621 The 9 Most Asked Questions About Accounting Firm AI [AI for accounting firms]

Today we're speedrunning the 9 most common questions people ask around accounting firm AI. Can I use this with secure client information? How do I get my AI to work QuickBooks for me? But also like, where do you even start with accounting firm AI? Also on the pod today, the US just got some really interesting security guidance on using AI inside of accounting firms, but it came with some problematic language suggesting that if we save time with AI, we ought to be passing those discounts onto our clients. What? Also, we're getting in the best way to post journal entries to the accounting system direct from Claude, whether there's a way to get AI to automatically fetch your client's bank statements from the web for you, and what to do when your clients— your clients think that AI can do the work that you do for them, and so they think you ought to be discounting what you do. Had a client ask you for an AI discount yet? If you haven't, it's probably coming soon. We're going to cover all this and a bunch more on today's pod. Let's get into it. How do you do, fellow accountants? My name is Jason Staats. Welcome to the Jason on Firms podcast where we talk all things accounting firm strategy. Whether you run one, whether you work in one, this podcast will turn you into an accounting firm strategy savant. Results not typical and not valid in New Jersey. Today, let's get down to the brass tacks of how and where to use AI inside of your accounting firm and just what exactly is inbound today, because we know this stuff is getting good. But how do your clients feel about this? How do the powers that be feel about, uh, the fact that there are absolutely not secure ways to use AI? Let's get specific and run through these 9 beats. Number 1, first and foremost, the most common question I get about AI, can I put client-sensitive information into it? There remain a whole, a myriad of ways to use AI insecurely. And this is why every day we're seeing headlines from don't put your medical images into Grok, This person put XYZ into ChatGPT. Here's what happened next. But, and this is very important to understand, the fact that there are insecure ways to use AI does not mean there are not completely secure ways to use AI, even with sensitive client information. Now, there are corners of the internet and the world where I will be burned at the stakes for saying this, largely by people that either don't understand AI or, or two, have a, a solution, a more secure solution for AI. A growing number of those people out there who will do the fearmongering while simultaneously have the solution to sell you to the fearmongering. But the reality is you can absolutely use AI even with client-sensitive information if we're doing it the right way and in the right places. Biggest thing here is to ensure that you are using the business plan, the most secure plan that you can use for your firm. For most people, the right answer right now for the right tool to use is Claude. It is the most powerful AI agent we've ever had for accounting firm work, and that means you are either using the Claude Team plan or the Claude Enterprise plan. Enterprise plan has a minimum number of users of 20, and if you are past 10 or 15 users, it's worth considering the Enterprise plan. In addition to a bit more security, you actually get a little more capability with the Enterprise plan, namely in models that have larger context limits. That is, it can kind of keep in its active memory more information at a time. Now, misnomers abound when it comes to AI, like the notion that AI has this inherent sponginess. If I put something into it, it's gonna know it and it might, might tell somebody else that thing that I gave it. And this is possible with AI, but only when the prompts that you're giving it are trained into the model. And this is never the case on say Claude's business plans. You have a commercial agreement that says your prompts are never trained into the model. So you don't have this risk of you telling the AI something and then the AI sneaking behind your back and telling that information to somebody else. That is just not a thing. Now, the details still matter here and how you set up Claude for your organization. It's worth getting an understanding of all the organizational settings in Claude. And this is where we dial in exactly what the members of our team are able to do and what they're not able to do. And today, if it's my accounting firm, there's a few things I'm going to disable, even though they are very powerful. One of them is Claude in Chrome, the ability of Claude to take over a Chrome browser, because while this is very powerful and really, frankly, novel that AI can control your browser, it is also probably the riskiest way to use AI right now because there are things on the web that could try to trick your agent and your agent just out of sheer ignorance could potentially go out and disclose something that you didn't intend for it to disclose. Now, agents like in Claude today, there's a whole bunch of reinforcement learning that actually goes into ensuring these things don't disclose stuff they're not supposed to disclose and navigate the web responsibly. But out of an abundance of caution, because in accounting firms we do work with a lot of sensitive information, right now I'm disabling Claude in Chrome altogether. Another thing that is worth considering disabling is the ability of your team members to upload their own skills to Claude. Skills are kind of like repeatable workflows in Claude, and they're very powerful, but with skills come the risk that there could be malicious instructions. And so if a member of my team goes out online and goes to, there's all sorts of like Claude skill libraries now. There's a world where they could download a skill that has something malicious inside of it. It could be a skill that's very useful, but hidden in there is, you know, an instruction to have it go do something that it shouldn't do, which means I personally right now in my firm wouldn't want to open up the ability for anyone to create a skill and add skills inside of Claude. Use your own judgment here. This may or may not make sense for you, but I'm probably going to disable this. This is worth going through with your IT provider, these organizational settings. Frustratingly, a lot of IT companies don't really have a strong understanding of AI right now, but take the time to go through these settings to get an understanding of what it means. Understand that Claude, 80% of Anthropic's business is enterprise, which is actually a much higher bar for security than is commonplace in the tools that we use inside of accounting firms. But make your own decision here and pull in a trusted expert to sort of make sure you got an understanding of what you're doing before you get into it. Most common question number 2, can I just use Copilot, Microsoft Copilot instead of Claude? Because it's just, it's right here. It's already in my Microsoft ecosystem. I do, I got to pay a little bit extra for it, but wouldn't this just be easier? It is. The answer is it is much easier to roll out Copilot than it is Claude. But the fact that it's easy doesn't make it worth doing. Doesn't mean it's worth doing. Really common thing I hear right now is, is Copilot can now serve up the Claude AI models. Hooray, this is great news. Does that mean Copilot's now good? No, no, Copilot no good. If you think, if you remember back to the very early days of ChatGPT, Microsoft took a 50% ownership in OpenAI and they said, huzzah, all the best ChatGPT models are now available inside of Copilot. Did that make Copilot good? No, it was never good. There is more to it as it turns out than just serving the model to make that tool useful. While Copilot has gotten better in the last 12 months, Copilot is still so far off of Claude. Like, when it comes to, like, driving real production work in an accounting firm, which— that is how I want to use AI, not for these cute little side tasks, but to actually drive production in my accounting firm— Claude is vastly more capable than Copilot. Yes, it means there's more due diligence in terms of onboarding another app, but this is by far and away the most powerful tool we have ever had for automating accounting work. And if we're not using it, if we're not leaning into it, then we're not only missing out on the productivity opportunity, but the bigger, more existential concern of am I learning how to use AI? Is my team learning how to use AI? That might be actually the biggest consideration here is we all need to be upskilling and getting comfortable with this. And in no world should you settle for the tool that is second best. I say second best, Copilot, it's not even second best on the list. It's, it's a number of ways down, but your time is more valuable than to waste it with anything but the best. The best is Claude. Question number 3, is it worth moving from ChatGPT to Claude? Yes. Do I need to say anything more? I'll just say yes again. Yes. Question number 4, where should I start? It's interesting the first time you plop somebody down in front of an AI assistant and there's just that blinking cursor in the empty box. There's some people like me and maybe like you who saw that and you were like, oh, where has this been all my life? Yes, this is incredible. But probably like 80% of accounts, they see that empty blinking box and they're like, uh, okay. Um, so what do, can you, what do I put in here? Can you tell me what to put in here? And you're like, what do you mean? You can put anything in there. Like that's what makes it so incredible. And they're like, okay, I'm gonna put my name in here. You're like, no, you don't get it. That is, that's the experience for most people I think is they don't really know where to begin. And the best recommendation I can give you is just bring a task to it, a specific job that you're trying to get done and ask it how can it help you to get this job done? Because it has all these skills that we don't have from being able to write and run code to being able to pull open source code libraries from the web to help it do all sorts of different jobs. We're using Claude and my team to like fully edit videos. So like we can't, like we'll be very bad at guessing what it will be helpful at. So the best thing we can do is bring a job to it and there'll be times where it's disappointing, but there'll be other times where you're blown away and you're like, holy smokes, it's actually very, very good at this thing. And that can oftentimes kind of be a wedge into like, oh man, like our firm can really use this for tax planning or for reporting. Talked to someone recently who's rolling out, uh, basically using Claude to replace QuickBooks inside of their CAS practice and getting a lot of early traction in this. I wish there was a, you know, everybody comes back to like the iPhone killer use case when they, when they had a camera app. I wish it was as simple as giving you like, here's a killer use case. Honestly, that killer use case might evolve into Claude being an accounting ledger. We've done some YouTube videos on this, the fact that you can genuinely replace QuickBooks on simple client ledgers with Claude. So maybe that becomes the killer use case. But for now, because accounting firms all do the same thing so differently, it's very hard to give you a single thing that will be helpful to you as much as share the stories of all the different places that other firms are having success and those ideas hopefully inspiring you with kind of similar things that you do. In fact, on my newsletter this week, We'll be sharing over 100 use cases that we have crowdsourced from accounting firms. Uh, and folks on my newsletter are getting that totally free this week. If you're not on that, jasononfirms.com/newsletter to sign up. You're gonna have an abundance of ideas. Most common question number 5, how do I build my first agent? And I would say an, an agent implies a level of, uh, autonomy, like the AI is going out and doing things and making some decisions for itself. And just to be clear, like, While there, it's true that AI can do this and agents can certainly make interesting things on their own, like just kind of given the boundaries and the guardrails within which they're allowed to operate. While it can do this stuff, this is not the place to start with AI. This is, if anything, the riskiest place to start. Most of the agents you see people talking about online are like banging around and doing stuff on the web and maybe making these kind of, these decisions on their own that in the context of the work that we do are a little bit risky. Like, could an AI agent process a payroll for a client? It sure could. Does that mean that it should? That, that would make me pretty nervous. And so rather than jumping straight in and being like, I'm gonna build an agent now, more realistically what most of us do is we identify a cool use case in conversation with the AI, and you use that over a series of weeks or months. And once that use case is really nailed down, then potentially there's a world where that use case is then set up to work a little more autonomously. Uh, for example, let's say there's a step in a process where every time you just go to AI and AI kind of does that step in the process for you, wouldn't it be nice if that could just run in the background so that you didn't have to have a conversation with the AI every single time you needed to do that? We can absolutely build that. It's actually not that hard to get AI to do some of these things automatically for you, but that's not the place to start. That is where a use case evolves once it's kind of well understood and you've done it in conversation with AI quite a bit. So the place to start is not building agents that go off and do stuff autonomously. It is in nailing down these use cases in conversation with AI that eventually evolve into something that has a little more autonomy. Are you still running your accounting firm on a spreadsheet? Puh. 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 shipping some cool stuff lately, like built-in month-end 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 of Financial Cents. 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 10% off the annual team or scale plans with 2 or more users. I'll put the link to claim that, uh, down in the show notes. Check that out, see if Financial Cents 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 1,800 1040s 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. PDFs. Oh my gosh, can't say enough about Sortman. We actually did a full demo of their platform on my YouTube channel not so long ago if you want to see it. But to learn more about Sortman, check out the link down in the show notes. Question number 6: Just give me some good starter use cases. All right, most common places I see firms using AI right now— email management is a huge one. Claude has very good connectors for, for Gmail, for Outlook, and it gives it read access to your email, which means it can monitor the stuff that's coming in, flag stuff that's spicy, even automatically triage triage, even automatically triage the stuff that, that maybe you don't need to worry about yourself. So email management is one. Second, creating import CSVs that go into QuickBooks. With QuickBooks, you can import, uh, entire transactions, journal entries. There's a number of different like bulk import functionalities in QuickBooks, and your AI assistant can create these CSV files that are imported into the system. And this can replace, I mean, a huge amount of work. Let's say you got to do a big, ugly monthly journal entry for a client that comes from 10 different reports. You can provide it past examples of that journal entry in the input reports and then give it the current month's input reports. And you will be shocked at how good of a job it can do at building that entry. Now, anything that you're getting AI to do for you, it's very important to think about what the expectation is in terms of the documentation it puts together. AI, left to its own devices, will put together documentation, but it's, it's humongous, like way too verbose documentation. For my tastes. The best documentation, the best work papers in an accounting firm are as concise as possible because their job is basically to convince a reviewer that the work was done correctly. And I don't want to get an 11-tab workbook back from the AI tool and then have to review all that, right? So also give it guidance for what good documentation looks like, and maybe it's just the past few months' documentation, and then AI just builds the current month's documentation for you. So starter use cases, email management, import CSVs. Also, I would say anything that involves a spreadsheet. AI is phenomenal at using Excel. Big, ugly, even technical spreadsheet workpapers, spreadsheet tasks. AI is very, very good at. The best way to put it to work is show it past examples of you doing the work and then cutting it loose with, you know, the current month or current period inputs. Question number 7, how do I get my Claude to use QuickBooks? Can it? The answer is yes, but it kind of depends. The most obvious way to get Claude to use QuickBooks is by it actually using it in the browser. And we're talking about QuickBooks Online. And it's true that Claude can control the browser on your computer. But as we talked about in the security question, this is actually something that I recommend disabling inside of an accounting firm because while it is very powerful, it is still a bit risky. It could make a mistake. And so the other more effective way to use QuickBooks with AI or, or really any tool for that matter, a web-based tool, is an MCP. And an MCP is the connector from your AI assistant to the tool. The trouble is not all MCPs are made equal. So right now, QuickBooks's Cloud MCP, it will let you connect to a specific QuickBooks file, one QuickBooks file, and read very basic information from the file. And that might be better than nothing, but MCPs are capable of so much more. And in the context of an accounting firm, when I work with a whole bunch of clients, like, I want access to to, you know, the 250 QuickBooks files that we manage for clients and a connector that can support all those companies, not just a single company. Now this is because the current QuickBooks MCP connector is really just made for business owners that have a single QuickBooks file. And so while this will get you some information from QuickBooks, it won't let you do all that much. And so what is the best way to get AI to work QuickBooks? Honestly, right now it is probably AI that uses a web browser. Just recognize that's probably like the riskiest way to use an AI agent right now. Little later, we're going to talk about how do you— how you can actually build your own MCP that will give you expanded Claude access to QuickBooks and let it do a little bit more. But there is a third way that you can use Claude to, to in some ways automate QuickBooks. And this is by building your own Chrome extension with Claude. So Chrome extensions, they kind of live up in the top right of your Chrome window. They're capable of modifying things on any website that you view, of automating certain actions on websites. And Claude can build you your own Chrome extension. This is actually a fun experiment if you've never done this before. Chrome extensions are fairly simple and they just, they live in a folder on your local machine and Claude can build them for you. And so you can have a conversation with Claude or like give it a screenshot of a page in QuickBooks and be like, hey, I very commonly do XYZ. Uh, let's figure out how to automate this in Claude or, or sorry, with a, with a Chrome extension. Or, hey, I wanna build a Chrome extension to check all of my clients' banking connections that I have access to in QuickBooks. Here are the steps that it normally entails. Let's build a Chrome extension to do this. And so the result will be you have this Chrome extension with like a dropdown menu and different options you can select. And when you select it, it will run code on your webpage and that is completely disconnected from Claude and you have full control over what are the capabilities of that extension. It's not. An agent going around and doing things autonomously. It is AI writing some code for you that you can then trigger from the, the Chrome extension whenever you want that thing to run. Um, if you're a user of Write Tool, Write Tool is a Chrome extension that helps you automate things in QuickBooks and really improves upon the user interface. It is not too dissimilar to kind of building your own version of that to automate things, not only in QuickBooks, but anywhere else in the web when you are doing repetitive tasks on a website. Web automation has been a thing for a long time, like being able to build scripts on top of web pages. It's just, we're not programmers. We don't usually write scripts, so it's not a thing we've had access to until now. Claude can write all these scripts for you, and then you can package them up inside of a Chrome extension to be triggered whenever you need them to run. Two more questions to go. How do I get Claude to automate my tax software? I'm going to assume this is a US person asking the question. Most tax software in the US is still legacy desktop-based software. Even if you use it in a hosted environment, it's still desktop software just being hosted. And this stuff is chronically difficult to automate because it's not in a web, it's not in a Chrome browser. It is just a local app. Now on the business plans of Claude today, there's no way to automate things on your desktop. On the consumer plans of Claude, there's a feature called computer use where it can like tippy tap on your keyboard and control your mouse and literally do anything on your local computer. This AI, it could do things in your tax software, but I don't recommend doing it. The consumer plans of Claude don't come with the commercial agreement that says they'll never train your prompts on the model or, or into the model. And so I just, in an accounting firm, I don't recommend you touch the consumer plans, only go with the business plans. So for that reason, right now we don't have a Claude feature that can control the entire computer. And even if we did, for the same reason that cloud on Chrome is risky, this would be tremendously risky. Chrome being— cloud being able to do literally anything that any of the apps on your computer can do. But is there a world where you could set that up on a standalone machine that only had your tax software on it? Down the road, potentially. But today, no, because the only way to do it is with the consumer plans that don't come with like AICPA SOC 2 Type 2 certification. I just don't trust the security on those plans. Now, the one outlier here is you have ProConnect, a browser-based tax prep platform from Intuit. It's, it's fine for many firms. It will work well enough. It doesn't handle that much complexity if you're doing business stuff, if you're doing multistate stuff, that's where it usually starts falling down. But I can tell you, having spoken with folks who will use Claude to automate their web browser, which again, I probably don't recommend, But there are a good number of people that are now using cloud to automate inputting things into ProConnect, and it works very well. And this is, I mean, if there was an argument to move to ProConnect, it's probably this. The fact that it's way easier for AI to automate things that are happening in a browser than it is to automate things that are happening on your local desktop. That might be a temporary thing. Who knows? Maybe we will get desktop automation to the cloud business plans and be able to trust that soon. If so, there's a world where it could navigate your UltraTax, your Drake, whatever you're using. But right now there are a good number of firms that are already automating ProConnect entry with Claude. Last question, number 9: Do I need to ask my clients permission to use AI in my accounting firm? Totally reasonable question, right? Feels like a new thing. A lot of clients don't understand AI and they might be spooked by it. But consider, uh, did Intuit ask you for permission, or any of the people that use QuickBooks, for permission when they launched Intuit Assist? Their little AI assistant. Did your practice management system ask you for permission when they launched an AI assistant inside of that platform? The reality right now is AI is in everything, like it or not. The IRS just released some, uh, initial AI security guidance for accounting firms. And like the second paragraph down, they're like, gang, everybody's using AI, whether you even realize it. And so in 2020, and so right now, do you need to ask a client for permission before using AI? To get work out the door? My answer here is a resounding no. Now, do I need to be 100% responsible for the risks that come with using AI? Absolutely. 100%. I'm accountable for the output of AI. I'm accountable for ensuring that we're only using it securely. In no way will I ever be able to say, hey, it was AI's fault. Like, no, no, no, no, no. The buck still stops with you. And ultimately, at the end of the day, our clients have never given a hoot how the work got done. They care about the output. And so in my opinion, like, like, no, I don't think we have any responsibility to tell clients like, hey, we use generative AI as, as, as part of how we put this project together. What would that even mean? Like you used Claude, you used a tool that had generative AI in it. Like almost all of the apps that we use now, 80% of the apps in your, in an accounting firm now are using generative AI in some way or another. The biggest risk of asking is if the client says no. What does that mean? You're not going to use QuickBooks then? Or like it or not, I think that toothpaste is out of the tube. Let's stay laser focused on like, what can it add? Like, how can I deliver even more value as a result and get our heads around like how to do so responsibly? Now, I mentioned this new guidance in the US from the IRS, sort of our tax agency around is it okay to use AI or not? And by and large, like the answer was like, yeah, as long as you're doing so responsibly. But inside of that bulletin, it said that you need to discount your clients if AI is saving you time in the process of doing the work. Are you kidding me? We're going to dig into this. Okay, let's talk about it. It's Nerd Corner. Nerd Corner. Let's nerd it out. Nerd Corner. This is something like whether you do tax work, whether you're in the US or not, I think we need to be paying attention to. Because, oh my goodness, if the powers that be, if the governing bodies of our profession are saying that we need to discount our services and are echoing like what some clients are saying now, like, oh, did you just use AI for it? Then it should cost less. We hear that and we're like, no, no, no, no, no, hang on. It's not going to cost less. If the governing bodies of our profession are saying the same thing, that's a problem. That's not great. Here's exactly what they said. Quote, AI can reduce research and drafting time such that billing clients for manual labor or time that was not actually spent, or double billing for AI-assisted tasks may violate Section 10.27 depending on the facts, e.g., a noticeable pattern across clients or the size of the billing differentials. I'm going to read the first part of that again because it's kind of a hard sentence to follow. AI can reduce research and drafting time such that billing clients for manual labor or time that was not actually spent, or double billing for AI-assisted tasks— that was a triple or sentence— may violate blah, blah, blah. It continues. Cost savings should be passed on openly with billing practices that reflect the efficiencies gained from the use of generative AI. Practitioners should not only disclose in general or specific terms as needed the AI activities performed, but also fairly credit to the client's account any cost reductions. To me, that sounds like not only discounting hourly billings when time was saved, like, you know, to not include hours that were never spent, which makes sense. But that last part, it says, but also fairly credit to the client's account any cost reductions. That almost sounds like if somebody is on a $1,000 monthly retainer and we used AI one month and it saved us a bunch of time, this is almost seeming to suggest you ought to credit their account for any cost reductions. It's not seeming to say that. That's literally what it says. How should we feel about this? Are you mad yet? Because aren't we all like, isn't everybody saying like, oh man, if AI saves us all this time, how are we gonna be able to charge how much we've charged in the past? Tell you right now, things haven't gotten easy yet. And the client who comes to you, we're actually gonna talk about this later in the episode. The client who comes to you and is like, well, can't AI do this? Shouldn't this cost less? What clients are paying us for is like the output, is the outcome, is, is this correct and are you going to stand by it? We've had tech to help automate that for years and, and for years clients have said like, couldn't I just use QuickBooks for this or Xero for this? Or couldn't I just use TurboTax? What am I paying you for? And this is a bigger problem than just AI. This is how clients perceive the value of what we do. And are we actually solving expensive, painful problems for people? But if that conversation is being led by, like, the people who govern our industry, that's, that's a harder argument for us to make, right? There's going to be a few more clients raising their hand and being like, hey, hey, the IRS said your work should be cheaper. Why is it not cheaper? Now, in this case, it's, it's the IRS saying this. And the IRS, I would not say, are like, are like going to be lobbying on behalf of professional accountants and tax pros. It's interesting. If anything, the IRS is like saying, hey, you know, all of our rules and just everything that sucks about the IRS, AI's fixing that. So your tax pro should charge you less. Thanks. Thanks, IRS. This is a pretty shockingly old-school perspective that it's almost written through the lens of someone that is exclusively familiar with the hourly billing and hourly pricing paradigm where we are like tradespeople and that's kind of it. We're trading dollars for hours. The value pricing conversation has been around accounting firms for eons, and most firms still say they hourly price, even though at the end of a project, what they actually do is they're like, whoa, what did we charge them last year? Okay, here's where the number came in this year. Eeny meeny miny, this number. And we call that hourly billing. That's not hourly billing. That's value pricing. That is, I'm pretty sure client will pay this, so I'm going to bill them this. And sometimes that's a great number for us. Sometimes that's a bad number for us when it's a stinker client that complains about fees. But where most firms still identify as doing hourly pricing, because usually there's like some sort of moral friction with value pricing and it feels like, feels like you are no longer the transparent, we're going to bill you fairly for your time sort of firm, you know, to go to value pricing maybe feels a bit like you're trying to squeeze every dollar out of your clients. Well, I think a record number of accounting firms just became value pricing firms this week. Now that the IRS is saying if AI reduces the number of hours you spend on a project, you can't still bill a client for that work. That's basically what this is saying. And so if you hold yourself out as hourly pricing work and you're doing tax work, do you really want to decrease your prices if you're using AI? Probably not. I always come back to the story of a, of a firm that uses AI to rename 100 receipts. So client drops off pile of 100 receipts, you point Claude at the folder and Claude renames all those files in your naming convention, like date, payee, and amount. And that's awesome. That's a huge time saver. AI can do that very well. Now a human doesn't have to hop into every one of those files and, and rename it. But if that saved you an hour of time, would you still bill out that hour of time the way you always have in the past if AI does that work? The answer is yes, and that is value pricing, because that work is just as valuable now as it was before AI would do that work. And maybe that changes over time. Maybe the value of everything we do in this profession gets eroded over time with AI. Seems unlikely, but maybe that happens. I can tell you right now, hasn't happened yet. But what this guidance from the IRS is saying is if you bill hourly and AI wipes out that hour, you can't bill that hour. And this is what makes you now a value pricing firm. Welcome. Welcome to the league of value pricing firms. You've already been doing it for years. Unless you were billing out to the dollar and to the penny of what those hours were, you've been value pricing for a long time. I told you. No, no, no. You always— you were always part of the club. Come on in. It's great to see you here. Who knew that it would be the IRS that brought you around to it, right? But here we are. Now, I've got some more thoughts on that, but I've actually got a related question. Sort of a similar angle on the same thing, a little more practical in terms of how we talk about this stuff with our clients. But it was a listener question. Let's throw to the mailbag. Mailbag, mailbag, this segment is called Mailbag. Got a bunch of great questions this week. Questions come from, uh, podcast comments, from social media DMs, from my email, from my phone. Text me your question at 1516-980-4968. First question: What software can be used to automatically get bank statements without giving Claude my client's bank logins? That's a two-parter. Let's talk about the first, the concern over giving Claude your client's bank logins. This is a totally valid concern, but interestingly, A concern for which there will be a solution probably sooner than later. If you look at OpenClaw, OpenClaw is an open source AI agent. We did a YouTube video on it a couple of months ago. It comes with a built-in ability to use 1Password, the password manager. So it's basically its own like little password vault that that AI agent can manage. And there's entire platforms now dedicated to this. They want to be kind of the secure credential manager for the AI agent. Because in order for an agent to log in and do something in a system, it needs to authenticate in just the same way that you and I do, right? And so this is something that there's a lot of people working on and we will have a more secure solution for hopefully sooner than later. But the biggest blocker right now with automating the fetching of client bank logins is the fact that letting an AI agent do this, it's just very risky. Like it is capable of doing it for sure. You can send it to Bank of America and it will almost certainly be able to pull that stuff correctly for you. But right now, as things stand, if I'm running an accounting firm, that's just a little bit too much risk for me to expose myself to because sending an agent out onto the open web, there's a risk that it could be tricked into doing something that it shouldn't. And that is a very, very small risk, but it is still a risk and a risk that's frankly tricky to monitor. The worst version of this is it going and doing something to try to complete a task for you, and it does something wrong without you even realizing it. That is the big risk, really. And so today, as things stand, what are the best ways to fetch bank statements for my clients on an automated basis? Uh, first and foremost, look out for statements that can automatically be fetched by the accounting ledger. Uh, QuickBooks Online— does Xero do this? I don't know— it will actually fetch bank statements from certain banks every month for you. This is because a growing number of banks support providing the bank statement via API. And so like, you don't actually have to log in and grab it because the ledger will just fetch that statement for you. So first check there. Second, there are services like Hubdoc that will let you connect Hubdoc to a bank and it will fetch the bank statement if that's something that the bank allows Hubdoc to fetch from them. So there are some services that like that, that allow for it. There's a service called LedgerSync where this is exclusively what LedgerSync does is you connect it to all of your clients' banks and then you give your team access to LedgerSync so that they're able to pull banking information without having access to the banking credentials. That's the whole point of the LedgerSync platform is for an admin to connect it to all these different banks, but then you give your team access to different clients inside of LedgerSync where all that banking information lives so that your team doesn't have to have access to the credentials. But for me, the best solution for this is still a combination of a cheap human and SSO. Single sign-on. AI hasn't quite replaced accountants yet, but if there is a part of accounting that is in an 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 Maker's Hub. They have taken automated extraction from invoices and receipts farther than I've I've never 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, MakerSub's got you covered. They handle payments as well. They're doing foreign 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. 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. MakerSub, worth checking out. Putting AI to work on the most hairy accounts payable workflows. Learn more about MakerSub, check out the link down in the show notes. Tax election 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 have been sitting at the tippy 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 tax intake to built-in month-end clothes automation to now. Recently they launched Canopy Coworker, a built-in AI assistant, uh, to make your wildest dreams come true. And they're going to give you $75 just for popping by for a demo to see what's changed. Head over to getcanopy.com/Jason, book a demo just, just to see the new stuff. They'll send you a $75 gift card afterwards. Okay, that's all you got to do. You just got paid for listening to an ad Good read, man. Go do it. I'll put that link down in the show notes. So most accounting firms still manage client credentials with, uh, best case for most firms, a password manager. Worst case, you've got bank logins like just stored in plain text somewhere, which is probably not smart. Not probably, it definitely isn't smart. But even using a password manager, if the members of your team have those credentials in a password manager, it does nothing to prevent the people on your team from being able to see all those credentials. Imagine a member of your team leaves tomorrow and they had access to 100 client credentials. Isn't that a problem? Isn't that like, isn't that a concern? It should be. Now, enterprise, like big businesses, solve for this via single sign-on. Like you log into one platform and you launch all your apps from that platform. And there's, there's one service that does this for accounting firms. It's called Practice Protect. It was acquired by Rightworks. A couple years ago, and to be clear, it's a pain. It's kind of a bear to set up, but in the end, what you have is your team members log into Practice Protect, and from that page, they can see all the credentials they have access to, like links to the credentials, not the credentials themselves, and then click through, like the Wells Fargo for John Smith link. It opens up Wells Fargo and logs it in without your team member ever seeing. The credential. And so the way I handled this fetching problem inside of my accounting firm, where you tick over to the first of the month, you got all these things you got to gather for all these clients, right? We had a centralized admin team that was responsible for doing this, and they could fetch all of those things, but they couldn't see the client credentials. Now, I used humans for this because there was a whole bunch of things beyond just fetching statements that we had to do every month. There was all sorts of loans where they didn't issue statements., and so we grabbed a screenshot of that every month. There were things we had to email clients to request from them every single month. We had to grab copies of health insurance invoices to ensure that the payroll accruals were done correctly. It's just a big hot mess of all these things you have to gather for your different clients. And the only complete solution for me was still a low-cost human. So we had admins that did this stuff and that got us further. But if you're really trying to automate this stuff, I still wouldn't recommend automating it with an AI agent. I would look at, does the accounting ledger already pull the bank statement? If not, look at something like Hubdoc. Or look at something like LedgerSync. But to me, the most complete solution here is a human plus a single sign-on app. Next up, Jason would love to hear your current thinking on business workpapers. I acquired a firm that uses CCH Engagement, which seems to be overkill. Most of our clients are on our Q— most of our clients are on QBO. Okay, the workpaper, ye olde workpaper. The job of a workpaper is to save the reviewer time. This is my opinion. While workpapers are used by preparers to like get to the final output, the ultimate job of a workpaper is to save the reviewer time. It doesn't matter if it takes the preparer more time to put that workpaper together. You are trying to save the time of the reviewer. Any shortcuts in this process, crappy workpapers, no workpapers at all, ends up taking the reviewer more time to review the work. But the goal in a firm is for preparers to put in more time to save the time of the reviewer. Now, workpapers, I would say, increasingly are being replaced by simply workflow solutions that do the same thing. Like a couple of good examples in the wild are Double, which a lot of firms use for month-end close automation. Tallyfor is a trial balance solution that now does some automated workpaper production. And these apps sort of walk you through all of the steps of the process of getting that work done and replace what in the past would have been like doc— like manual documentation of getting that work done. Now, to be clear, workpapers still exist all over the place in an accounting firm and are something that we don't talk about enough when it comes to automating work. We talk about getting work to a state that is done, but that's not really enough in our line of work. It's not enough to simply have a finished project. We have to have the documentation that validates that the end result is correct. And this is something that AI can help us with. AI can produce documentation. And I would actually argue that this is what moving a project forward actually looks like, is the AI is producing the documentation we need to get a project from step 1 to step 2 to step 3. Now, this still requires inside of our accounting firm agreeing upon what that documentation ought to be. It should not be up to every single human to decide in a firm how something ought to be documented. This ought to be a firm procedure. Here's the expectation for exactly how this thing ought to be documented. Similarly, if we're pulling in AI to help us produce that documentation, it should not be left up to the AI to decide how to document that thing. This could look as simple as giving it, you know, the last 3 months or the last couple years of that documentation and saying, this is the style that this documentation should be produced in. Uh, left to its own devices, AI, when it creates documentation, creates way too much documentation. Like, I don't want to spend my whole afternoon grading AI's homework. So provide it with exactly the documentation it should be using. The goal of that documentation of those workpapers is to as concisely as possible prove something. And we're asking AI to do the same thing. It's not enough to trust a black box answer from an AI. You got to treat it like an accounting intern. You got to make it show its work in as concise a way as possible for the reviewer. They brought up CCH Engagement. I would say, yeah, CCH Engagement, usually it was an app that historically was built for test teams like doing sort of large volume of workpapers in the context of an audit. I used engagement as well in my tax firm. It is overkill these days, I would say, unless you're doing a test work. Maybe you can make the argument for it depending on the type of work that you do. The number of workpapers that are required has been reduced, but in most cases this can be simplified down to something that just works in a normal file structure. You need to have a locked-down, like, folder structure and expectation around the naming conventions of the workpapers that live within there. But these days, most people have moved past something like CCH engagement unless you do a test work. All right. Here's the question digging into clients who have concerns about the fees we charge if we're using AI. How do you address clients who are open to AI and okay with you using AI for their account, but believe that your fee should go down since it can now all be automated? Parentheses. Yeah, right. On permission. I don't think we need to ask our clients permission to use AI. For the same reason that Intuit didn't ask permission to launch Intuit Assist, their AI assistant. AI is just like, it's in most of the tools that we use. For us to somehow be disallowed from doing the same thing feels like a cold take to me. But what about the client that says, hey, AI is automating all this accounting and tax work. Like, you're going to, you're going to give me a discount this year? At least in the near term, the answer here is a very clear no. And it's worth considering. This conversation has been happening for eons. Like the first time a ledger went to a computer and people aren't doing 12 columns anymore, they're doing the accounting in a computer. People are saying, hey, is that going to— is that going to— is that— should that— that should be— that should save us some money, right? And it's going to save us some time. Imagine when TurboTax first came out and Intuit is like, hey, normies can now prepare their own taxes from home with this, with this software. A one-time purchase, pick it up at Costco. You can now do your own tax return. The number of people that are like, what? I would do it myself in TurboTax. I just don't have time. So why does it have to cost so much? We have always struggled with our perceived value of what we do versus the client's perceived value. You will have clients who will gripe about any bill that you give them, but you also have clients who will happily pay a huge bill with a big ol' smile on their face. Why is that? It's because not everybody values what we do the same way. And this usually comes back to how valuable are the problems that we're solving for them. If someone pays $500,000 in tax a year and you can find them a way to defer some tax or save some tax, that's probably going to be a big number. You send them a $600 invoice for their 1040. By the way, don't do this. But if you did, they're going to be like, this feels like a pretty good deal. Entrepreneurs who will spend more money on tax in their lives than anything else. Think about that. Anything else in their lifetime, should those people be price shopping their tax preparer? Mm, that doesn't feel like the move. Now, that does not mean that there are not people on the other end of the spectrum where, like, there's really not that much that we can do for them to save them money. They don't have expensive problems, and so they don't value what we do very much. That is oftentimes not their fault. It's just their circumstances. It's the industry. For whatever reason, there's not as much value that we can provide them with. And so this ultimately comes down to how we position our accounting firms. Are we solving rare pains that are expensive? Rare because I don't want the guy across the street to be able to undercut my price. And expensive because if I save you $1 million, regardless of the inputs and how much time it took me, that person will be willing to pay me more than the person that I can only save $100. Enter value pricing. And so is AI a factor here? Like, will AI influence maybe the perceived value of some of the things that we do? Absolutely. But I would argue tech has always been doing this. It is just yet another signal for how we ought to be positioning our accounting firms to ensure that we're still solving expensive problems. Next up, I'm looking for a tool or MCP that can write into QBO specifically for journal entries I've created in Cloud. Okay, so right now QuickBooks is MCP for Cloud. That is the connector that you can add to your Cloud account and connect to QuickBooks. Doesn't allow you to push journal entries over to QuickBooks. Now, the easy workaround here is there is a CSV import functionality into QuickBooks like we've talked about. And so your Cloud can create a CSV and then you just take it over to QuickBooks and import it. But if you literally can't be bothered to do that, I'll give you two other options. One is, did you know with Cloud you can build your own MCP connector? What? I know a number of people that have done this now because the QuickBooks MCP on its own is just very, very limited for accounting firms. But you can build your own QuickBooks MCP that sits on top of— another incoming acronym here— the QuickBooks API. The QuickBooks API is basically every programmatic way that you're allowed to automate a QuickBooks file. And this includes writing data to the platform and reading data to the platform. And that API is much more complete than what the MCP allows for. I'm sorry that I have to talk in this way. This is just the world that we live in these days. Okay. OMG. And so you can build your own MCP that sits on top of the QuickBooks API and adds a whole bunch of sort of net new functionality, among them being able to post journal entries to QuickBooks. So this is absolutely a thing that you could build yourself if you want to. It's one of those things that Claude can do for you that you wouldn't otherwise be able to do yourself. Just go to Claude and be like, hey, I want to do this specific thing with QuickBooks. Help me build an MCP to do so. And you'll be surprised how far that can get you. Otherwise, one off-the-shelf tool that could be worth looking into here is called Booker, B-O-O-K-E-R. It's by a company called FinOptimal. What Booker does is lets you sync journal entries to and from Google Sheets to a QuickBooks file. So you can have a journal entry inside of Google Sheets and make changes to it, and those changes will sync back to QuickBooks. Some accounting files just by their nature are a whole bunch of huge, really hairy spreadsheets. Booker was originally created for companies where they've got all these tabs and tabs of spreadsheets and the resulting calculations become a spreadsheet. With Booker, you do all those calculations inside a spreadsheet, but then the journal entry dynamically updates in the accounting system. So you can either, I think you can either push to sync it to the accounting system or have it automatically sync. But by and large, Booker just gives you a way to have journal entries inside of a spreadsheet that automatically sync to a QuickBooks file. Which means then that you could use Claude to automate the spreadsheet itself, and any updates you make to the spreadsheet are then synced to QuickBooks. Obviously then you're paying for Booker per client, but question asker, you're the one that got us here and was too lazy to import these journal entries via CSV to begin with. So that's what you get. Hey, quick word from one of today's sponsors. Me, me. Yes. I run a community, Realize, that is for accounting firm owners. And Realize is kind of the secret behind how I've done all of my content and been able to like financially turn thought leadership into a business. About 100 firm owners in there. Realize is a systematic way to build connections with other peer firm owners. It's not a framework for how to run a firm. It's a way to introduce you to other super cool firm owners. Right now, We're putting together small groups so you can have a group of like 5, 6, 7 other firm owners just like you that you meet with every single month, run very similar firms. Having this for me totally transformed firm running. My north star, like my goal for you coming into this group is to make a new best friend, somebody that runs a firm, because that was what I needed in my journey. The whole community is built around that. It's called Realize, link down in the show notes if you want to learn more. Hey, real quick, I want to invite you to my super excellent weekly newsletter. My newsletter was like the start of my thought leadership journey. It's where I try to collect what I think are the most important things for you to see every single week, sent to you every Wednesday morning. Make sure you don't miss any of the good stuff, right? Increasingly, we're also building into that some, like, of the best evergreen thought leadership I've ever done, kind of make that, like, the most action-packed, highest quality email you get to your inbox each week. Because listen, we're all busy, all right? You're not gonna see every single little thing that I put out. Goodness sakes, there's just too too much. Mm. If you're not on my newsletter yet, check that out. JasononFirms.com, or I'll put a link to it down in the show notes. What's the future of CCH and Thomson Reuters with AI? Are they just slow to the party? Sure Prep feels obsolete and I see nothing in Thomson Reuters' horizon to replace or supplement their OCR, uh, tools. Uh, yeah, don't, don't hold your breath. What have CCH and TR shipped in like the last two decades and what makes us think anything's gonna change? I mean, they all came out and they said, They had like a, an AI measuring contest and said, here's the amount of money we're ready to spend on AI. And what's it, what's it done for you so far? Now companies, these companies specifically, CCH and TR, their innovation comes through acquisition. There will be a cool company doing something novel and they will acquire them and the team has to, you know, work some sort of earnout, 2 to 5 years, but usually as soon as the acquisition happens, the tool kind of goes to crap. SurePrep being a great example. They acquired SurePrep a couple of years ago. By all accounts, everything I've heard is that thing has gone absolutely nowhere in the meantime and service has gone off a cliff. Now CoCounsel, which is sort of TR's big AI thing that they're pushing, and it's really just a research platform. This was a tool called Materia that they acquired a year and a half ago. Maturi was doing some cool stuff. I know the founder's still around at TR, but you got a little less faith in it now that it's under the TR umbrella, right? Even with all the resources and all the investment of TR right now, the leading US tax research platform, that is AI research platform, it's BlueJay. It is not CoCounsel. So Thomson Reuters will use all of their leverage, uh, to make the same arguments they always do around, well, we're the only tool that integrates with all your other Thomson Reuters tools, and so we're the only one that can do it. So good. But they've literally never delivered on that premise before. And so my recommendation has been and will remain to be don't buy anything but tax software from these jokers. There are better alternatives for literally everything else in your firm, and I could get you thousands of accounting firms that will co-sign on this fact. Don't listen to your TR or your CCH sales rep. Please, please, please. They are the last people in the world that we should be listening to. Two more to go. My partner and I are building a fractional CFO firm focused on strategic advisory for small trades businesses. I know you think you're launching a fractional CFO firm, but get ready to be a bookkeeping practice or build a really good partnership with someone who wants to do bookkeeping for tradies. We're just starting out, but we anticipate staying fairly small, working with up to 20 to 25 accounts on the high end. At that scale, what do you recommend regarding IT security infrastructure? We're not doing tax work, but we may end up offering services in high-risk areas like processing payroll and bookkeeping type work that involves edit access to QuickBooks. We've done some research and it's hard to parse out what's truly needed and what's overkill at our size. There's practically no rules governing bookkeeping firms in the US, even for tax shops. Frankly, the rules are so, so vague. They basically just say, yeah, if you don't do a good job at this stuff, you could be in trouble. And next to no one has ever gotten in trouble. But the rules themselves are like Yeah, you should take appropriate safeguards, dude. There should be safeguards and they should be appropriate. And we're all like, what does that mean? So it is not as if there is like a black and white answer for like, here's exactly what this should look like. In my mind, the move here is just follow the industry best practice. Look at who are the IT groups that just serve our industry, because if there's 10,000 other firms using them, that in all likelihood means I ought to be covered. In this case, like we just got IT groups that work exclusively with accounting firms.. And that is just a really easy way to tick that box in my mind. Biggest ones right now are probably Verido and Wright Networks. My opinion, like, don't make it harder than it has to be. Don't pay the IT guy down the street to do your stuff for you because he doesn't actually understand the tools that we use. And where there are rules specific to what we do, he doesn't know them. I mean, I wasted— I spent years doing this and you'd have one thing break, CCH engagement, and you get an IT guy in and he's like, What on God's green earth is this application? And you would then pay them 3 days to learn how that app works and to fix the thing that was the problem in 20 minutes. Don't do that. Hire someone that just works with accounting firms so they already know the best practices for these arcane tools that we sometimes are forced to use. Uh, you're going to pay a fixed monthly, like per employee headcount fee generally for working with these IT companies, but it's an easy solution. Don't reinvent the wheel yourself. Last one, does AI threaten our creativity and, and tax planning? Let's broaden that out to any type of strategic planning, which normally starts with the conversation, flows into the bookkeeping, and ends with some sort of report or advisory call. Does AI threaten this given its black and white nature? Here's my take is what AI is doing more than anything else is compressing the work itself, simplifying the work itself, and you almost have to build your business as if the AI is going to compress the work itself to nothing. For production as it is sort of defined right now to become very trivial. And so what remains? I'd say what remains after that is your network, your connections, and taste, your perspective on what is the right way to do something that may otherwise be subjective. And sometimes we lose track of what a privileged position we are in getting to talk with entrepreneurs all day, especially if you serve a specific industry. You have such valuable conversations that the rest of your clients aren't privy to. And this gives you sort of this decision-making model for what should and shouldn't be done that these other entrepreneurs don't have access to. And that is not a thing that an AI will step in and replace in my mind, that nuanced expertise. I always come back to this, this sort of thought exercise of imagine tasking AI with decorating your home. Let's say you move into a new place and you're like, Hey Claude, I want you to go out and, and here's the floor plan. I want you to buy everything that I need to decorate my home. It could absolutely do that. And all those things would start showing up and you'd set them all up. And what you would end up with is something that looked like it was, I don't know, a JCPenney ad. Everything would be fine. It would work, right? But what if you wanted something more interesting? Well, then you could get, oh, maybe somebody like gave a prompt for like how to do this cool, specific, funky styling. And you went back and you're like, We're going to return all this crap. I want you to find me stuff that is in this really cool specific style. And you got a prompt that kind of captures that style. It goes out, buys all these things. You set it all up and you're like, oh, this is nice. This is different. This isn't, isn't quite so trad, not so JCP. And AI can absolutely do that for you. But there will still be a day where you walk into somebody's house, somebody else's house, and look around and go, dang, I didn't know that this existed. That is taste. There will always be taste. And once you see that taste, it's possible you could reverse engineer a prompt that gets to that taste. But even that is only possible for the people who walked into that home. And if your accounting firm becomes that home for taste where people don't even know what they're missing until they come and work with you, then what we're left to do when AI reduces the production to nothing is be purveyors of taste, a specific perspective. Hey, you're running an HVAC company. Here's a whole bunch of stuff that you may not be aware of. Here's what some people are doing and having success with. And the AI people will be like, well, AI can have taste too. You can make AI do anything. And that's true to a degree. But for the people that never walk into that home and have that moment of, oh dang, I didn't realize this was possible, they would never be exposed to that taste in the first place. They would never be able to get AI to create that version of taste, right? And so it's raising the floor of like how tech-enabled a firm can be for sure. But in many ways, this is already what accounting firms are because under the hood, aren't most of us all kind of using the same apps anyway? If you explain what an accounting firm was to someone pre-computer and you explain the future and you're like, yeah, we're all going to use these computers that can like do math like crazy and there's like 20, 30 apps and pretty much every accounting firm is using this more or less same combination of apps. This person would be like, so aren't they all just doing the same thing? Like, why don't they all just get consolidated? What, like, what makes them different? You would be like, well, I guess the difference is all the humans on top of those apps. It is how the humans serve people and how they communicate with the clients and what their perspective is and how it's different that makes that accounting firm unique. And when you have an AI that's very, very good and can do the work, what remains? Well, it's all the humans. And their perspective and their taste as to how the work should get done. There's a very strong argument that on the other side of AI, humans are more valuable, more scarce than they've ever been before, because that's all that's left. That person that was in the office that just cranked for 10 hours a day that the client never even saw, they didn't even know they existed. They didn't understand or see any of the work that they did, but it was part of the end product that was handed off by somebody else. When that role goes away, When back of house is reduced, what remains is not the doing of the thing, but the, what do you have to say about the thing? What is your perspective on that thing? And that, I actually think that sounds pretty fun. That is a product of, of having 100 human conversations that that other person hasn't had. That is a snapshot of, of your individuality and your take on this very nuanced subject and an escape from the mean, from the average, right? And so I would argue what remains is creativity. There's perhaps a greater emphasis than ever before on, do you have anything to say? Because the computer can, can get you to the trad, the average, but what's your take? What are you adding to this? Do you have something to add? That sounds fun to me. That sounds like an escape from the tedium, from the menial onto something that feels more human and more interesting, or just a fanciful way to sugarcoat end of days. Maybe we'll see, right? Either way, you and I, we're gonna be hand in hand staring into that, that creative oblivion together, aren't we? Friends, thanks for being here today. Got any questions? Pop that stuff down in the comments. We're gonna do this again, okay? And I'll see you there. Take care.