see more about that MCP as well. To learn more about Carbon and their new AI assistant, Chi, check out the link down in the show notes. Listen, I want to talk to you about your IT, okay? Do you have IT help that really understands you? Unstands what accounting firms do, the rules specific to firms, or are you trying to do it yourself? 'Cause you're an IT expert. You're not. You're just not. Or maybe you hired the guy down the street? How are you enjoying paying him to learn how these awful, or paying the IT guy down the street? How's that going? Paying him to learn about this software he used in an accounting firm gang, there's a better way, okay? That better way? Be-e-r-i-t-o, there it is. I talk a lot about like my private community realize and how you need a place to go to just ask questions. True story, if you go to my community right now and you search the word, there it is, you will see a long list of people that are like, this is a no-brainer, it's really good. Verdo will do your hosting, Verdo will be your entire IT department and get this. Verdo will now help you deploy Claude inside of your Verdo hosted environment. That's right buddy. I'm afraid about like what's okay to use AI with and what isn't. Imagine your IT company taking the lead and being like, hey, this is the right way to do this. Here you go, here's an environment where you can cut loosen and make your wildest dreams come true. That is the sort of proactivity I want out of my IT company. Verdo, they only work with accounting firms so they know the rules, they know the right way to do this brother. Stop fistfight near IT help and go with the obvious answer. That is Verdo. If you learn more, check out the link down the show notes. Kind of interestingly, a company we've done some sponsorship stuff with just to disclose Verdo, an IT group for accounting firms. In fact, I think we actually may have a sponsor role on this episode from them. But one thing they're touting now is they will let you set up a cloud on your virtual desktop. You'd like your host a desktop that you use from Verdo which you could then log into from anywhere. And so you're actually remoting into the same desktop where you're same cloud instances from any machine. You still have the limitation of like, you'd have to be remoting into it from a proper computer, a laptop or a desktop workstation. Wouldn't work from your phone. And if it did, that would be terrible. I don't recommend that. But yes, cloud chat can still do some really cool stuff. But, co-work, man. Not the beats co-work. Have you read any of Alex Hermosi's $100 million books? If so, what book would you say is the most impactful for firm owners? If you haven't read any of them, read them. Yes, I've read them all. I have read the little bonus workbooks that came with a $100 million money models. People sleep on Alex Hermosi stuff. His, if you do, I can get like his, he can be a lot in like his social media content. He's just hard to take seriously. Like, I don't know. He's a bit of a walking meme. The guy ate Chipotle for a year. I can't imagine what that would do to a person. Maybe we would all be like that if we ate Chipotle for a year. I ate Chipotle until Wednesday. And over the counter strength is not enough. I got a call on the audible on the whole rest of my week. But if you had to read one Alex Hermosi book, read the first $100 million one, $100 million offers. At the time he wrote this, he had virtually no social media following, but it's so good. Like, there's so many things to be taken from it around. How to better frame what we do to be enticing to people. But also the writing. Pay attention to the writing itself. It's written at a fifth grade level. And it's just reduced to be so easily understood. And at this, we could do such a better job of this. I'm not talking over the top of people. I mean, it's really, really like intelligent stuff. If you can see through sort of the broy-ness and kind of the culture around his business and all that. But yes, I love the Hermosi stuff. It's really, really good. I would say there's a few thought leaders in our space now that are definitely born of the Hermosi ification of like kind of the business model and all that, which is fun to see. In Clawed, how do you use projects? Do you use projects as like a department or a project for each client? Great question. So projects inside of Clawed, they live up on the left-hand side. If you are in chat, projects can be shared with the rest of your team. If you're in co-work, projects cannot be shared with the rest of your team because that stuff only lives on your local computer. Don't use projects as either of these things, a department or client. So don't have a project for a tax production or bookkeeping. Don't have projects for client 1, 2, 3, 4, John Smith, John's car wash. Don't do that. I don't want to use Clawed as a storage place. That is not the job of Clawed. Use connections for that. There ought to be a home for where all that information lives that is not inside of Clawed. Clawed's job is to sit on top of that stuff. So for example, I would never have a project that is John Smith. I would have a Google Drive folder or a canopy folder or something like that, a SharePoint folder where all that stuff ought to live and then Clawed connects to that information. Projects are actually not a super heavy user of. It's mainly for a femoral like one-off projects where it's like, hey, we're doing this thing over the next 10 days. Let's pull four people into this project. Well, how's everything right here? And we will work on this. For that, it is useful. But it is not for the long-term storage of any sort of information or even context about the firm. Park all that stuff other places and then connect Clawed to where that stuff lives. Don't use cloud as a home for any of that data. Is chat GPT dead? Will we be seeing a swing? A swing back to you being in love with it before the year ends. Would you quit acting like I swing back and forth between chat GPT and cloud? I don't. I told people for three years from when chat GPT came out. I said, "Use chat GPT. Don't worry about anything else." And along the way, there were other tools that were better very briefly, but chat GPT was the only one that was at the top or very near the top that whole time. This last December, all of that changed when cloud launched some very good new AI models and nailed the harness. The harness, that is, the thing that deploys all these sort of sub agents to get the work done, and it took a step way past chat GPT in a way that nobody expected. Anthropic is now worth more money than OpenAI. Nobody would have ever guessed that would have been the case 12 months ago. There were light years behind OpenAI, but now cloud is just better. Also, it's what everybody's using. So like, maybe the strongest argument now for why you should be using cloud instead of chat GPT is cloud is what all of your friends are using. And we don't talk about this enough inside of accounting firms. How much value there is in using the same things that your friends use so that you're a text message or what's that message, a quick call away from like, "Hey, how you doing x, y or z?" If they're using all different stuff than you are, it's very hard to get answers. Now, we'll chat GPT swing back to being the best by the end of the year. Right now looks pretty unlikely, but I will say we have not made a habit of swinging back and forth. For three years, we set chat GPT now since the beginning of the year. We've been saying cloud. And for just to not have somebody share this clip with me later, I hope that cloud is the best until the end of time. Alright? We are a small fractional controller slash fractional CFO firm. We are opening a bookkeeping division that tail as old as time. Hey, let's do fractional CFO work. 60 days later. Okay, so it looks like we're going to have to get into bookkeeping. I'd love to hear more advice on the approach to new client onboarding and cleanup. How do you charge for that? We're not sure if we're doing it the best way still in the learning phase. Alright, whether you do bookkeeping work or tax work, you got to have a process to get new clients's stuff into the state that will work with sort of your ongoing systems. Now, a lot of debate around how do you price this because you don't really know what's there until you have pulled back all of the layers and you see, oh my gosh, there's a body over here. There's a body over there. Why did I just quote it up from fee for this? So three different ways you can do this. You can just go all hourly. This is good for you. It's bad for the customer because the customer has their own idea of how much time this should take. After all, I mean, their mom was doing the books before you, right? So surely you're not going to send them a $6,000 bill. So first approach is hourly. Good for you, bad for the client. Second approach is fixed fee. Great for the client. Risky for you. Third approach is give them a range. And I don't think we talk about ranges enough. When I rolled out fixed upfront pricing for our tax work in my firm, we had a category of like big projects that were hard to nail down a specific price for. And so instead we gave them a range. And they would sign off on that range, saying anything within this range was fine and we had permission to pull payment for a number within that range as soon as the project was done. And people didn't mind that. So I probably wouldn't do hourly just because there's a lot of risk that comes with hourly. Even if you do hourly, the client's going to be like, okay, what's your hourly rate? And you're like $650 an hour and they're like, boom, the next question is, well, how many hours is it going to be? And so these still have to like give them something that is almost a fixed price because they're going to ask how many hours it will be. So for me, I just prefer giving them a range. Now some folks sidestep sort of dodged this problem altogether and will actually in the case of bookkeeping work do a full months bookkeeping before they even do the cleanup. And this is for good reason because cleanup work can be frustrating. It can be this long sort of drawn out painful thing. It's like the worst part of the customer experience. And you may not want that to be somebody's first impression. Since the argument for actually just doing a month's accounting and that being where you start. Now, the accounting purists are like, well, you can't do that. Like what about the balance you? What about all the stuff that's wrong? And so for sure this comes with some trade-offs. But if you're optimizing for the client experience, if there's a pretty good argument that this is the best way to start with someone new, I learned the lesson the hard way in building our cast practice that it is okay if everything isn't perfect by the first month because we are taking something that was just very wrong and not right at all and getting it to a place where it will eventually be correct. If it's your end reporting, that's one thing. But in a month over month books, I ultimately got to the point where it's like, I would tell the client, hey, we're going to, it's going to be a three to six month process to get things totally dialed where I'm really happy with them. Now oftentimes the client coming to us, they were pretty happy with how it was and they like, they didn't even realize it's something needed to change. But what I fell on my face trying to do was make month one perfect and then it would take 70 days to get the first month out and it's just this big, frustrating experience for the client initially. So to me, giving a range is the best of both worlds. But also consider like, is there a way that I can give them a quick win? Would like doing the first 30 days actually be a quick win to get more buy-in so that they're then more likely to be like, okay, yeah, we got to do the cleanup work. Definitely don't give the cleanup work away. And this is hard sometimes. We get excited about a client and then we're afraid to put a big number in front of them in the beginning. But consider what types of people opt out when they see that big number. They're the people who are like, yeah, my stuff's a mess. That is going to be a lot of money to clean up. I don't want to pay that. Or the people who are like, I don't have the money for that. I just want to say those people don't still need help, but it's probably not the last time that they're going to say, I don't have the money for that. And so this can be a hard thing to judge fairly on our own business. So let's change the perspective. Imagine you're buying an accounting firm. One accounting firm charges a $10,000 upfront fee to join the client list. The other does the cleanup work for free. What accounting firm do you want? Pretty obvious, right? One firm is going to have a much more premium client list than the other. So as much as it pains us to like the idea of like, oh, I might lose this client if I put a big price upfront, what's the alternative? Protect that client list, fam. Okay, two to go. What's the best way to filter resumes and applications on Indeed using cloud? Okay, so you put out a job listing on Indeed and you get 1200 applications in 24 hours. And all but three of them are actually people that are qualified to do the job. How can we use AI to filter through more of this stuff? Use and foremost, make sure your job description is accurate. Like what are the required qualifications? Have we actually put enough thought into this to make sure that it captures everything I want to see from an individual? That is step one. Do you have a good job description? Step two is do you have a grading? What are the day? Rubric. We should have a process for how we score folks at each step of the hiring process. From how is the resume and the cover letter scored? How is interview one scored? How is interview two scored? Is the only way to build a system that improves with every bad hire that you make? Because you can look back at the system then and say, how did we miss this? Like how did this individual slip through and they weren't to fit? How can we build a better system to ensure that we don't miss the same issues next time? So we have to have a rubric for how that resume is scored. And then third, in a perfect world, we give cloud a pile of examples. Here's some past resumes and how we graded them. And at that point, honestly, it is as simple as just. As simple as dropping those resumes into a folder and having cloud work through all those different resumes, grading each one. What you actually want it to do is let through the ones that have a passing score or ones that have sort of things that are maybe fringe situations where they might actually be able to still pass your rubric. And cloud can do a pretty good job depending on how you prompt it. If I say, only let people through who have one, two, three, and four, then it will be very strict in that regard. But what if you have somebody that comes around and actually has an even higher level certification? If you kind of loosen that language a little bit, it will also give you fringe folks. So we'll give you the people that are a clear pass. And we'll give you a short list of people that are still worth considering. You can also have it go out and like research these people online. Get some back information about them as well. So this doesn't have to be anything more than just like putting a pile of these things into a cloud conversation. The harder thing for firms is actually the system. What is the firms system for how we do this? There's a way you could fully automate this. So like every time a new resume comes through it automatically gets graded. Easiest way to build stuff like this right now is cloud code routines. If you're in the desktop app, flip over to code and check out the feature called routines. Routines are workflows that can be triggered from external services. I don't want to get too nerdy. I'll start you there. Go check that one out. Last, you said we need to reimagine our job statuses for the AI era. What job statuses will exist in the future? So what does it look like to put an AI into production? Like to actually trust it to get parts of the process done inside of our accounting firms. I think that probably looks like rethinking what our work statuses are. Right? We've all got these steps that projects progressed through from intake to prep to review, to partner review, signature delivery, whatever all that looks like for you. If we're trusting AI now with parts of that process, that probably needs to be carved out of the statuses. A good example of this that I had pre-AI where we de-skilled part of a process is when in our bookkeeping business, we had a team of admins do the pre-accounting work. That is all of the gathering of information, bank statements, fetching stuff online, chasing clients down for stuff that we need. I had all that stuff get spun out to a team of admins. so that the work didn't go to the accounting manager until everything was in. And in practice, in our workflow system, what that ended up looking like was the step that was prep got split out into a step that was pre-accounting followed by prep. Now, in an era of AI, where there's parts of the process we can now carve out for AI to do, we're gonna do something similar. We're gonna carve a step out of prep. We're gonna carve a step out of review or out of intake. We ought to be able to see in that process mapping where the AI is stepping in to take over part of that work. Now, where does this go eventually? It's hard to say because I could see a world in the short term where our status is like explode out to be this longer list of different more granular things 'cause AI can come in and do these different parts of the process, but I can also see a world maybe further down the line where that then contracts again, right? I could see a world where like everything up to review is just something that the AI managed and we are just now reviewing it when it comes out the other side. Maybe that's what it ends up being. Maybe it gets more nuanced as we get closer to that and new status is emerged, like new human sort of checkpoints emerge. I do think as AI does more of the work, I hope that more of the touch points return, the checkpoints at which we interact with the client. I think a lot of touch points with the client have gone by the wayside as we got over capacity and started doing too much work. Most of us have stopped doing intake meetings, many of us have stopped doing delivery meetings and an age of AI where the value that you and I add is the conversations, the context we can glean from drawing things out of our clients. When that is what's left, that stuff is even more important. So I think the process probably evolves on a whole as we put more emphasis on that stuff, but you could see a world where those statuses sort of explode and it becomes a longer list of things as the AI gets pulled into more and more, but then maybe down the road, there's actually a contraction again, as the tech just handles more of the production, more of the process. I don't know, hard to say, all I can say right now is what do I do next? What you do next is you look at the whole process and you say, can an AI agent do this? And if there's a part of the process that it can do, then that probably needs to have its own status now. For the same reason that we previously designed statuses to go to humans of different levels of expertise, like that's why we created those divides, it went from a prepare to a reviewer because a reviewer had a different skill setting could review the return. And then went to a partner because they had permission to sign the return. AI is like another worker that we have inside of the firm, and that ought to be reflected in the steps that are involved in getting that work out the door. What a time to be alive, man. Weird time, AI stuff, it's getting into your slack, paper trash bags, huff and gymnastics, chalk, what a whirlwind we went on today, gang. Appreciate you turning out, and I wanna see you here next time. Do we reduce again? Okay, wouldn't miss it for the world, can't wait to see you there. Until next time. (upbeat music)