552 This Tech Unlocks the Future of Accounting Firms

I don't think we're talking enough yet about how AI is set to change how professional services firms deliver expertise. More people now are starting to use AI meeting note takers and they're quickly becoming like the brain of your firm because they're capturing all the different ways that your entire team dispenses expertise. And I don't think we're thinking enough yet about how this tech enables knowledge automation. Kind of a new and novel form of automation that AI has now unlocked. We've seen some cool developments recently in the note taking apps for accounting firms. More people are entering into this space. And while note taking is like step one having a transcript of your meetings, step two through 100, which I think we're starting to see the beginnings of, like what is the next step for these tools? It looks really, really interesting. So what we're talking about today, knowledge automation for accounting. Fir. If you're new here, welcome to the Jason on Firms podcast where we talk all things running small accounting firms. We recently went deep on this in the newsletter. My newsletter is kind of the best way every week to just keep up on what I'm doing. I'll put a link to it down in the show notes. But we now have three sort of leading AI meeting note takers for accounting firms. We've got Vinyl, Ping and Abicor. And they're all coming at this sort of meeting assistant thing from a different angle. I'll put links to all three of those down below, but to quickly level set on what this is. These are basically AI meeting assistants that will sit in on your virtual meetings. Some of them have mobile apps for recording in person meetings. And some of these tools will even integrate with like your VoiceOver IP system so you can record phone calls. Now compliance there is a little more complicated. There's more regulation around recording phone calls. But the whole value of this is, is every day we are running meetings with our clients. Like in many ways advisory meetings are the sawdust of a professional services firm. And this is really, really valuable stuff to capture because it can be used in so many different ways from how you train your juniors to eventually how clients can even like self serve and get access to some version of your expertise 247 around the clock. And it's not to say that it is as good as you or ultimately a replacement for you, but this AI assistant that can see into every conversation you've ever had with a client and how you advise this, you know, say 80% version of you, while it is not the hundred percent version of you. It can also do things that you can't. It doesn't sleep. It can serve as the first line of defense for the rest of your team when they're stuck on something, they don't want to bug you. Again, there's a huge number of ways that we can use the information, kind of the wisdom that is dispensed in these meetings. But step one is to capture it. Every day that we aren't capturing these meetings is a missed opportunity. Because right now, if you could go back a decade and have a transcript of every meeting that you ever ran with a client, not only would that enable perfect recall, it would give you all sorts of really cool opportunities to like, you know, ask AI, like, help me build a framework for how I usually run sales meetings, onboarding meetings, monthly advisory calls with clients, tax return delivery meetings. The value is in more than just the transcript because AI and large language models can do really, really interesting things with those transcripts. Before generative AI, when you had a transcript, which, by the way, they weren't even particularly accurate back then, they've gotten much better. All you could really do with those transcripts is like this control F, like find a specific word and, and even that oftentimes didn't work because the transcription wouldn't nail what the word was. And so if I then go and search for that word, it may not see it. But large language models are way better at this. For me to put that transcript into, say, ChatGPT, something like that, and then be able to say like, hey, was anything around here discussed? Or put 10 transcripts in there and be like, hey, walk me through. Or develop a general outline for how I usually run this type of meeting. Recently I put out my 30 favorite AI prompts to use with your meeting transcript. So you put your transcript into ChatGPT and then you give it a prompt to get something useful out of it. I'll put a link to that down in the show notes as well. I can tell you probably my favorite AI prompt to use with meeting transcripts. It is to extract from the transcript anything the client said that could be used as a testimonial. You know, when clients do like, say that nice thing and we're like, oh man, that would be nice to have to be able to put on the website or something like that in most of our firms that some degree of that stuff is happening, like every day, but we're not doing anything to capture it. This is a great use case for AI to be able to pull anything the client said in that meeting, that was, you know, sort of a glowing review, and then kick off a workflow to ask them permission to be able to use that as a testimonial. Now here's the rub with AI meeting transcript tools right now, as was kind of illustrated in what I just explained, in order to do that right now, it requires somebody going out, taking that transcript, put, putting it into an AI tool, giving it the prompt, and then kicking off that workflow. But wouldn't it be nice if that stuff just happened? And that is the beginning of a, a much more powerful platform for managing these transcripts and even beyond that kind of the collective knowledge of our accounting firm. Because if you think about what makes your firm unique, it's the people, it's the voices, it's the ways that you advise. At the end of the day, all our accounting firms, like we are by and large using the same tech, kind of built on the same stack of tools. But what makes one person's firm different to another's is how the humans use those tools and the differences in how the humans deliver that service. By the way, in the future, when we have super powerful AI tools that will get way more of the work done, what will be the differentiator? It will still be machine plus human versus machine plus human. All else being equal, everybody has access to the same tools. The differences will still be in how the humans deploy the tools and do the human interacting stuff. But as your mind starts to think about like, what are all the different ways that this context and like these transcripts could be used, you're thinking about internal training. You're thinking about, well, this like staff training meeting that we had, like we covered six different topics. There's like an HR thing, there was a tax training thing, there was this how to use this specific piece of software thing. Like, wouldn't it be nice if my tool would sort of strip those out and tag them in specific ways? So I could just have like this HR assistant that pulled in all the different HR things. Even if one transcript had eight other things that were unrelated, that level of organization is going to be really powerful. One of the things that assistants specifically for accounting firms do well right now that you don't get from a mainstream median transcript tool, like a Fireflies or a Fathom is these tools for accountants will map these transcripts back to the clients and contacts that they relate to. And normally this is pulling from your practice management system. So ideally your meeting transcripts tool, it integrates with your practice management system. So when it Sees your meeting with person xyz, it can associate that back to a client and a set of contacts that belong to that client. Ideally then your transcript it gets pushed back to your practice management system. We're already seeing practice management systems who will happily integrate with these meeting assistants and then other practice management systems who are more all in one position, like they want to do everything themselves. Choosing to not integrate with these meeting tools, which is frustrating. That is one of the trade offs of going to a platform that's all in one is they're usually going to be a little bit slower to ship that next thing that you want. But when they do, then at least it's all inside of the same thing. Right? And hopefully it's good enough to be useful. But, but if you're weighing like why would I use an accounting specific meaning transcripts tool versus a generic meaning transcript tool? One of the big upsides is that integration with your practice management system. And so what we are buying right now with these tools is the ability to have these really good transcripts. But what they represent in the future is kind of like the brain of your firm. Imagine you have somebody who's this technical subject matter expert just, and they're really, really good in this one domain and they come and work with you for three years and they advise your clients on how to do this one very specific thing that they're an expert in. And maybe when you first hire this person, let's say it's R and D credits, I don't know, some arcane thing that's, that's hard to do well, that requires very specialized expertise. Before this person came to your firm, you didn't have the firepower internally to be able to deliver this service. This person works for you for a few years. Some of the other folks kind of like learn the ropes and get up to speed with this thing and then this person leaves, they go on to do something else. Wouldn't you like to have those meeting transcripts? If you're thinking like this is the thing we're going to continue to do with the team that we have, wouldn't it be nice to be able to query those meeting transcripts to be like in situation this or that, like A, B or C. How have we advised clients in the past? That's super interesting. It's also a little bit dodgy. Like you look at the creative fields and sort of the original sin of AI, the notion that it's trained on materials that it doesn't have license to train on. I think this Whole, like, knowledge automation thing. There's going to be a very real conversation here where folks that work for a firm, are you going to feel good about the firm owning those conversations in perpetuity? The notion that you dispensed that advice and you don't have rights to that conversation that you had because you did it while you're under somebody else's employee. And you know, to take it to an extreme example, this podcast, we've got an AI assistant that can see into all the episodes of this podcast, and it dispenses advice very similarly to the way that I do. Imagine you had something similar inside of an accounting firm and you're like, I don't have 550 episodes of a podcast, but you work for an accounting firm for a few years. I mean, you'll probably have 500 meeting transcripts and then you leave. The spooky thing is that firm can create an AI assistant that will act and talk and advise very similarly to that person who left. Do they have rights to, like, that person's, I don't want to say likeness, but the assistant that is backed by all their expertise. Kind of weird, right? Now there's, there's an element of, like, personification here that I think becomes fundamentally problematic as we're seeing more tools that will, like, look visually like the person and talk in a voice that, like, is nearly indistinguishable from that person. I think we can all agree that sort of stuff crosses the line. But if you are the big boss, if you're the firm owner, this really gets your wheels turning on just how much like, IP our firm is generating that we maybe aren't even capturing yet. Because at the end of the day, what is, like, what are the assets that an accounting firm owns? Practically nothing. Like, you've got some systems like quality control, like that sort of thing. Maybe there's some value in those. But I would say all those things pale in comparison to the value of those transcripts. Think about M and A. Think about selling your accounting firm or you're buying another firm. How valuable would it be to have the transcripts of the last three to five years meetings with these clients? That would be tremendously valuable, right? And so what is the future of these tools start to look like? I personally think this stuff is fascinating to go back to the whole knowledge automation sort of concept. Like, you look at, you know, an accounting ledger, and an accounting ledger is kind of this perfect record of every transaction that's gone through a company. Right? Like, we understand how all that looks, it's all these transactions. And the accounting ledger is sort of the collection of all those. But what if you had something that was more like a knowledge ledger, a perfect record of every time that you dispensed expertise. What would those quote unquote transactions be? They would be your meaning transcripts. These AI notetakers right now, they're almost becoming the accounting ledger for the mind. Like they are a collection of all the expertise that you've dispensed all in one place. Now what do you do with that next? I mean you can think of all the different places you could use this. Somebody writes you an email and you want AI to draft a response for you. Boy, I mean that, that's really valuable context to have, right? How I've advised people for the last 10 years, how I advise the client that asked me the exact same question six months ago. You could start to see how this kind of becomes like a centralized firm wide brain that hopefully brings your team a little more in alignment. And how we advise folks if we have an effective way that we can interface with this brain, this machine. And so ultimately what becomes the home of all of that context, all those meeting transcripts, all that data? Do we want the meeting transcript tool to be yet another place that we have to go and we have to connect it with everything else? Because right now my email can't see into everything that's in that meeting transcript tool? Is this a really strong argument for practice management systems to pull this stuff in house? Because that's where the most context for the rest of the firm already lives. It is usually all of our client communications. It's oftentimes where all of our work gets managed like outstanding requests. So if you had to pick a single place for all this context to live, I think you probably want it in the practice management system. And it starts to feel a little bit like an integration nightmare, which we already have this problem in so many places in accounting, all the these siloed sources of information, every tool now has its own little AI assistant and it's like, dude, I don't want one more chat assistant to talk to that can only see into part of my data. So at its worst, like it makes this whole problem worse if we can't find a way to bring all this context together. But I can tell you if I'm somebody that is running a one of these note taking companies, like I'm pretty bullish on my platform, arguably becoming like the smartest platform there is. You look at an accounting ledger it's like, it's a record of all the transactions, but it does not in any way reflect how I advise you. Look at my email history. That's another super valuable source of context. Right. So there's some real intelligence there. But you look at the practice management system, it's like, well, it's a record of like all the work we've done for folks. Oftentimes it's the file system. But is it truly reflective of like the expertise of our firm and how we dispense expertise? Not really. So it's gonna be really interesting to see how these platforms evolve. And we're already seeing some platforms communicating what they do as sort of the brain of the firm, practice intelligence, whereas others are still communicating what they do as like note takers. And the note taker path, like, you can see how that can quickly become commoditized. Right? Like your job is simply to get me an accurate transcript and then I will go off and choose what I want to do with it next. But the platforms that are leaning more into like practice intelligence, like, you can kind of understand how down the road that's a really, really powerful thing for our firm. And even the ability, ability to like, build out customized workflows of every time we have a new meeting transcript. Do xyz, like the testimonial example, like, how do I build a custom workflow to run this prompt on, on a meeting transcript for every single meeting and if this thing came up, somehow flag it and kick off this other workflow. This is a thing right now, frankly that you could do with like zapier or a make and kind of build your own automation. But that's out of reach for most people. In that interview with Dan Luthi we did on, on the main channel a couple months ago, we were talking about how they have sort of conversations with their clients and there are talking points that fall into the bucket, that is bookkeeping, talking points that fall into the bucket, that is controllership, and talking points that fall into what is clearly CFO work. And depending on how the client is scoped, you want to keep it within the lane of what's appropriate. Right. And in Dan's team, they even have different people in each of those seats. And everybody kind of knows, like, oh, don't like, wade into what your teammates are doing, especially if the client's not scoped to do that type of work. If your meeting assistant can see into the terms of the engagement and has an understanding of when is a conversation wading into CFO territory or Controllership territory where they are only scoped to do something less. Then there's a very real element of like revenue recovery here. Like how do we, how do we ping the manager to be like, hey, just so you know, like, there's some stuff coming up here that's probably outside of scope and it could be worth having a conversation with the client. You could see how like that actually becomes super valuable really fast. Now last concept here that I'm super excited about and where I think these tools inevitably go. And Fireflies, a mainstream meeting assistant, actually shipped a very basic version of this recently. It's not like as mature as the version that I'm excited about, but it's a signal that they're working in that direction. If you think about like call centers, folks that are doing high volumes of calling, this stuff is done very much on rails where they're kind of being like they have a set of prompts that they need to work through and kind of have to guide the conversation through this branching sort of sequence of logic. And the idea is you're trying to build a system to get a team of hundreds of folks making phone calls to make these phone calls in the same sort of standardized way, right? And eventually you learn like, what are the best ways to sort of guide these conversations? What works the most. And then when you hire one more person, you're plugging them into a system that ought to already work. Hey, real quick. We recently did an updated demo day of one of this episode sponsors, Client Hub over on the main YouTube channel. Ran through the really cool stuff that they're doing for bookkeeping and accounting services firms building sort of the centralized brain where you manage and oversee all of the work. But also Client Hub is getting more into like the doing of the work, the actual workflow itself, doing some really cool stuff around like AI month end close review and doing so with a deep integration into QBO and Xero. So like pulling all the transactions into Client Hub so you're able to do more of the review from inside of the practice management system. And it happens right now at a time where it's kind of annoying. We've got these really cool workflow tools redefining how efficient we can be with getting the work done. But then you've got your practice management system which is like disconnected from all of this workflow and seemingly just overseeing like the status management of the work Client Hub. They're kind of trying to do both of those things in one platform. It's been really interesting to follow if you Are a bookkeeping firm or client accounting services firm. Client Hubs, one of those platforms that I think you just kind of have to check out. Like I said, we just did a new demo of it on the YouTube channel. I think you like it. To learn more about Client Hub, check out the link down in the show notes. Hey, real quick attention. If you have clients that do project work, do ugly stuff like manufacturing, like inventory, construction stuff, anything where you got hairy itemized bills that have to go to specific destinations, when it comes to bill pay, there is a new sheriff in town. It is this episode's sponsor, Makers Up. Even if you don't do bill pay, this is something that is worth showing your clients. Actually, we've done a number of videos on my channel. You can start sharing my videos with your clients. What fun. Gang, we are light years beyond what the old guard like Bill ever did with automated bill extraction and get like getting these things through the approvals process, which is oftentimes much more complex than just is this bill over $10,000 or under $10,000? Man, I worked with a church that had this like multi level like just this crazy hairy combination of approval approvals that was dependent on like a mountain department, all these different things and they had this physical thing that had to get passed around for every single total nightmare. You probably got a bunch of clients with situations like this and you can absolutely be the hero with Makers Hub. It is seriously good. Worth checking out. Learn more about Makers Hub with the link down in the show notes. Now, inside of an accounting firm, most of us are still running our meetings in a not as. Not in a specially structured way. But as soon as you need to go delegate the running of that meeting to somebody else, all of a sudden it gets way more important to make sure like are they running this meeting the way that I run these meetings? You look at say a discovery call. What are the questions we need to make sure that we ask and the details we need to make sure we get on every single discovery call, that ought to be something that's a pretty explicit list. And then the next time a new client slips through and they've got some really problematic thing or you're like dang, I wish we would have known this when we were doing discovery. Great. Like that learning then cycles back into the items that you need to capture at discovery to make sure that you don't make the same mistake mistake again. But in the course of having that discovery meeting, wouldn't it be nice if you had a meeting assistant that was literally Updating you with what still needs to be covered during the meeting. So this is a user interface that is transcribing the meeting live. And as each of these items are discussed, it can resolve those checklist items, showing you the remaining things that need to be covered. Now that, that's pretty cool, a little bit sci fi, right? But that is honestly just scratching the surface of how valuable these meeting assistants can be because they can also give you advice around kind of the overall tone of the meeting. And you know, a client brings up some, I mean, this is a dumb, dumb example, but a client references like, you know, an old TV show or a thing that they can't remember. I mean, the AI can near instantly surface that to you to be like, oh, they're talking about xyz. Or you bring up some past conversation that you had as you're talking with the client, the AI can go and fetch that and provide you with additional context to guide the meeting. It's pretty weird. Like it's. And in some ways it's kind of overwhelming. But you can see a version of this that makes these meetings vastly more productive. Kind of having this live note taker who's with you going out and fetching details and like enriching the meeting along the way. But also if you have a number of folks in your firm who are running these meetings, it becomes an invaluable tool to get folks running those meetings in a more standardized way. And so that is not just giving you a transcript after the fact. That is that assistant being in the call live. And you have a user interface where you can see it's kind of like guiding you through the prompts that you got to get through, making sure that you don't miss anything, maybe even giving you like helpful things along the way. That seems inevitably to me to be where these things are headed. You know, a client asks a technical question and then immediately having a readout of, here are the ways that we've advised eight other people in a similar situation and how similar are they here? They are ranked by similarity to be able to do that sort of stuff real time. There's some of that that you won't want to do real time. Like, you'll just, you need some time to think about it and do some deeper digging. But there's a lot of stuff that are more mechanical, a lot of things that are more mechanical where you're like, I just need to remember to ask them X, Y and Z. And then the nice thing here is as soon as the meeting's over, the AI can see what the answers are to all those things and automatically take those answers out, resolve the requests, and then, like, you know, deep link that stuff back to. Well, the answer to this request came from this part of the meeting. That's really exciting. This is an example of a type of tedium that I think we just aren't aware of or aren't thinking about right now. If I have a conversation with a client and they give me an answer me then having to go communicate and document that answer somewhere else, like, that is tedium. That's something that we would happily get out from under. And there's so many examples of this across the work that we do from like, futzing with PDF files and annotating them and naming them specific ways, all these different things where it's like, if you pointed out any of these things, you'd be like, yeah, no, actually, I'd love to not have to do this stuff anymore. But this tedium takes up a huge percentage of our days still. There's a ton of this tedium around meetings, I would say. So I'm super excited to see how this category develops. I mean, the biggest action item right now, if you haven't already, is you need an AI meeting note taker. Like, you need to be capturing and documenting these meetings every day. We're not capturing them is a missed opportunity. It really is, because you can never go back in time and get those meetings back. And wouldn't you love to have the last 10 years of meetings now? The biggest objection folks will make is, what if I don't want to be recorded? Like, I don't want this stuff to go on the record. What if the client has an issue with it? The client conversation is usually pretty simple. It's usually like one conversation with them around. Why you use meeting note takers and you're like, listen, this is going to make your life easier. Like, it's easier for the team to stay in sync. Like, we don't forget things at that point. Like, it just makes everybody's life easier. And if the pushback is, well, what if I don't want to be recorded? Like, then that's probably not a conversation I want to be having with you to begin with. Like, if I can only ever call you on my burner phone, then is that really a good client for me? Now, the flip side of that, the accounting firm, if you've got, you know, a partner in the firm, that's like, well, I don't know how I feel about being on the record, in every single one of these conversations, if you pose that same thought experiment to a staff person of that partner and you say, okay, that staff person over there, are they allowed to say the same thing? Are they allowed to like, pick and choose when they want to be on the record or not? Most of us would be like, no, I don't like the idea of like my team going out and giving out that advice that maybe we couldn't stand behind. Well then like we, we gotta draw the line somewhere, right? Like, that's the trouble with fraud is it doesn't scale. It just doesn't. You want to do that stuff and shoot from the hip. Well, as soon as you need somebody else to come in and do your job for you, delegate a client to somebody else, they're gonna be like, well, I miss Gerald. Gerald used to tell me the things that I could get away with in xyz. Like, you just, you kind of just. You don't want to build a firm that way. Like, the juice is not worth the squeeze there. There's only so much of the putting my own neck on the line that becomes a good business decision when I'm doing this for a thousand different people. The much, much easier way to run the business is in the one that you can stand behind. And even within that, there are very good firms and very bad firms that are like just not plugged into what is the best advice that we can actually stand by. So, so the whole argument of like, I don't want to be recorded, for me, that's pretty weak. Like if we're relying upon giving away advice that we wouldn't want anyone to know, that feels a little risky to me. So if you haven't yet, super recommend picking up a meeting assistant. I've linked the three down below that are for accounting firms. The two non accounting firm mainstream ones are called Fireflies and Fathom. I will generally steer folks to the accounting category option because if we don't support the people who are building software just for accounting firms and your accounting firm ends up just running on tech that anybody and everybody can use, like that's a, that is a dark timeline. Like, I'm all for supporting the folks that are building the tools that will make us more efficient. Like, I want the version that will integrate with my practice management system. I want the version that will integrate with my proposal tool so it can see into the terms of the engagement and raise its little robot hand anytime. The conversation, like, veers into a place that's out of scope. I want all that stuff. So that's why in a situation like this, I'm usually going to go for the option that's built for accounting firms. You got any cool novel stuff that you've done with these transcripts yet? Drop that stuff down in the comments. I'd love to see it. It's interesting even having an AI that's very good at answering questions like you like, which I have with this podcast. The one thing I have found having this assistant trained on all the podcast episodes is humans still won't use it. Like, people don't want to just go to the robot for most answers. They will start. I mean, I'm in Washington D.C. today hosting an event and 80 people are going to come out and ask me questions. Why would you come out and ask me questions here when you can just go ask the AI assistant questions? Right? For as powerful as all this AI stuff is, I think we're still learning what aspects of this interaction need to remain human. For some reason, even if there's a robot that will give very accurate answers, there's some aspect of human connection here that we still need to like, fully stick the landing on an answer. And I come back to this when people are like, well, what? When am I going to give people answers? If everybody just comes to my robot for an answer? Like, I. Right now, I really struggle to see this scenario where the robot answers 100% of things because some aspect of the human experience and psyche, like, still wants to connect with a human to get that answer. I don't know, maybe that's a short term thing and we just have to unlearn that or something. But I would say right now, even though these assistants are getting very, very good at answering questions the same way we would our clients, they, they still need the person, they still need to sit down and have the conversation for whatever reason. And I would say, like, that's a good thing. At the end of the day, we have to be able to like confidently communicate what is the thing that we can't do, not can do, what is the thing that we can do that AI cannot do. Then the easy answer there is like, sit down and have a conversation. Like, connect on a human level in the way that we can't with AI. And I think there's actually a lot of evidence already that even in situations where AI can be a tremendous advisor and extremely accurate, it still doesn't fully, like, close the loop of what that person needs. There's still an element of like, coming to the human expert and having that conversation that seems to get us to a more satisfactory answer, right or wrong. Anyways, been a really exciting last three months in meeting assistants for accounting firms. A couple of these tools are brand new, but I already have integrations with practice management systems, which is something we've kind of been clamoring for for the last couple of years. Just using mainstream off the shelf meeting note takers. Super recommend you check those out. You got any wisdom to share here? Drop that stuff down in the comments. And thanks for being here. I'll catch you in the next.