The Wize Way

Episode 180: The Most Dangerous Myth Accountants Believe About AI

• Wize Mentoring for Accountants and Bookkeepers

Share episode

Most firm owners hear the noise about AI and panic. They’re told their jobs are disappearing and the industry is on borrowed time. But the reality is simpler: AI isn’t replacing accountants,  it’s reshaping how firms think, make decisions, and deliver value.

In this episode, Brenton Ward and Thomas Phabmixay break down what AI actually does in a firm, why it’s more “translation engine” than threat, and how it can scale judgement, not just automate tasks. Thomas shares how he uses AI inside his own practice to analyse work, improve communication, and speed up decision-making and why the biggest wins come when AI is woven into workflows so the team barely notices it’s there.

You’ll learn:
✅ Why “AI will replace accountants” is a myth that’s been repeated for decades
✅ How AI turns financial and client data into instant insight
✅ Where AI creates real leverage in a firm’s workflow
✅ Practical ways firms can use AI today without building custom tools
✅ Why critical thinking matters more than ever in an AI-enabled world

If you’ve been overwhelmed by the hype or unsure where to start, this conversation gives you a grounded view of how AI can strengthen your firm, not threaten it.

________________

PS: Whenever you’re ready… here are the fastest 4 ways we can help you fix and grow your accounting firm:

1. Download our famous Wize Freedom Strategy Map for FREE - Find out the 96 projects every firm owner must implement to build a $5M+ firm that can run without them - Download here

2. Need to Hire right now? Book a 1:1 FREE discovery call with our WizeTalent hiring coaches to help find your next team member the Wize Way – Click Here

3. Book a 1:1 Wize Discovery Session – Spend 30mins with our Wize CEO, Jamie Johns, a $7M firm owner who is ready to give you his entire business plan to build a firm that can run without you – Find out more here

4. Work with Jamie and our mentors for 8 weeks - Build a custom business plan for your firm - Apply here

SPEAKER_01:

Welcome to The Wise Way, the show for accounting and bookkeeping firm owners who want more time, profit, and freedom and a business that can run without them. I'm Brenn Ward, your host, and each week we deep dive into the real stories, proven strategies, and battle-tested tools from successful firm owners just like you. Our WISE mentors want to share their journey of how they've scaled and systemized their way to freedom so you can too. If you're stuck in the grind or you're ready to scale smarter, this is your blueprint. Let's get into the episode. Hey guys, welcome to today's chat on the Wise Way, the AI Edition 001 episode. Thomas, welcome. Good to be here, Brenton. Great to have you. Would be keen to have a bit of a 20-minute power session and would like to bring our audience some more of these conversations as uh the noise really heats up. Well, it's been heating up for quite some time, but it's pretty loud out there now around all things AI. There's no avoiding it anymore because even my mother uses the the words AI in her conversations, which scares me. Um this is where we're at. So I would love to, for this conversation, start with some fundamentals, some basics. Um I I really for because we're going to be having these conversations pretty regularly going forward, um, for those who haven't heard uh too much from yourself, either on this channel or on other platforms, give us the give us how much you're um nerding out on AI and connect that to um your role as CEO of a family run uh public practice, public accounting practice in Sydney. Give us a quick soundbite of Thomas, the AI guy.

SPEAKER_00:

I think if I had to sum it up in the phrase, Brenton, um I need to touch grass. Like I need to, I need to open the window and go to the park and do something else. But I've just been deep in it. I, you know, this is the sort of thing we've been hearing since um university. People always saying accounting is going to get automated. Uh, will we even have a job anymore? And with all things that are getting hyped, it's you know, we we people want to hear um what is it really? Like what is it, what is it actually beyond the hype? And the experiences I've had so far have been um getting getting experience, uh getting my hand in code, like trying it, writing some Python scripts, uh using GPT and Claude to explain things to me. That was you know how it was early on. And just kind of exploring what it can and it can't do, and and what does it actually mean for accountants? Um what I've been doing so far is just been working with software engineers and chief uh chief technology officer. And I I just wanted to get deep in the world. I I really wanted to hear from the people who do it uh for a living because you know, being accountants, we don't come from a technical background, and it's really easy for us to just kind of exaggerate what it would do or just get the wrong idea. And yeah, it's just being building our own tools, releasing a few things, working together with firms to see how it can be adopted. There's a lot of potential here, uh, and there's just a lot of misconception as well. I I just I don't know, I don't have too much to share. Like, yeah, I don't have too much.

SPEAKER_01:

It's a great intro because already in that um you differentiate yourself from you know a lot of um a lot of firm owners and business owners right now because you know, let's face it, most of us are still very much stuck in the weeds of running our business, haven't had the chance to even pop our head up and really understand what this could be looking like or doing for our firm. So it's I and this is why I'm so interested to get your perspective on it, because one, I I love the the overlap between you as a firm owner, you as a wise mentor. So, you know, having preached uh the wise way for quite some time now and this new obsession, almost healthy obsession with um this revolution. And I'd want to touch on quickly your point on you know what you heard heard at university. Uh we're losing our jobs, something's always going to come after us and take the job. And I find it interesting as a remark because everyone's hearing it at the moment about jobs disappearing, et cetera, with AI. But if you listen to the great Ed Chan, our wise master yoda, um, without giving away his age, he's been saying for the last 40 years that someone or the noise has been telling him and his team that they're going to be extinct or they're going to be eradicated and that they won't have jobs at every point in technology technological evolution. Um, and sure enough, still around 40 years on, still profitable, team still there, maybe slightly smaller and team like producing more fees with less people, but still huge demand, um, still a massive talent shortage in the industry, et cetera, et cetera, et cetera. So I find that rhetoric interesting. I highly disagree with it. Um, I see both sides of the fence for the conversation on AI, the fearists and the concerns, and also the opportunists and the optimism around it. Um, and I I aim to explore that from lots of different angles with you uh over the coming months and years. I think it's an unfair question of you to ask where you see all of this heading uh over the next couple of years, because it's just moving so fast. I think we can all agree on that. Um, but trying to kind of boil it down to what what is what is AI? What is it to you? Uh and fundamentally as a technology and as a shift in how we do things, where is it best placed for us right now?

SPEAKER_00:

The way I've been able to mentally reason around AI, Brenton, is thinking about it in terms of as a tool for translation. And that's what AI does really well. No, AI is trained on public information. It can't like it can work within frameworks to try and get coverage around things that we might have not invented before, like they're trying to use it to discover new medicine. But fundamentally, and where it's most reliably used is in a translation tool. And where have we seen it work better than when people have started to vibe code? You know, make me a website that looks like this and that, and then all of a sudden code starts showing up. And it's AI translating someone's query into another medium. It works really well when feeding it financial statements, PLs, and balance sheets of multiple years. And you're asking, okay, so uh take this in. I want you to give me an analysis. What you're really asking is you're asking it to translate it into insights. So it looks, it's, you know, we call it reasoning. It's really just translating these financial statements, these P ⁇ Ls, these balance sheets into another form. So where where I find it best placed is when it's in places where you would need to make a judgment. Okay, and it's it's it's I think it's confusing to can to think of judgment together with translation. But for example, you want to check someone's work. You want to go and um they've given you like a like a tax planning document letter, they've given you some completed work. You want it to be able to check it.

SPEAKER_01:

Um You're feeding it information, client information, uh, team information, data, and you're empowering the team to do what with it? Like what's the ultimate objective? How do how do we leverage it in that sense?

SPEAKER_00:

Yeah, you're you're basically leveraging AI as a unit of intelligence that you can purchase. Like it is now intelligence is now a single unit that you can use. Or where we're we're running code, where we're using it for automation. AI is essentially intelligence. Okay, and now you can start to actually apply it. Where it's actually best placed is in the parts of the workflow where you would have to make a judgment, where you would have to reason over something, where you would have to translate something. All those activities require a form of translation. You're you're asking it to uh take a look at something that's already there, context and prompt and instruction, and then think it through in order to get to an outcome. And you know, many would argue that it's not actually thinking, it's more of a probability machine. But it just hallucinates very useful things uh most of the time, and it seems to just be getting better. And where our workflows exist today have always had these points where a human had to come in and then uh make a decision, do a review in order for that workflow to carry on. In those spots inside a workflow, we can now plug it with AI. And so what does it mean for accountants? It means that judgment, reasoning, and intelligence is now a scalable thing. Um, for one example, if we finish our work, we have our working papers and tax returns and financial accounts done, you would go and feed it into AI. Get get it to do a broad check across a checklist. And of course, it's not going to be fully correct. But right now, it's the equivalent of asking an intermediate tax accountant to take a look at the work in detail. So you still have to go and check it. Only except it comes back in less than a minute.

SPEAKER_01:

Yeah.

SPEAKER_00:

Right. And that's that's really the power. You know, reflecting on why's and the 80-20 rule, we don't need things to be 100%. Uh, we don't need uh it's better that something does it 60 or 70% well rather than the person doing it 100% well. It's leveraging our our most senior skill sets. Right? It's a lot of people are thinking that use AI for the entry-level tasks. It's going to reconcile transactions, it's going to sort out your receipts and bills. And that's definitely true. It can do that really well. But AI does not care if the thing that it's doing uh was a task that belonged to a person getting paid$200,000 a year or someone that was getting paid$50,000 a year. As long as information was needed, context was needed, and instruction needed to be performed, looking at an input that is constantly changing and evolving, it can it can really begin to pull it off. Um accounting.

SPEAKER_01:

Yeah, what I'm hearing there between the lines is also it's a value creation tool for client experience. So in a practical sense, you know, the everything you've spoken about there, you know, is exciting and makes a lot of sense, but bringing that to life um amongst everything else that you've got to do to run the day-to-day of a firm, how how are you guys managing that process at the moment? And what what, if any, pushback are you getting from the team? Um, how is it coming to life in your own firm?

SPEAKER_00:

What I found was this, Brenton. If you have a team of 10 people, you're going to have like one or two of them who are just super keen about AI. Like you'll like, you know, I think the common advice right now is give everyone a GPT license or a clawed teams license. If you could actually go in and see how much the team is actually utilizing it, there'd be like one or two people in there who are really actively using it. And I'm seeing that across pretty much every firm. And I've been thinking on that. You know, I've been thinking on so how do you get it adopted then? And it it kind of made me think to why Chat GPT was even created in the first place. They wanted to prove that AI is here. They presented it in its first example, which was in like a chatbot, uh chat box on a website, and you can start playing with it. So it was there to demonstrate that AI can work. I think just everyone assumed that this is AI then. This is AI. You'll find that any serious uh engineer, any serious person who's trying to implement it within their firm, it's not Chat GPT or Claw that's going to be part of our day-to-day. It still needs to be placed and woven into something. It needs to be built into a tool. People don't want to feel like they're using AI. They just want to feel like they're getting their job done. Okay. So AI makes a lot of sense when you're not even noticing it there, right? It's it's inside a workflow management tool and it's automatically assigning it based on who the right person it is to go to. You're we don't think about it, but you when you go to your Gmail and an email is a categorized into like promotion or marketing, that's AI being used, but we don't think about it. We don't really take notice of it. So where AI makes the most sense is just, I guess, when you don't see it there. I like that.

SPEAKER_01:

And in terms of low-hanging fruit, so without having to go down the path necessarily like you have done working with engineers and creating your own tools, um, what are the what is the actual low-hanging fruit that an everyday firm owner and their team should should be really just taking advantage of straight away? Like dipping the toes in the water if they haven't already. Like everyone, as you said, everyone's basically played with chat GPT at this point and maybe one or two other off-the-shelf tools. Um but yeah, what are some of the low-hanging fruit that are actually going to have a decent impact on uh workflow management, production, profitability, you know, the the real drive and forces of a firm?

SPEAKER_00:

I so some very low-hanging fruit that everyone's pretty much done already is just getting getting it to help you write better communication. Like that's a very obvious one, but that's been the that's already been proven value. Just if you look at any large org, just the quality of communication has generally improved. But where I find it the most powerful is when you give it two pieces of information and then you ask for the differences. Okay. So for example, give it a transcript from a meeting, but then hand it also the engagement and ask, what are they asking for that we aren't already doing for them that we can add into it? Um give it financial statements from two different years and ask for the differences. If you, if you basically give it two different things and ask it to analyze, you're already using more of its horsepower than if you just asked it to like improve your email. And I've tried this myself. If you if and anyone can really try this, you can go to OpenAI API playground or Claude's Playground, give it a give it like a very small input, like a not well described input, and then ask it to output. And you would see how many tokens it's spending. Tokens being Yeah, one's a token for everyone.

SPEAKER_01:

Yeah.

SPEAKER_00:

Yeah. Okay, so what a token? Token is basically the unit, I guess you can call it the unit of energy or power that AI uses. That's not the best explanation, but I think that's something we can all um get around. It is one unit of energy that you're asking it um to spend in order to give, to do something for you. If you and so essentially, if you give it not much input, not much context, and you ask it to produce something, it won't use a lot of energy. Okay. If you give it a lot of input and you ask it to do an analysis, you ask it to produce a very dense or concise summary or analysis out of it. A very common one is if you gave it an entire transcript and asked it for the pain points and action items out of it, the token spend, the energy use is just so much, it's so much greater. And the game then is use as much tokens as possible, find ways in which you can just make those servers that are sitting somewhere in America just scream rather than not even break a square. So how I try to like teach my team to understand this is I taught them a framework uh called the Bloom's Taxonomy of Intelligence. And there are six levels to the Bloom's Taxonomy of Intelligence. The first level being remember, okay, that's just literally just to remember it, memorize. Um the second level being explain. So your ability to just explain something, do you remember it, and now you kinda you can explain it. The third level being apply. So meaning you could read a recipe on how to bake a vanilla cake, and then you'll do it. But in these first three levels, it does not really require that much intelligence. Because the first one is just memorizing. It's like you're reading a dictionary, you don't you don't even understand what it means. It doesn't really take that much brain power. The second being explain. Now you have to try to actually explain it to someone so that they understand. So you're using a bit more brain power and then apply, which is just follow these instructions. You can memorize the instructions, you can explain it, and then you can apply it. But these first three levels don't really use that much intelligence. Where you actually start to use intelligence is in levels four, five, and six. Level four is analysis. And analysis is compare and contrast, experiment. Like put two things together and then see what happens. Try and tell me the differences. Try and tell me how it's similar. You're already now using a level of intelligence that is orders of magnitude greater than the first three levels. The fifth one is evaluation. And evaluation is you're you're you've done a bunch of experiments. Now decide what sort of criteria do I need to create? What sort of framework do I apply or note down that allows me to judge whether these two inputs going together actually produced a good result? Like how do I actually evaluate it? Is it good or bad? You can analyze, tell the differences and explain and experiment, but doesn't mean you can tell if it's good or bad. And the sixth level is hypothesize, which is to create, which is to find a gap in this world, in a knowledge or in a domain, and then put something there that has never existed before. And how this relates back to token usage is it's artificial intelligence. You want its brain to work the hardest that it can possibly can. If you ask it to explain, you don't need AI, just go to Google, like pick up a dictionary. If you ask it to analyze, you can't really do that in Google, you'll probably come up with an article of someone doing it already. If you ask it to evaluate, it starts to fail. It's not, it's not a very good use of it. It's already very challenging. And if you ask it to hypothesize, it's it it can do that reliably well, maybe like 5% of the time. So if you can just stay in level fours, five, and six, if you can try and use AI in a way where you're, you, you can, you're prompting it to use intelligence in the most strenuous way possible, that is when you're getting max value out of AI. Okay. So it's asking it to do tax planning, giving it a transcript and going, here's context around the entity structure, here are some tax strategies, evaluate for me what is the right strategy to use, or pick and choose from it. See how, see how you can like piece these things together to give me ideas. And the goal is to then teach the team, don't just run with its answer. You yourself also now have to analyze, evaluate, and hypothesize. And more importantly, always be evaluating. Hypothesize different ways, like find gaps and then try to use AI to explore this really quickly. Oh, we're in accounting. We're we're a knowledge-based industry. There is no shortage of use cases where we would be trying to reason through things, trying to um um trying to come up with strategies for the client. Wherever those tasks exist, you want to encourage your team.

SPEAKER_01:

Um yeah, that's that's uh that's gonna hit a a good few levels. I I think anyone listening to this is uh probably gonna pick up 80% of what you said um because of the practical nature of it. And I I love those um six levels. Like that, that's a brilliant framework to consider in the context of this conversation. Um to wrap up, and I don't wanna there's a we've got heaps of ground to cover here. So anyone listening in, uh, if you'd like a particular topic to discuss, DM either Thomas or I and we'll we'll add it to our run sheet. But um we've got plenty of areas to explore uh in these weekly conversations, Thomas. But I want to ask you a a question outside of the firm. To date, since you've started really going down this rabbit hole of, you know, um AI and everything around the impact of it and the potential of it, has it enriched your life in any significant way? Incredibly.

SPEAKER_00:

Incredibly. I always have Claude open with me. I always have perplexity open with me. I just every single second, I am I'm all I'm doing something with it. I'm trying to constantly refine ideas. I'm trying to, I'm throwing transcripts of all the meetings I've ever had and like analyze how I could have said that better. What was the sentiment of that person? What did I say that that resonated really well? Or I'm having these thoughts. Um, help me try and like, help me try and help me try to consolidate this into a concise way. Like, I think we can obviously tell, yeah, I don't have uh a good ability to be able to consolidate it together into concise ways. It it feels like my second brain nowadays, Renton. I I don't use it to think for me, but I use it to help me just explore all these different paths and go, yeah, I like a bit of that, I like a bit of that, I like a bit of that, pop it into a notepad. So notepad's the other thing I have open, probably just as much as Claude, because I'm picking and choosing from it and go, all right, here's here are the things I like from what you said. Go into a new chat, throw it back into there and see how we can explore this. It's it's just, I feel like my the way I've been able to process through ideas has just increased dramatically. And of course we have to be careful. We can't just whatever it's telling us is some sort of, it's it's not actually thinking. It's this black box, it's uh it's probability, it's hallucinating things, it just seems very useful. Uh the way we have to use it is we have to just use it to inspire ourselves to think about things that we haven't considered, you know, to be able to break out of that box and go, oh, I didn't think about it that way. Or you're right, there are parts to this analysis that I haven't figured out yet, uh, haven't thought about. So I should probably dive into it a bit more. Um it's always there, Brenton. It's I I don't know what I'll do. If when Claude's not working, the day kind of feels a bit dull. When it's there, I feel like I just get so much more thoughts processed.

SPEAKER_01:

Um tell me, I I want to keep going on that actually, because I find this, I find um this whole subject interesting from uh a psychological perspective and how we operate as humans. So part of me thinks everything you've said there is like amazing. Like it's enriched your life in those ways. Part of me goes, there's there's like a there's there's something slightly unhealthy like bubbling there, you know, in terms of our dependence on it and uh how much blind trust and faith we put in in these things already. And I'm guilty of that as well, like in terms of what these tools are telling me. I just go, oh, I accept it as true pretty quickly. Do you do you see that playing out um with any concern? Or um, yeah, what are you what are your thoughts on that? Is there any layer of unhealthiness to it? Yeah, you gotta be really careful about that.

SPEAKER_00:

Uh maybe the most immediate thing I've been noticing is but my natural numerical arithmetic has gone worse because I'll just type it amount, I'll give it like a numeric problem, like uh kind of work through it. I'm like, oh, I can't, why can't I recall eight times six as fast as I used to? I don't know. I never ask it those sort of questions, but why is that? I'm just noticing like, why my why is that harder? Like that's that's strange.

SPEAKER_01:

Because have you heard like, you know, I it's interesting because I only was I was laying there the other day, and for some reason, all these old telephone numbers from my childhood, you know, my old home landline number just kept coming into my head. And I was listening to someone recently, and they were like, try and you know, your your four best friends tell me their phone numbers, and you just don't know. Like back in the day, you would know, but there's no need to know anymore. And uh in the context of that, it was, you know, what is the version of not knowing your phone numbers anymore of this whole new phase of technology? And in some ways, without being a a doomist or a or negative about it, I do feel like a bit of a frog in a pot, you know, of the water heating up slowly. Um, but uh, but at the same time, there's so much good that comes with it. So I was just interested to get your thoughts um on that. I think the only solution, Brenton, is neurolink it.

SPEAKER_00:

So it's not really lost in my hand. Yeah. But I think, you know, it's you can look at, you know, I've met people who go, oh, GPT told me my IQ was in the top 0.5% in the world. And I would say, hey man, you need to you need to be careful with that. Like you, it's it's it's you know, it it it's not actually it it's not actually assessing you. Okay. It doesn't actually, it's not us, it's not an actual psychologist. It's not actually doing a test. And we're already seeing it on the news, right?

SPEAKER_01:

People are doing you actually want to play that person. I know you, and I know that's not fair.

SPEAKER_00:

Yeah, well, it's it's there is it just gets dark, right? Like, you know, you got people um where they're talking to GPT and it's just confirming with them that their mother-in-law is poisoning them, and then they add all those things.

SPEAKER_01:

You do start to hear some of the the other side of the stories which don't have great endings, which with any technology is gonna come that side of the the game, right? So I find that very interesting. I I will park that for a separate conversation because I would like to look at it. Like, I think it's only right to look at that side of it um without pay placing too much attention on it. But it is important to look at that.

SPEAKER_00:

Yeah, you have to you have to keep exercising your critical thinking. I think whatever it gives you, the first question you should be asking is how is it how is it wrong? Like, what is it missing out on here? It's it's what what parts of the picture is it not considering? And I think we all have felt that using GPT. You talk to it for a while and chat gets a bit long and it's forgetting pieces from before. And we're all building this sort of intuition that it's not it's not taking everything into consideration. Uh it's not giving you the it doesn't always have the full context, which is the challenge of it. So I guess the first question we should always ask when we're using it is it it doesn't know everything. Like it's only giving you it's only processing part of an aspect of a question that you're asking it and based off what it's known. So yeah, if you I don't know, we'll see in five years. We can't see. We can't see if we're here in five years. Yeah.

SPEAKER_01:

But that was a really good conversation, thank you. And um yeah, I'm looking forward to these. Um and just seeing how this landscape plays out. It does I think it leaves no one untouched in some way, shape, or form. So it'll be it'll be interesting to see how we can apply it to our audience's benefit as well. Like uh ultimately we want the people listening to this to take this back to their business and in in input it in certain some way that helps them build a business that gives them more freedom. So thanks for tuning in to this episode of The Wise Way. If today's episode sparked an idea or helped you see things differently, please don't forget to leave us a review. And if you haven't subscribed to the podcast on your favourite platform yet, please go ahead and do that as well. Let's continue the conversation here through YouTube or any other social platforms that you can find us on. And just remember, if you're not a subscriber of our weekly Friday tip newsletter, you can get that to your inbox every week going forward. Whether you're starting out or scaling up, you don't have to do it alone. Let's build a business that works for you, the wise way. We'll see you in the next episode.