614 The Three Types of Accounting Firm AI [And where to deploy each in your accounting firm]

Listen up, 'cause I hear that Jason on FIRMS is going on tour. Buckle up, it's the show you love now with snacks and an open bar. The tour's halfway through, but we still need you. For tickets and more, head to JasonOnFIRMS.com. JasonOnFIRMS.com. Now it's time we end this song. On to the show! When people talk about AI today in accounting firms, they're oftentimes talking about wildly different things. Today we're exploring the 3 most likely timelines for where all this stuff goes, the types of AI we end up using, where those are at today, and the pros and cons of each. Plus a bit of accounting firm strategy. How do you get off the hamster wheel of always being overcapacity and then hiring but still being overcapacity because you got to pull in clients to, to keep that hire busy? We'll talk about the best ways to hire an AI expert to support your accounting firm, plus maybe the biggest announcement we've ever made. I'm the co-owner of a new company. Action-packed episode today, gang. Let's do it. Okay, AI. What are people talking about when they say, uh, AI is going to be the future of XYZ in the accounting profession. It's 100 different things. Our software vendors are explaining it as one thing, but in practice, most people, what they're doing, they're using Claude. Like, it is not the AI built into a platform. It is the generalist AI. So who's going to win? Like, what does the future look like? We're going to look at 3 different versions of this, all of which can be played with today. But understanding this actually has a huge impact on the tech buying decisions we're making right now. So number 1, first type of AI that folks are really going hard on right now, generalist AI systems. Historically, that was ChatGPT. Now for almost everyone, that is Claude. I'm in the middle of a tour of 20 cities in the US hosting events for accounting firms. In every city I ask how many people are using Claude at least once a week, and then how many people are using Claude at least once a day? Probably half of these rooms, people are reporting they're using Claude at least once a day. Half of these rooms. Like, there's very few things in the accounting space that have that level of saturation. And obviously we are attracting, you know, a certain type of demographic out to these events. But there are people who believe that the future is just this. It's a generalist AI assistant and you kind of won't need to like really have anything else from a software standpoint. These things are good enough to make up a bulk of the tech stack just themselves. That's what a lot of folks believe. Maybe they're not there today, but in the near future you could see a world where they were, and anytime you need a specialist tool, it just kind of like whips up that tool, almost like this just-in-time approach to software where it just builds what you need when you need it. Now, is this realistic or not? Time will tell. I've definitely got opinions on this. I will say right now, it is shocking how far we can get just with Claude, but doubly so when you layer in open source libraries. At the time this podcast comes out, hopefully you've seen a video we just released on the main YouTube channel around how to replace QuickBooks with Claude. I know that is like scandalous to even say aloud. That's not possible. Here's the thing. What makes that possible? What makes Claude a vastly more powerful accounting system is actually pulling in a couple of open source libraries, stuff that's totally free where you can literally just utter its name into Claude and Claude goes out and finds the source code and can now use that stuff. Stuff. A simple modification of Claude with something like this, at least as far as making it, you know, an accounting ledger goes, is a massive improvement in what we can actually use Claude for when it comes to doing accounting for our clients. And so as we're considering like, all right, are these generalist AI assistants actually going to be all that helpful? It's not just the AI assistants themselves. It is the vast world of open source software that's out there that it can leverage to get a job done. And there is an open source version of almost everything from accounting ledgers to practice management systems. They are usually not, you know, as good and as dialed in as the proper sort of leading platforms in a space. But when you have a tool as powerful as Claude that can customize it for your needs, will that get you far enough? A lot of people think right now that it will. Now, historically, these generalist assistants, they just lived in your internet browser. This was ChatGPT for the longest time. I mean, the big breakthrough we've had in the last 12 months is Claude Cowork, which works on your local machine, on your local files. All the conversation history lives locally. And so it feels like something that is a little more yours and like working on your context. Another huge breakthrough here, which hasn't come to business plans yet, and even when it does, it's going to be a really challenging thing to roll out securely. Another breakthrough here is computer use. The ability of the tool to use your mouse, use your keyboard, have access to literally any application on your machine. It's one thing for it to kind of work on your desktop and be able to access your files. It's a whole other thing for it to be able to use the apps and all the systems on your entire machine. In reality, like what, like while this is very exciting because part of the appeal of generalist AI assistants, I think, is that they sort of work most similar to us humans, where they can work across different apps. And maybe that's a different way to visualize automation when it's like, oh, it's just doing the work the way that I do. But if we, I mean, if we consider like the risks and the fears around an agent potentially making a mistake or getting tricked into disclosing something, plugging that agent into like be able to use everything on your machine is just about the most risky way you can deploy an AI agent. I mean, you've got the keys to the castle. Like you have the most access. Realistically, that's not the first place that we would roll out a desktop agent. In practice, that would probably look like having a machine that is kind of this sandboxed environment that has very limited things on it from limited apps to maybe even limited parts of the web that it has access to. Hey, maybe that's the future. Oh my gosh. Big brain idea alert. Here it comes. You know, I, I don't know, like a Wi-Fi network, you can lock down the internet so your kids can only access certain things. Ah, is this the future of rolling out AI agents and giving them access to the web? Is this firewall that's like, no, no, no, you're not allowed to access that website. I'm kind of joking. That's actually a tremendous idea. Right now inside of Claude's organizational settings, you can identify any websites that the agent should never be able to go to. But the more useful version of that for me would actually be the opposite. Let me identify the websites it can go to. Like, don't, don't make me put every website on the entire internet. Like, let me just say this can only go to, you know, bluejay.com or QuickBooks or something like that. When it comes to putting an agent onto a machine and that machine has limited access to things, that I think is the most realistic way that we're actually going to roll out like desktop AI agents that are actually using a computer of their own. Now, Pros and cons of generalist assistants and kind of some arguments for and against how this may or may not be the entirety of what the future looks like. Pros, we kind of touched on, it works kind of like an accountant does. Like what we do right now is we're sort of like human integrators. We do all the stuff that slips between the cracks and we take stuff from app A to app B and we wish those things integrated just the way we wanted them to. But the reality is right now they don't and so we are left to do all that work and chase down any of the humans that have information that needs to be taken into account in the process. This to me is the most interesting thing about generalist assistants is they can work across all these different apps the same way that you and I did. It's a, it's a solution to a problem that I wish didn't exist. But the fact remains, a lot of accounting stuff is putting stuff that's in one system into another system. And that is not a sexy thing to say. And oftentimes there's some logic involved in, well, what, what number should be taken out of that system and doesn't need to be modified before it goes into the other system. But this is why we have so many monitors on our desk. Like more of what we do right now than we would like to admit is taking stuff from one thing and putting it into another thing. Generalist assistants that work across the top of our apps, they can do this as well. This is kind of this agent layer that sits across the top of all the stuff that we already use. Now, the biggest con with generalist assistants like this is deployment is hard. How do I deploy this securely? There's a lot of nuance to this. And even the idea of giving it access to all my apps, like, is there inherent risk in that? There is. This is almost a new, I mean, not almost, it is a new layer of tech that comes with a new level of security challenges and how do we roll this stuff out responsibly? We talk about that a lot on this podcast, like more of the nuance around the right and wrong ways to do this. But this is and will remain the biggest argument against, at least in the near term, firms being able to really go hard into these generalist assistants is they just can't get their heads around making sense of the security side of it. Now that is the first type of AI. But if you listen to like software companies right now, that's not what they're talking about. Like our software leaders are not talking about the power of clawed because if anything, they see it as an existential threat. And so we're gonna talk about the AI that is built into the tools that we already use and whether that is the answer for the future or not. But before that, there's another one in the middle and, and there's a very good chance that this sort of version of AI in the middle ends up being what takes us through the next 5 or 10 years. And that is a generalist assistant that is making heavy use of MCPs. So an MCP, if you're not familiar, It is a way for your AI assistant to connect to the other tools that you use, not unlike how APIs were how, uh, you know, and remain how software different platforms talk to each other. So if you pay a bill in your bill pay software, it pushes that to the Ledger via the Ledger's API. An MCP is an AI assistant version of the same thing. It's kind of a more natural language conversational way to get information in and out of other systems. Now, Not all MCPs are made the same. For example, QuickBooks. Intuit has released an MCP for QuickBooks that anybody can use, but it was made for the normie small business owner and is very, very limited. So if you connect the MCP to your cloud right now, you can connect it to a single company, which isn't realistic for the way that accounting firms work, right? Again, this wasn't made for accounting firms. It was made for your normie business owner, but also It is a read-only MCP that has very, very limited access. But the power of MCPs is they can be read, but they can also be write. And so what does a write MCP look like? That actually looks like Claude going and being able to do real work in that platform. And so a great, an exciting example of this, I think we've talked about this last week maybe, Qik is like an AI-first accounting ledger, just to kind of stick to accounting ledgers as an example. Kik launched an MCP that is both read and write. What it— like, what does that mean? An accounting ledger with an MCP that's read-write? It means you can go to Claude and give it like a pile of bank statements and be like, okay, now go and do the accounting, go do the bookkeeping in Kik. And it can not only pull information from that platform, but it can write information to that platform. So it can classify transactions. It can import transactions into the accounting system. That, I mean, you wanna talk about baking your noodle, like this is an entirely different level of capability from AI agents where they are totally good enough to do this stuff right now. The problem, the limitation has been like the interoperability, like their ability to interface with other programs. MCPs in many ways are the best of both worlds. You've got a generalist assistant that's very capable and it's doing for you all the things that we talked about kind of in the first path for AI,, but can also work your apps as well. Now, limitations here might be obvious as well. What if the apps don't want to play ball? QuickBooks being an example, who has the most to lose in this timeline? I would argue it's probably the incumbents, because what happens as soon as there's a really good MCP connection with that piece of software? You're probably not logging into that piece of software as much as you used to. In fact, you know, maybe 80% of the times you would have logged in before, now it might just be the agent going in there. That detail you need, that pay stub, that whatever it is, your first point of contact is probably going to be Claude and saying, hey, go grab this thing for me. And it just does it. That is an example of where you are no longer logging into the platform itself to do that thing, but you're relying on Claude to do that thing. Who has the most to lose from this? My argument would be incumbents, because as soon as we stop logging into a platform, the perceived value of the platform I think it probably goes down. And this is an opportunity for new entrants into a space, but a huge risk to incumbent folks in the space. And so what if these platforms just refuse to integrate? Let's start with QuickBooks as an example. They launched a very limited MCP, right? Well, there's already people building their own MCPs for QuickBooks. You can actually go to Claude and build your own MCP. And so what people are doing is they're looking at the API for QuickBooks, which is the way that other apps have historically interfaced with QuickBooks, either pulling data from it or pushing data to QuickBooks. You can build an MCP that sits on top of kind of the full developer API of everything that could be automated in QuickBooks. And everything in that API is much more complete than what QuickBooks now has built into their MCP. So people are basically building their own better MCP that makes use of the entirety of the QuickBooks API to not only read information from QuickBooks, but be able to push information to QuickBooks as well, and connect at the QBOA level. That is the accountant level where you have your whole list of 500 clients instead of the QBO level, just connecting to a single client. And so a more beefed up MCP, it can access any of those clients and it can do anything that is allowed in the API., which is fairly robust when it comes to QuickBooks, but stops short when it comes to like the matching screen, like posting transactions, that sort of thing. They've never opened that stuff up to the API, but still, that's a vastly more powerful MCP, right? So as it stands, anyone can go out and build their own MCP that will get them as far as anything the API will allow for. I know there's a lot of acronyms. I got a little bit nerdy,. But here is like kind of the next step of this, which is even easier to understand, is, is to say, okay, what if a company won't even put an API out there? A very, very closed, like lockdown system where they're like, we're not willing to integrate with anything else because we want to build all the best stuff inside of this platform. There are very real trade-offs with that, obviously, because you're not gonna be the best at everything. Goodness. But in an age of AI assistants and MCPs and this, The idea that I want my agent to be able to access all my systems. What will the defense be for these companies that want to be fully locked down? It, it's going to be tremendously hard for them to maintain this position because AI agents can log into systems and use them as a human user would as, as kind of a worst case scenario because it's very slow and very clunky for them to do. But you can still log an AI agent into any system. It can navigate it the same way that a human user would and do anything that is allowed for inside the app itself. And so if this is possible, like what argument remains for that software company to not be like, okay, we just need to give you an MCP because a whole bunch of bots banging around in our platform is way less efficient than just giving you an MCP for those bots. And people will say like, well, wait, what? It's against the terms of use for that software platform. For a software company to start kicking people off its platform because their AI agents are using it. Is tremendously risky. They already have the risk of, well, why would I just not go and use the alternative system that has an MCP? Very obviously, I want my agent to be able to interface with your platform. That's why it's logging in and using it. But to then take the extra step of I'm going to kick you off my platform because your bot is using it. While it may be against the terms of service, I think that is exceptionally unlikely and will just create negative PR and force people to that other platform that will allow for an MCP. As I read this ad, I am feasting my eyes upon talented accountants that you could literally, literally book an interview with right now. Like Ruru, a CPA, a tax pro, $3,500 a month. He's got experience with CCH Access, bless your heart, CCH Engagement, QuickBooks Online. Gang, he is one of many accounting pros that you can peruse at the Cloud Accountant Staffing portal. We'll link it down in the show notes. It is a veritable smorgasbord of accounting talent, folks that they have pre-vetted where you can go in, you can see, see their experience. You can literally see an intro video that they've recorded and you can book an interview with them without ever even having to hop on a call with Cloud Accountant Staffing. Amazing. We got tax pros, bookkeepers, auditors, executive assistants. Oh my gosh. I got a lot of people in here right now. 45 candidates, because I know you need that next dream hire. You really do. What do you got to lose? Okay, go check out the portal, book an interview. Feels like a win-win, right? You're only going to move forward if you find somebody you love. Cloud Accountant Staffing, run by folks that used to run an accounting firm. They make it as easy as possible for you. To learn more, check out the link down in the show notes. Hey, if you're done licking your wounds from tax season, if you run more than a few hundred 1040s a year, you got to 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. 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. So honestly, right now, I think it's kind of not even an option for people to not have an MCP on their roadmap. Now, how robust of an MCP that is will become very, very important. You can imagine if it's read-only, That's frustrating. That's limiting. And there's this balancing act here where, like, you want to mitigate the risk of an agent breaking stuff and doing things that it shouldn't inside that system. But you look at a company like ADouble who is very quick to market with an MCP, maybe the first one that I saw in the accounting space. Their version 1 of the MCP let you control what aspects of the platform do you want your agent to be able to access. You can turn all that on and off. And should it have only read access or should it have read and write access? You can set all this stuff up yourself. You can start on the training wheels version and, and give it more access over time as you get comfortable with it. But that's an example of a really well done MCP that gives the user the control over exactly how much they want to let the agent go in there and bang around. And right now there is nothing stopping the other companies in the space from doing the same. Now it's not to say there aren't like some, uh, technical challenges in doing so, but I think the business case for For a company to be like, no, we have, we're not gonna allow for this. I think that's pretty much out the window now because agents can just log in and use software. I mean, literally you could be away from your computer interacting with your Claude and from your desktop, your Claude can go and log into that system and fetch you information. Like, like what? Like why would those companies continue to not give you an MCP? At that point, all they're doing is just making your agent slower. That's annoying. That is like kind of disregarding the reality of the world that we're living in right now. Most likely to take that position are the ones who have always said like, we're gonna be the all-in-one, we're not gonna put an API out there. But maintaining that position long-term just feels wildly unrealistic to me. But there are other like even more peripheral tools that I wanna see greater MCP support from. BlueJay Tax, for example. In the US, BlueJay is probably the leading tax research platform right now. But this is something To me, where unless they start doing some net new novel stuff, like getting into maybe technical review or something like that, the research component of that platform, I'll be honest, I just want that as an MCP. I don't wanna have to log into another app to do that. What's more realistic to me long-term is that Claude, when it is working through a problem for me and it needs a technical tax answer, will go to its BlueJay MCP connection to get a higher quality answer, that's pulling from authoritative sources rather than going to the web and pulling an answer from NerdWallet and then like continuing on with that answer, right? That is an example of, I think, an app where at least in its current form, I want, I'm more than happy to pay for because I do like kind of locking down the agent to just the authoritative sources, but is not something that probably justifies this entirely separate login when an MCP connection can handle this just as well for us. Now the pros of kind of a future timeline where we're all just using our generalist assistants with all these connectors. The pros are it feels like the best of both worlds. You can pull in other apps when you need them while also having the simplicity of doing the bulk of your work in your AI agent. Another pro is this is a positive timeline for software companies. If we go back to the first timeline and we rely on our assistants to kind of build software as we need it, you know, it does all of its tax research on its own or something like that. That, I mean, a lot of software companies are in jeopardy in that timeline. And so this feels a bit like the best of both worlds. And there are the really extreme AI people who are like, oh no, all this just goes to zero eventually. Everything goes to the AI assistant. I don't know. Like, I don't know if it's like Luddite thinking to think that's not inevitable, but I have a, at least in the short term, like 5 years, 5 to 7 years, that feels unlikely. Now the cons of this approach, is MCP support. Is how realistic is this? Is everybody in the space going to play ball or am I gonna be stuck with a bunch of systems that my AI agent has to log into to get information from it? When we came into the cloud accounting era, everybody said this will solve integration problems. Everything will talk with each other now. This will be perfect. And we won't have a bunch of accountants putting data from system A to system B. Did that ever come to bear? It really didn't. We have integrations for sure, but we are so far from a fully integrated tech stack. And even the integrations that exist, you know, like there's so many limitations. So maybe it is naive to say that we're headed to a timeline where everyone will just put out these really good MCPs and your agent can just use all these tools. I will say even in the more negative timeline where maybe these companies don't put out very good MCPs, there will be more pressure on them than ever before to do so. And you will have entrants in the market that are like AI first, like they are designed for for the agent. There's many people right now that say if you're building software in 2026, you're either building agents or you're building software for agents. That's it. To start a software company today that by design is going to build users by humans logging in and using your platform exclusively, honestly, is probably not realistic with where we're headed. And so even if a company does take the position that we're not going to do an MCP, there's going to be a tremendous amount of pressure and all sorts of alternatives that are more AI-first. Now, the other con with this approach, the notion of your AI assistant using a whole ton of other, uh, MCPs is potentially, how do I know when I still need that app or not? This is another question I've been asking out at my live events is how many people have already replaced apps in their stack with AI? Quite a few hands go up. I've done this myself. And so if you're paying another subscription for something that's just an MCP, is it tricky then to know like, is that still worthwhile or could Claude just build a version of that for me. Time will tell. Last type of AI, third type of AI, and this is the type of AI that all of our software companies are talking about, but is largely not the type of AI that we're talking about on this podcast, is the AI that is embedded inside of the tech that you already use. Let's take a really bad example. QuickBooks, the built-in AI assistant in QuickBooks. How's that treating you? Pretty impressed by that? I mean, this suffers from the same thing as the QuickBooks MCP where it was not built for the accountant user. It was built for the normie small business user. And so it's very hard to make a version of that that's valuable to you. Also, when you're a power user in a platform, oftentimes, like, you will almost certainly recoil when there is an AI agent because you're like, I don't need this. I already know exactly where to go to get all this stuff out of the system. This, this AI layer is kind of just an annoyance. But you could see a version of that where it is more, more valuable for the lay user who's not the expert in the platform. But also that person has a pretty strong argument that they would rather just do that stuff in Claude, right? Not go to QuickBooks and ask the QuickBooks AI assistant, but just go to Claude. Now, long term, will most of the AI that we use be the AI that is baked into, say, your practice management system or all the software that we are already paying for? Arguments for this timeline, way easier to roll out. Vastly, I mean, you could flip the switch on AI that's built into a platform right now, and that is a super easy thing to deploy. Now, it is a bit a misnomer to say that this is more secure. This is like one of the most common arguments against, say, cloud adoption right now is people are like, this doesn't feel safe. I don't think my clients would be happy with me using quote unquote generalist AI. I'm going to wait for AI to get rolled out in the tools I already use. I mean, the reality of what we're saying here is like when, when, say, your PM, for example, uses or has AI built into the platform, that is not their own AI model. Like everybody is using the same LLMs behind the scenes. They are making calls to the API of Anthropic to use Claude's models or to the API of OpenAI to use their models. And so if we say, I'm just going to wait for the AI that my tools deploy because that's more secure, in reality, when we use that AI, we're giving our data to the platform, say your practice management system, and then from there it goes to the API., which is the exact API that Claude hits when you're doing work in Claude or that ChatGPT hits when you're doing work in ChatGPT. So using it through the tools that we already use, you can make an argument that's actually sending that data through two different platforms rather than just one if you're using Claude or ChatGPT directly. So from a security of like the large language model standpoint, that's a pretty weak argument. Where it is a stronger argument is kind of the net new risk of an AI agent they can just go around and do stuff and can potentially get tricked, can potentially disclose something externally that it shouldn't. This is like the most important security challenge to be talking about with AI right now. And this is the biggest pro of embedded AI is when it is only inside of that system, you don't have something. I mean, you could, but in all likelihood, you don't have something that is going out and using the rest of the web, logging into other systems. It only exists inside of that one system. Where this is likely most powerful is in the systems that have their fingers into the most context in your accounting firm. That's probably the practice management system. That is the people that have been going hard on all-in-one. And so in the last section when we talked about, you know, a generalist AI assistant being very reliant upon MCPs and all-in-ones potentially locking down and saying, no, we're not going to give you an MCP, those same people are the people who could build the most capable AI agent inside of their platform because it probably contains the most data about your accounting firm. Now, this is— of all of these different paths, this is the easiest thing to roll out tomorrow. And we do have to bear in mind, if you're like a big brain AI nerd, and if you listen to this podcast, you probably are to a degree. If you're that person, your team, in all likelihood, they are not those people.. And so putting these big sort of open-minded Claude ideas in front of them, oftentimes people will just be like, man, I don't even know where to start. I just know I gotta get this work done. Like, tell me how it can help me with that. This is a good argument for embedded AI being a more comfortable entry point for your team who's maybe not already just like staying awake at night thinking about all this stuff in, in like an exciting way. So the biggest pro is it's much, much easier to deploy. The con is that its context in some ways is inherently limited because it's designed to just work inside this platform. Now it doesn't have to just work inside that platform. It could go and work in other apps just like a generalist assistant does. But as soon as it does, you lose the pro of it being easier to deploy because now it's out banging around doing stuff on the web. Now this one, I mean, will come to bear and it is already coming to bear AI inside of the platforms that we already use. If I have a concern here, it's that software companies control so much of the mindshare in our industry because they have big marketing budgets and are responsible for the majority of education in our space comes from software companies. And all of that is coming through a very specific lens where they want you to be excited and thinking about the AI they have inside of their platforms and how it is quote unquote more secure and all that. And in some ways it is, and in many ways it is easier to adopt team-wide, but what that kind of flies right past is right now the reality. Well, come to a room of accounting firm folk right now and say, how many people are using Claude every single day? And half the room raises their hand. That is a completely different type of AI adoption and a completely different type of AI than software companies are, are telling us about that they are introducing in their platforms. I think they both have their own pros and cons, and I'll be totally honest, of these three timelines, I don't know. I don't fully know where we're gonna end up, if it's gonna be different based on, well, what's next 12 months, but what's next 5 years? I don't know if one of these will end up being the runaway winner. I don't know if our profession will end up paralyzed by the security fears of AI agents for the next 5 years. It's hard to say, but those are 3 wildly different things. I do think timeline number 2, maybe we aren't thinking and talking about enough. The notion of your generalist assistant just being a connector for a whole bunch of different apps. In fact, As you're making software buying decisions, if there's one kind of buying decision piece of advice to come out of this, it is the importance of not just MCP, but really good read/write MCP in the tools that you're picking. You can imagine changing, say, practice management systems right now and putting all of your data into this system that decides we're not going to open it up for anything. I mean, that just handcuffed your accounting firm big time, right? Like, that is a huge, huge risk. And so if there's one thing that you ought to take into consideration as we are kind of in software buying season right now, it is what is this company's positioning towards? Building AI inside their product is one thing. That's inevitable. If anybody's not saying they're doing that right now, that's a red flag. The more important question for you to ask is, okay, but are you going to build a really good MCP? Because it is risky to go all in on the embedded AI timeline. You don't want to go all in on a platform like that when there is a very good likelihood in 12 months most accounting firms will be leading their work from cloud and the connectors that Clod gets into. Hey, I know you love Clod Cowork as much as I do, but have you heard the news about Canopy Coworker? Subtle. It's a new AI assistant they launched inside of Canopy, one of this episode's sponsors. Canopy, a practice management system that is a feature nerd's dream. They're trying to cover as much of everything you need in an accounting firm in a single place. But as we talk about on this podcast often, it is much easier to roll out an AI agent inside of a closed system than it is to roll out an AI agent that works across like all of my systems, right? That's the trade-off. I want the AI that works across all my systems, but we're struggling now with like, what's the most secure way to do that? The AI agent that works inside my all-in-one system, buddy, that is easy. Flip the switch, count me in, strap in, buckle up, I want it. Canopy is so convinced that you're going to love what they got to show you, they're going to cut you a check for 75 big ones. They'll give you a $75 gift card. Gift card to Bed Bath Beyond. What's that? Okay, I'm not sure where the gift card is for, but if you book a demo with Canopy via getcanopy.com/jason, I'll put that link down in the show notes, they give you a $75 gift card just for coming and demoing. Do it. What do you got to lose? Hey, real quick, the most important question accounting firms are asking right now in 2026 is how do I use AI securely in my production workflows? I've said this many times, don't take security advice from a YouTuber. What you need is actual, like, proper IT help, like this episode sponsor, Farodil. If I had to shoot from the hip, uh, with what I've seen from folks, Farodil might be the fastest growing IT solution for small accounting firms. I say small, I mean that like 100 employees and under. You need an IT group that just works with accounting professionals. Otherwise, you are paying this IT group to learn accounting firm things. Like, I had a local IT guy that I once brought over and spent 2 days watching him learn CCH engagement, like how the, like, database worked on my server. Oh my gosh, please don't do that. Work with an IT group that only supports accounting firms, particularly right now as it is so important that we are beginning to implement AI effectively inside of our firms. You want to do that responsibly too, right? You need better help with this stuff. Look no further than Barredo. They do hosting. They'll be your entire IT group. They'll do your whisp. Can't recommend them enough. Check out the link down in the show notes to learn more. Now, I have a hard time believing in like fully in the first timeline where all we will need is generalist assistants. I mean, for sure, cloud will, I think, replace some of our sort of peripheral tools, and it already is doing that. But do you still need a practice management system for complex companies? Do you still need an accounting ledger? I think in the near term you almost certainly do. Otherwise, What is the organized way that you can store all this stuff? For sure, there's, you build, I like, your own backends for managing all of this stuff. And the super nerds will probably do this, but that is not accessible for most. Right now, we're kind of in this messy period where a lot of people are really excited about Claude, the generalist stuff, going all in on that. There's a lot of people the other end of the spectrum that are like, I'm waiting, I'm waiting for my tools to deploy AI for me. And more people getting excited about the second approach, Cloud with a whole bunch of connectors, but we're sort of waiting for the market to catch up. I, I mean, applaud Double because they were very quick with an MCP. Anchor also now has an MCP in general availability that is both read and write. Anchor is a proposal platform. That's really exciting. You literally go from a meeting transcript to a proposal. How cool is that? So really take this into account as you're making tech buying decisions. And I'm curious your take. In fact, let me know down in the comments, what do you bought into? The timeline that it's just like all generalist assistants and open source software, is that the future? The second timeline where it is a generalist assistant plugging into a bunch of other tools, or the third where you really think it's going to be hard to get our profession just running on generalist assistants, maybe for security reasons or otherwise. And so what makes more sense to you is the embedded AI that's inside of a more traditional tech stack. Let me know down in the comments. Hey, I mentioned I've been on this tour. Let's take a little stop in Nerd Corner here for me to catch you up on some really, really cool AI use cases that I've seen in the wild. It's Nerd Corner, Nerd Corner, let's nerd it out, Nerd Corner. I'm gonna give you 7 use cases that I think are really cool that I've heard from folks just in the last couple of weeks. These are real use cases submitted from actual accounting firms. Number 1, Claude built out a fleet tracker for an executive transportation company. Think like limousine, that sort of thing. To track vehicles, their trips, and allocate proportional fixed and variable costs. A fleet tracker. You're like, I'm an accountant. Why? I'm not going to build a fleet tracker. You know what this is? Is advanced reporting. That's all this is, is taking all that financial data and creating a new and novel way to visualize it. And this is so darn cool. Like if you're trying to get more into advisory, doing CFO, this is like what the really, really good version of this looks like. Is what is a novel way to produce, to present financial and non-financial information that will be actionable, that will give us some really valuable insights. Now, there is a risk here. In fact, I was sitting at a table of accountants just last week who were talking about like, have you guys successfully been able to get clients to like actually pay you more for advisory work and kind of batting that around? And one person shared the story of they went through all this work, all these lengths to put together this really, really good reporting for a client. That was like, I think it was a contractor, and they're like looking at all this data and they're like, very clearly, these are the projects where you're crushing it and these are all the projects that are bleeding you dry. And they gave all this information to the client. They did all this work and you know what the client did? Absolutely nothing. They kept doing all those different types of projects. They didn't double down on the more profitable stuff. And the whole table's like, yeah, man, clients suck. They're so dumb. But then it's also like, yeah, Okay, well, what if we hired a coach and somebody did the same for our accounting firm? And you may not even need the coach to tell you this. You might already know which ones you're crushing on and which ones are bleeding you dry. But do we change or do we keep doing the same thing? That's, that's all much too existential. Let's go back to the AI use case. Fleet tracker. Cool. Okay. Second use case. When someone fills out our web form, Our agent will research the individual and the company, so we have a ton of background info. That is very cool. So your intake form, still in your intake form, ask all the qualifying questions, the stuff you need to know to decide, do I actually want to take a call with this person or not? Or do they even get access to my booking link? But then from there, have an agent go do some, some deep digging, some sleuthing, a whole bunch of context, but also like, Is there even problematic stuff about this individual? Pull up the mugshots. Get me all that stuff, man. Maybe that's bad advice. I don't know. Use case number 3, we used it to improve our pricing calculator. We have talked pricing calculators at length on this podcast in the past. They're a divisive thing. I think what's more interesting here is you can kind of use AI to improve just about anything. If you have a pricing calculator, if you're using this,, but AI could help you make some improvements to it or turn a spreadsheet into a tool, a, a more capable tool, something that can't be done inside of a spreadsheet. Let me give a call out here to specifically artifacts inside of Claude Chat. So not Cowork, which is the version of Claude that does all the work on your desktop, but Claude Chat, which can be used either from the Claude web app or it can actually be used on the desktop as well. In the top left, you just pick Chat. The cool thing about artifacts in there, which are kind of like mini apps you can build with Claude, artifacts in chat can be shared with anyone else in your organization. And so inside my company, we've got a number of cool little apps that people on the team have built, and then they share it with the other people on the team. Pricing calculator, or really any spreadsheet that you use, you could build this into a little app with Claude. Literally just give Claude the spreadsheet and say, build this into an artifact. Like, that can be the entire prompt. We should make a video out of that. Hang on, writing that down. Okay, it's been written down. It will turn it into this little baby app and then you can continue the conversation and have it make improvements and all that. But then share that with your team. That's super cool. Use case number 4. Had Claude build my firm's website and Google Ads landing pages. Snipped the images of the site it built in HTML, hired someone on Fiverr for $550 and they built the entire 15-page fully functional website using the images. And the ad landing pages actually converted to booked calls. If you haven't seen yet my video on the main channel about Claude Design, it came out a couple of weeks ago. We built a full accounting firm website with Claude Design. The video is very silly, but the website that it made was so, so good. We pull in some concepts that we talk about on the podcast here around framing what you do, not through the lens of the services that you sell, because that is a race to the bottom as soon as I say, I do bookkeeping, I do tax prep. When they don't know the difference between what you do and the guy across the street, they're gonna price shop. And so instead, we need to sell the opportunity, the notion that they can give me $1 and I can give them back $2. How do we frame what we do for a specific person through that ROI lens? Buddy, Claude crushed that. The website was so cool. It had all these examples of like, oh, the average contractor XYZ, they lose money here, they lose money there, and then it like all built up to this, ROI calculator, and it was just such a great visualization of escaping the normie accounting firm website that is just this list of services. The bar has absolutely been raised for what anyone can build into their accounting firm website because Claude can build fully functional sites. This person, they used someone on Fiverr to build out the site. I mean, you can have it build the code and deploy it yourself, like, like with a little bit of technical banging around, and asking Claude, how do I do this? How do I host this code that you just made? It's totally doable. And then you can improve upon it. Like you can keep pushing updates to it yourself. All right, use case number 5, 3 to go. We use email notifications from our practice management system to trigger actions in Claude code. So PM sends out an email notification when something happens. Wouldn't it then be nice to automate something that happens next? Now, specifically what they're using for this, we talked about this a few episodes ago, and this is a little bit more of a nerdy Claude use case, is they're using routines inside of Claude Code and routines in Claude Code right now, which can all be set up in the desktop app, are the easiest way to trigger an AI workflow from an external service. So when you set up that routine, I mean, the easiest way to trigger it is probably with a webhook. I won't go any further into like that technical nerd hole. If you want to, you could literally just repeat those words to Claude and I'll walk you through how to do it. But where in Cowork you can do scheduled tasks that run on a calendar schedule, that's not great when I need something to happen immediately when something else happens. The easiest way to set up that, set that up right now is routines in Claude Code. Use case number 6, extracting stock trade data from a 1099-B PDF. I mean, it's 2026. It's really annoying that that use case is exciting as it is, but it is. You know it is, unless you're not in the US. Which brings me to use case number 7, replacing platforms by moving work into cloud and having better results than the platforms I was paying for. I've been surprised how many people are raising their hand for this at the live events. They have already replaced systems that they were previously paying for with cloud and are not looking back. And that's not gonna stop. Like, we're just gonna keep going that way. Now there is almost a net new type of software being built right now, which is what is the software I'll pay for that is something I'm just buying for my agent? Might be a system that I never log into. And so there might be some net new spend here, but spend on this in all likelihood is vastly less. It is usage-based, not per user, and it is making my agent better. It is like a utility belt for your AI agent. Rather than your agent having to like throw a brick through a window and figure out how to get into the tools that you use that were built for humans. That is a really good metaphor. Did you, wow, we just had a moment there. Okay, I can hear something outside. What's that? Okay, nope, mailman's here. Let's do it. Mailbag! Mailbag! This segment is called Mailbag! Questions are sourced from social media DMs, from my email, from my cellular telephone. However you wanna get these questions to me, fire 'em over, okay, bud? First up, how do you get off the hamster wheel of having a team at capacity and then hiring and then getting more clients and then having the team at capacity? Just repeat that until the end of time. This might be the biggest, most existential question in running an accounting firm because most people go through this progression from somebody at some point starts a firm and they're like, I'm just happy to have clients and they're trying to put food on the table and saying yes to everything. And then they hit capacity and you're like, crap, in order to fix this problem, I actually need some time to work on this problem, but I can't because I'm at capacity. I know what I'll do. I'll hire somebody. So now you have someone join the team and they could take on some of your work, but you now have to manage and train this individual, which is yet another job. And you're like, why do I, when am I going to be able to catch a break? And then it just never ends. Now the biggest problem here, the mindset that perpetuates this issue, is hiring for growth rather than if you are the big boss hiring for your job. If you are at capacity on client work and you think the solution is to hire another human being, what you haven't done is accounted for the fact that that human being is going to take a bunch of work to find, to train, to get up to speed. You better hope they don't flame out after 6 months. And that is a net new job for you. Maybe they can take some client work off your plate, but if all you ever do is hire those people, We aren't solving for all the jobs that you have. And yes, like, is part of the job that you have doing the client work they're taking on? Sure. But if you haven't learned this lesson already, they're not going to do it like you will. It will probably take them longer because they will do it in a way that you don't expect. And then you're like, well, dang, now I got to build systems and now, now I got to review this stuff. Don't lose sight of the fact that you have a whole bunch of jobs and you also need to hire for those. More the jobs around the running of the business. Maybe the job's around managing people or training people. Simply hiring another body to do prep work, that just creates even more of a monster to manage. To me, the only way I know how to get this under control is to step back and say, what is the type of business, the size of business that I want to run? And first you have to consider, am I trying to build an accounting firm that's just built around me and around my identity? If so, most people will hit a wall at $1 million in revenue per person. If you really take this to the extreme, you could probably push this to $1.5 or $2 million of revenue around you. There is a more alternative version of this where you are sort of the visual, almost thought leader. And this is almost more like a creator brand, like what I run, and that you can scale to $5, $10, $15 million, but you are fully out of the client work at that point. Most people, they just want to build the firm around themselves, at least initially. And if that's the case, you're probably not going to build a team of more than 5 or 10.. But in that group, you need 2 or 3 people that can not only do the technical stuff but manage people, because 10 people being direct reports to you will leave you no time to do actionable work. Like managing 10 people well is a full-time job. That's why nobody should have 10 direct reports, not even managers where it is their job to manage. But at least if you know that you want to be that size, that gives you a notion of like, This is around about how many people that I will have and what types of people I need. And then we right-size the client base around that. We're not always looking for the next person. We're always looking for better, which means on the flip side of finding better, we have to then cut the stuff that is no longer good enough. And it is in this optimization of pricing and positioning and all of this stuff that all the work becomes more profitable and you escape the capacity treadmill. Many people they don't even get on the capacity treadmill because they're like, "Ugh, I'm overworked, but I don't even have the money to hire." Like, we don't even have a hamster wheel to get on because we can't even afford hiring. All these are pricing issues. And the great news in accounting firms is like Geraldine Carter has shared this data with me. She has helped people. She's like an accounting firm coach, written a couple of really good books for accountants. She has now helped loads of firms go through this really hard price change transition. And she said pretty much across the board, most accounting firms don't see resistance to increasing prices until they get to like a 50% increase or more. A 50% increase or more. And so if you have an accounting firm that is, say, a 30 or 40% margin firm, if you increase your prices by 30 or 40%, you just doubled your profitability with the same amount of work. Now, in small teams, that profitability, that margin yardstick, it's tricky because you're not factoring in the owner time usually. But this is where optimizing proposals and like putting way more thought into how we price work is, is the solution. But if you know what type of business you want to run, if it's like just around you, then we are right-sizing around, you know, 8 to 10 employees. If you want to build something bigger than that, if you want to build a whole management team and escape kind of the traditional partner model, that is about like building a layer of middle managers. Once you get past $1 million, that is eventually like building an executive team. That's a trickier one to build because, I mean, you can operate literally at any size at that point. But if you find that the company is not making adequate money and so you always feel as though you have to hire the next person and you have to increase capacity to get to where you're trying to get, then that is still just a pricing issue. As much as we try to big brain all thought leadership and all these different things that go into like, dah, dah, dah, dah, dah, an accounting firm and automating this and workflow and that and picking the right tech and all this, at the end of the day, all that matters is your price. And the biggest thing I stress with people is abandoning the old adage of double your price, go home at lunch. It is sweeter than that. If you run a 30% margin firm, a 30% price increase doubles your profit. And that's great news. It means you're closer than you realize to a vastly more profitable firm. 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 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. 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. All right, next question, which this actually brings me to our big announcement that we teased up top. Can you discuss any additional disclosures or compliance considerations when hiring offshore talent? Also, is it possible to retain a specific person to reduce turnover-related issues? I assume this is in regards to like offshore turnover. Is it possible to retain a specific person? Build your offshore team the same way you would build your onshore team. The really fun thing with offshore is when you have enough folks to where your next hire is trained by your other offshore folks. I hit this point and it felt amazing when you could hire people and the training was handled by the other members of the offshore team. But around disclosures that are required with compliance, I can only speak to this in the US. If you do tax work in the US, Section 7216 requires two different types of disclosures depending on if they're doing 1040 work or entity work. If it's 1040 work, then you have to get the client to sign a standalone document that is pretty well defined. The font has to be certain size and all that. That is them giving you the permission to share their tax return information with an offshore group and that specifically named offshore group. Now this, you can put basically any term, any length on this. So what most people do is it is just one of the docs. They have them sign when the client first comes on board and it says, you know, they have permission for the next 20 years or something like that. But it does have to be a standalone document, not bundled in with a bunch of other things, but a standalone thing that they sign. Now, on the entity side for tax prep, this same disclosure can go inside of the engagement letter. And so obviously for people, they're like, well, this is way easier to bury now. Not that I think like anybody needs to be secretive about the fact they're using offshore staffing in their work. But for sure on individual tax work, the disclosure is much more in your face. We've had offshore staffing advertisers on our platforms for a while. I've shared before I had tremendous luck with offshore staffing. It was a huge unlock for my accounting firm, not as a replacement for hiring onshore people, but because I couldn't find the onshore people that I needed. And those offshore teams actually helped me upskill my onshore folks faster to get them out of the tedium of all of the work and more into the stuff they actually wanted to be doing. There's a lot of like versions of offshore that suck. Like when Google lays off an entire accounting department, they're like, "Oh, we're just gonna offshore all this stuff now." There are too risky versions of this that like Big Four does where they're like, "We're just actually gonna chop the whole junior level of everything that we do and replace it all with offshore." That's not what I did. I did offshore along onshore to get the best of both worlds. But you may have seen the announcement. I'm now co-owner in Cloud Accountant Staffing. Cloud Accountant Staffing historically has done offshore staffing for accounting firms. Still doing that, but we are now doing onshore placements for accounting firms. So I mean, the big— I mean, my whole business is really built around like, how do I bring people together so they can meet each other and grow from that? I mean, my whole online community, has always been that. Realize it hasn't been access to Jason Staats and hasn't been learning videos or anything like that. It's literally been peer networking. I found in my accounting firm journey all the answers that I needed were best sourced from other accounting firm owners, not from software companies, not from social media. The highest quality answers I could get were from real human beings that are running accounting firms better than coaches, anything else. And this ought to go out without saying, it's kind of obvious, right? And that was the whole reason we built this online platform. Is you are like one good question away from your life changing. But also like making friends was a really powerful thing for me. And then, and we facilitate a lot of that. My events are all small, like local workshops. And the whole reason that we do local workshops instead of yet another national conference is when I go to a national conference and I sit down and I'm like, oh cool, another 1-hour or 50-minute slide deck I got to sit through. But I look around in the room and I'm like, I want to play with all these people. I don't want one more stage talk.. And so I started hosting local events. I've spent over $1 million USD on hosting local events already this year. We're not even halfway through the year. And these are rooms of, you know, 60 to 120 people all coming out within a city to see me speak. But the best part of these events is they're meeting each other, making all these local connections. Just from this year's events, I have— there have been mergers that have happened from firms that attend these. There have been umpteen new friendships made of folks connecting in local towns. Actually got the story just this last week of hires that were made from last year's run of events when people met. So just amazing things happen when people connect. But maybe the most powerful version of this is when an accountant finds my content and they see this sort of like more modern approach for running their firm and they're like, this is all really cool and really exciting. But this is not the reality that I'm living right now. I'm in a firm that has not bought into any of this. And I mean, that was the start of my career journey, was in a firm that was just awful, was soul-sucking. And what I would have done for someone that could have helped place me into a cooler firm, a more progressive firm. That's the entire reason that I've done this, that I'm now co-owner in Cloud Accountant Staffing, is pulling accountants out of these awful situations. And this was me. I mean, 15, 20 years ago, this was me and being able to place them into a firm that can give them an entirely new life and set them on a path of doing something really, really cool. We took a stab at doing this in-house 2 years ago. It was called Jason Careers, and it ended up being— I mean, that business ended up being a tremendous amount of work to get everybody on the same page. And the timing of placing someone into a firm and all that was just way more work than we'd planned for. But what we did learn when we pulled like 350 accountants through that process was that the people that tune into this content are so high level. Like the folks that went through that are like the type of people that will run part of a practice. Like they've got their own reasons for why they don't want to go out and do this on their own. They want to do it as part of a bigger team. But these are— I mean, even manager level probably isn't framing it as high level enough. Like these are people with ambitions to like build a practice within a practice, to take over part of a practice, to run a team. Like, these are incredibly capable people. Like, half the folks that went through this pipeline, me and the guy that were working on this were like, I would have killed to be able to hire somebody like this. Like, this is the type of person I would spend years trying to find in my firm. And we literally had hundreds of these people come through. And so it's a unique advantage that I have to be able to source those folks and draw them out as they're looking for something better. What I needed are the resources to be able to then connect those people with you, that like the folks running modern firms. And so I'm super, super excited about this to have kind of the operational side to now be able to drive that. If you haven't ever worked with Cloud Accountant Staffing, they're easy enough to find online. All the onshore stuff is in beta right now. We're still figuring out like what's the version of this that would be most exciting for you. We're building up a big old list of folks right now that are looking for something better, which is really, really exciting. That is maybe like the most transformational, life-changing thing that I think what I do in my content can enable is making some of those connections. So I'm very, very excited about that. Let's bang out these last few questions. Thoughts on hiring an in-house AI expert to improve efficiency? Here's the biggest trap with doing this right now is I think a lot of people are waiting to themselves learn AI and get their team to learn AI because they're looking for some sort of unicorn person that can bring this to them. The reality is, is it is today and will remain everybody's responsibility to learn AI. Like, I can't delegate the act of learning AI. Everyone on the team has to be plugged into this stuff. The best framing I've seen for pulling in an AI expert, I think I shared this a couple of weeks ago, is actually something I picked up from someone on the tour. They hired an AI expert that will just come and sit next to you and help you through the stuff that you're working on. They use this person as a forcing function, almost like a personal trainer. Because they've paid for this person's time. They have a scheduled time when they're going to sit down with them. And because of that, it happens. Without that accountability, it doesn't happen. To me, this is more a question of accountability than it is finding the right person, because the way you use Claude is literally come to Claude with a job and say, how can you help me do this job? That is not something that we need somebody to show us how to do. For sure, there's people that will get more out of it. And so there's even better ways you could do that. Maybe that's justification for pulling in a coach. But the biggest thing that most of us need is accountability to make sure we're setting aside that time. How do you feel about creating new accounts in QBO with fewer permissions to use Claude in Chrome to log into QBO? So your Claude is logging into a more restricted account that doesn't have maybe all the power user features in QBO. I love it. You are thinking down the right path. Realistically, right now, the biggest issue with giving Claude and Chrome access to QBO is that there's no way to restrict it to just QBO. I mean, you can say in the prompting, never leave QBO, but that is the degree to which it can be controlled. And honestly, it's exceptionally unlikely that it disregards those rules. But I will feel much better about Claude and Chrome when I can limit it to specific websites. This is why kind of in my security best practices right now, I recommend in your organizational settings having Claude in Chrome turned off because it is the single greatest risk of Claude going out and disclosing something is it banging around in Chrome. But you're totally thinking on the right path when it comes to giving your agent access. Only give it access to absolutely what it needs. In our business, we use Notion for a lot of stuff. My agent has read-only access to the stuff in Notion that it needs read-only access to. It's very rare that it actually needs write access to my Notion.. So I'm not giving it any more access than it absolutely needs. This is the right way to think about AI agents. When we talk about AI agents someday using the entirety of the computer, this is why we talk about setting it up in a sandboxed environment where it doesn't have any apps that it doesn't need access to, where you've got this like, like firewall for kids where you like, you only give it internet access to the stuff you want it to have access to. You're totally thinking down the right path where the more we can restrict what it can do, the better. That realistically is the responsible way to navigate the next 36 months is really dialing it back to only what it absolutely needs to mitigate the risk of it doing something unexpected. What do you recommend we tell our staff who are carrying a lot of fear around job security in the age of AI? It's a great question. I had this conversation with somebody in person last week and they said, you know, I've told my team like their jobs are safe and XYZ and they don't need to worry about this and that. And I asked him, I was like, what percentage do you actually believe that? How, like, how sure of that are you actually? It might be a disservice to promise our teams their jobs. Like, honestly, I don't know that any of us are really that sure about this right now. And I totally understand. I want to be able to say that. We want to be able to tell our teams that. I think it misses the point. Because if anything, what that leads to, in my opinion, is inaction. We all want everybody, somebody to tell us everything's going to be okay and that we're going to be safe and all of that. The reality is somebody telling you that does not make it so that your business is guaranteed 5 years from now. You know, that things won't change in a way that you never expected. I think the better thing to tell people right now is we're living in a weird time. But here's the thing. Is we're going to learn this stuff together and you should be so lucky to find yourself at an accounting firm where I will give you the time and the resources to learn this stuff. If there's a tool that can automate your team member's job right now, your job is to put that tool into their hands, not to convince them of some reality where it's not going to replace them. We need to meet reality head on. Stop fishing for like statements that will make me feel good. Just learn it. Like be the ones to put it to work because the ones that are putting it to work are the ones redefining what the next version of this profession is. Not the ones who are like afraid to roll something out because it might be able to do my job right now. If that's the case, we gotta meet that reality head on. And if you are the employer of people, in my opinion, you have a huge moral obligation to bring this to your team, to not protect them from reality, but to show them reality so that we can all figure out together what this next phase looks like. The, I mean, the only alternative to me is like almost wrong, like incredibly risky. Last, can you go deeper on how I could build a Chrome extension with Claude? Did we talk about this use case on the Profession? Did we talk about this use case on the podcast? So Chrome extensions, the little things that live up in the corner of your Google Chrome, I think about like Write Tool. Write Tool's a Chrome extension for QuickBooks Online. Chrome extensions can do all sorts of cool stuff from like modifying a user interface on a web page to giving you like dropdown options in a menu to automate things on a web page you come back to often. But from a programming standpoint, Claude extensions are actually not that hard to build, which means ol' buddy Claude, Claude can build you your own Chrome extension. All right. Ready to have your noodle fully baked? Think about the websites you use the most. What are the processes that you go through most commonly? Oftentimes going to the same pages, clicking through the same sequence. A Chrome extension can do all of these things for you. And they're not hard to build. Literally go to Claude, say, build me a Chrome extension. Here's the first thing I want you to do. And it can be basic. It can be change the text from this or that, change the color from this or that on a specific page. Start with something really, really simple to get it set up. But then you can point Claude to the folder where that Chrome extension lives to give it permission to modify those files. And then have an ongoing conversation with Claude over time to be like, hey, anytime I'm on this page, give me this button to automate something or change this or that about the site. I mean, literally, and it will go and it will change that Chrome extension live as it is working inside of your Chrome. This is a Chrome extension that only you use. It lives locally. In many ways, this is a safer, more responsible way to automate a web browser than to cut an AI agent loose in a web browser. But gang, like this is the era that we are living in. You can literally have this side conversation with Klaudik and build this stuff for you on the fly, specifically to how you work. How cool is that? Whew! Give it a try. That's something we should totally do a video on, building Chrome extensions inside your accounting firm. It is like AI has given us 18 arms, but we still only know how to use the 2 that we were born with. Man, this, So many great analogies, metaphors, not sure. So many really good metaphors in this episode. I hope that you found it as profound as I did. Gang, it was great to see you today. You got any questions, thoughts, any of that stuff, pop it down in the comments and I can't wait to see you back in the next one. All right. I'll catch you there.