David Mitchell [00:02:10]:
Well, good morning, everyone. Welcome to today’s webinar. Today we’ll be exploring the practical applications of AI tools for bookkeeping and we’ll be looking at the opportunities, tips and possibly some lessons learned from an expert in the industry. So let me welcome Cassandra.

Cassandra Scott [00:02:32]:
Hi, David. Thanks so much for having me along today.

David Mitchell [00:02:35]:
You’re very welcome. Just to let everyone know, my name is David Mitchell. I’m the CEO of grassroots it. I have a background and special interest in helping teams thrive through proven technology. Now, AI may not be proven technology or considered proven technology by some, but it is here and it is definitely here to say stay. So I am excited to invite Cassandra here today to help us understand how it is being utilized in the world of bookkeeping and how we might leverage it for our own business. I’ve worked with Cass for over five years and she’s always professional, insightful and a recognized leader in her field. So welcome, Cass.

David Mitchell [00:03:18]:
Could you please let us a little bit more, know a little bit more about your journey so far and what you are passionate about.

Cassandra Scott [00:03:24]:
Yeah. Thank you, David. Cassandra Scott. I’m based out of Brisbane, and for the last 20 years, I’ve actually run my own bookkeeping practice. And out of those last 20 years, for probably about the last ten years, I’ve been heavily involved in the class accounting space, particularly Xero, and sitting on some sort of national panels with Xero, but diving deep into everything technology within the accounting and bookkeeping industry. And in the last few years, I’ve really started to explore opportunities with AI technology within bookkeeping practices and also the services that are provided to SME’s, either internally by their own bookkeeping teams or for those businesses that use external teams to provide that support. That support can be streamlined and leveraged using AI technology. So one of the things I’m actually very, very passionate about is actually increasing the professionalism of the bookkeeping industry as a whole and making sure that small businesses are getting the best information that they can around their financial records and data at the right time, at the highest level of quality.

Cassandra Scott [00:04:36]:
And so that information can be used by the business owners with their business stakeholders for whatever purposes may be there. And AI is becoming a significant component around doing that successfully these days.

David Mitchell [00:04:50]:
Awesome. It is great to have you here. So without further ado, we have to acknowledge that AI has exploded. It’s enhancing nearly every app that we use in some way. Tell us a little bit more about what you’ve seen in the bookkeeping field.

Cassandra Scott [00:05:08]:
Yeah, that’s an interesting one, David, and I think one of the things that happens quite universally is there’s the general term AI is actually used for a combination of both AI and automation. And I actually don’t think that’s necessarily incorrect. I think that’s probably quite relevant in that automation has been something that our industry has been exposed to for probably the last ten to 15 years. That the processes that we used to undertake in collating and providing financial data and working through the accounting and bookkeeping processes have been automated heavily. Number of the tools that are actually in the market and you know one of the simplest ones we can look at is the accounting software, the xeros, the myobs, the qbos and leveraging off of things like automated bank feeds bringing data in in an automatic way. The last few years though, the marketplace has actually seen, as you said, the evolution of AI within the industry. And this is where it overlays against automation and AI is where you’re starting to use the data and use the process of automation and allow the technology to start to make decisions on your behalf. So if we simply define automation it’s tech performing tasks or processes with minimal human intervention.

Cassandra Scott [00:06:29]:
But when we overlay AI on top of that it’s actually the machines and the intelligence, sort of mimicking human intelligence, learning from data, recognizing patterns and making decisions on that learning. So the AI in its purest form is now sitting on top of automation, which has been around and which most people I think have been using quite significantly and well for at least the last ten to 15 years.

David Mitchell [00:06:55]:
Great, great. And as you’ve got up here, does it enhance productivity?

Cassandra Scott [00:07:02]:
Absolutely. And how so? Yeah, so it definitely enhances productivity. So as a bookkeeper in practice for the last 20 years, one of the most frustrating things that I ever had to come across was dealing with those paper based receipts that you get off of clients or even working within businesses. And I spent 30 years working within businesses SME’s before stepping into practice. And the bane of my life was always dealing with the paperwork and the mundane and the routine and the, to a certain extent the simplicity of the paperwork and you know, you, it’s a bit like monkey see monkey do in some instances it becomes very repetitive, it becomes very time consuming and AI and automation has seen those things simplified and being able to be sped up in terms of the bookkeeping in the account cycle. So if we look at, you know, simple tools like document extraction or data extraction tools and we’ll talk more about those a little bit further on, instead of me having to sort of take in each individual piece of paper and take the data off of that and manually key it into the accounting system. We now have technology and tools that do that. And over the years, those tools themselves have evolved in that they used to just sort of do some OCR recognition, take the data out, punch it into a certain field without any intelligence around what it was doing, but it was still simplifying the work process.

Cassandra Scott [00:08:30]:
And now today, those same tools are applying the AI logic over it and actually making decisions based on other information that’s available to them. So what, you know, for a client maybe used to take me two or 3 hours in terms of processing paperwork, could now, in theory, be done in between 30 minutes to an hour using technology. Bank feeds are probably the most common one that most people at some point have been exposed to. Having to worry about getting that information from your bank or waiting for the postee to arrive at the end of the month with the bank statement, which could be missing the two days before your end of month or end of year, but having that data coming forward to you on a day to day basis. So real time work with real time data for real time information. So absolutely, it enhances productivity.

David Mitchell [00:09:19]:
Great. Well, let’s dive a little deeper then. So before we do, you’ve already given us some practical examples how well let you go a little bit further, deeper into the tools to run your business. Yeah, and I’ve got some questions for you after you do.

Cassandra Scott [00:09:42]:
You’ve always got questions. Good, fun, love questions. So, AI tools for bookkeeping, I think when we’re talking about bookkeeping, and if, you know, the audience will either have somebody internally providing and undertaking the bookkeeping work on their behalf within their business, it might be an employee, it might become somebody that comes in, you might actually even be doing it yourself, or you might be engaging with an external service provider, and that might be done through a contracted bookkeeping practice, it might be your accounting practice who’s doing it. But fundamentally, when we look at AI tools for bookkeeping, I don’t like to just restrict it to what are the tools that are used to facilitate the bookkeeping process? Because I think there’s a big discussion around also the tools that run your business and that the bookkeeping tools are actually a subset of those. And what we’re talking about here, if we link back to the conversation with productivity, is actually being more productive with our time, and if we’re more productive with our time overall, that frees us up or it frees our team up, or it frees our external service provider up to do other things for us and within our businesses. And those things can actually be far more strategic than literally sitting here and just punching data in utilizing skill sets that may at the moment be untapped, because time has been committed to what I call the routine financial management processes within a business. So if we’re looking at, I might jump to the tools to streamline your bookkeeping. You’ve got your obvious ones out there, you’ve got your zeros, you’ve got your myobs, you’ve got your qbos online.

Cassandra Scott [00:11:24]:
They’re all working off of cloud based. You’ve got immediacy of information, you’ve got the ability to store documents in the cloud. Smart learning already occurring within those tools. That’s here and that’s now and that’s being worked on today. But where this is moving forward within those accounting products, and it’s not too far off, and there are actually already some instances of it in the market, is how to then look at the jobs that need to be done within the bookkeeping cycle. And how can we automate and simplify and iterate those jobs to be done. So if we use Xero as an example, they’ve just come out with a new tool called Jax. Jax.

Cassandra Scott [00:12:06]:
It’s very, very new. And I think in some instances it’s still in beta testing. But if you’re a sole trader or a business, a trades business perhaps, and you’re out seeing your clients, or you’re a consultant and you’re working away, imagine being able to go into your accounting software, open up a screen and just saying to it, look, based on the information that was in my diary, I’ve actually finished working with Bob Smith, my client, in delivering those plumbing services that were in the calendar invite. I did 3 hours with him. Can you raise the invoice and send that out for me? Now all you’ve done is had a very, very simple conversation with the tech around doing that. But the output of that from a technology perspective is an invoice is raised to the client, it’s raised with the correct accounting treatment around it. It’s raised at the number of hours that you’ve iterated, it’s raised automatically at the hourly rate that you’ve perhaps stipulated somewhere else in the process, or that your tech is actually looking at and extracting. And it’s sent out the door automatically to your client.

Cassandra Scott [00:13:11]:
And you’ve actually not even put your hands on the keyboard. So when we’re talking about bookkeeping tools, what we want to correlate them back to and actually think about in terms of their utilisation is jobs that are to be done. If we look again at that from an accounting and bookkeeping financial management process, you know, imagine logging into your and accounting file, whether it be QBO or NYMB or Xero, and imagine that you’ve got what we call an agent, a bookkeeping agent. So at the moment, you’ve potentially got a person within your practice or externally to your practice that does this. But imagine if you’ve got somebody or an agent, a tech agent, that actually jumps into your email account to start with, extracts out of that email account. Any supplier invoices that have been sent to you overnight automatically moves those through into your accounting system. And that may be through another document extraction tool that actually pulls out the relevant data, the supplier name, the amounts, the dates, all of that information. Based on the learning that it’s undertaken by utilizing data that’s already in your accounting system, it codes that data and gives it the general ledger categorisation and also the GST treatment categorisation that is appropriate.

Cassandra Scott [00:14:31]:
It puts it into your bills to be paid. And you may also have a rule that sort of says, at the end of every week, I’d like a list of all of the bills that need to be paid and have a batch payment file that’s actually prepared for those bills that need to be paid. Now, in theory, that may be done without any human intervention, with the technology that is currently evolving into the marketplace, and I can imagine there’s a number of people sitting there screaming at the moment about the human side of it. What happens to the people? People don’t actually disappear in this process, but what happens is their role in the process actually changes. So that person that normally would log in instead of logging in and having everything from scratch to deal with is potentially in there, dealing with any exceptions that have been flagged. So something may have come through that’s actually not quite able to be determined by the AI agent. Based on everything that it’s learnt, the human comes in and actually starts to look at it and makes the human decisions that need to be made. So, look, this sounds really simplistic and it sounds really scary in the concept of removing people from a process, but this is where technology is going.

David Mitchell [00:15:48]:
Very interesting, Cass. So a couple of questions come to mind there. One is human oversight and potentially recognizing errors. I mean, we’ve all seen the OCrs that get it wrong. And then my other question is around workflows. Now, you call those jobs to be done. It seems to. How do you change from where we are now using Xero to implement? You know, we’re talking about zero at the moment to implement something like Jack’s, there must be different workflows, different training.

David Mitchell [00:16:20]:
How would you go about that?

Cassandra Scott [00:16:21]:
Yeah, there certainly will be different workflows, David, but what you’re going to find is the, the technology suppliers that are implementing this will actually take you on that pathway with them. So if we use Jaks as an example with Xero, and as I said, it is in beta at the moment, it’s going to be released to a cohort of users in certain industries where it’s potentially been identified that there will be a higher level of uptake. What will happen is people will start to work with that technology and start to redesign and reconfigure their workflows, and that information is actually going to come back. And whether it’s AI that starts to bring this information back or not, my expectation is that it will, but that information is going to come back and be disseminated out through businesses. So I think one of the things, interestingly, we’ve seen with the increase in technology and cloud based technology and some of the AI tools that are in the market is that there’s a lot more open source sharing of knowledge and information. So whereas how we did things and why we did things and the methodology that we did used to be very, very stoked piped. There’s a lot more hive mind thinking about what is best practice and what is standard practice across a particular industry. So when you’re looking at these sorts of workflows and features, a lot of the tech that’s developing, and if you look at the three, three incumbents in the accounting space, zero, nymb, qBo, they’re all about data.

Cassandra Scott [00:17:55]:
And when you’re looking at the data, don’t think. They’re not just having access to your data. They have got hundreds and thousands and millions of data pieces and they’re able to see how people are using the technology and they’re going to come out and say, we’ve implemented this new piece of AI or tech within our product that is going to do this for you and your business. And here are our recommendations about how you and your team can interface with this to gain its maximum potential. Now, this isn’t going to happen in, you know, two years or five years. It’s going to happen in, in ten years and 15 years. It’s going to start now. Absolutely.

Cassandra Scott [00:18:37]:
But it’s actually going to be an iterative change in those workflows as we, I think, intellectually overcome the concept of this is something that I need to sort of have my hands on and do myself. So the way that we think about our work and the way that we think about our workflows is actually going to change. And we, it’s a ride that I think that we’re all going to be going on. And, you know, I’m the first one here. I’m sitting on that ride already with some of the things that I do. And it actually, you know, in some instances, you know, it can be a bit daunting and scary. And, you know, I came from an era where when I started working, we didn’t have computers. So I’ve seen the transition from, you know, manual ledgers into desktop computer technology into cloud based technology, and now I’m sitting on the advent of, you know, AI technology in that process.

Cassandra Scott [00:19:28]:
So it’s an interesting ride that we are going to go on and how we’re going to adopt that and those workflows in our business is still something, I think, that’s to be surfaced. But, yeah, it’s absolutely going to change. David.

David Mitchell [00:19:43]:
Great. I do want to dive into an example of a couple of the tools, but a question for you first, you do the bookkeeping for grassroots it. So this is how we’ve worked together well, and I know there’s been a few times that you’ve raised anomaly in our numbers and you do that via an AI tool.

Cassandra Scott [00:20:02]:
Yeah.

David Mitchell [00:20:03]:
Can you tell us a little bit more about that?

Cassandra Scott [00:20:05]:
Yeah, so we’ve been using a particular AI tool called expert, and it’s on the list there. There’s a couple of others out there that do it, but expert is one that we have utilized. An expert’s been in the market for probably about four or five years, so it’s not new. And, you know, when working with, with the likes of grassroots, providing the bookkeeping services to them, one of the things that we’re always highly focused on as a bookkeeping practice and, you know, whether we’re working for David or even my own business is about the concept of data quality. And our methodology for reviewing that historically had been very, very manual. We’d run big reports, we’d have a look at the data. We’d try and sort of dive into it and go, oh, that looks a bit anomalous. Flag it, open it up, have a look and go, actually, no, that’s okay.

Cassandra Scott [00:20:52]:
Close it back down and then move on to the next one. So that was really, really time labor intensive. For us, we connect up data quality tools, and what it has the ability to do is review that data against a number of different algorithms, AI driven algorithms, for things that are specific to the bookkeeping industry. So for those of you that work with bookkeepers, we can be a little bit anal sometimes about the things that float our boat in terms of data quality and data accuracy. So, you know, this tool has been developed with the bookkeeping compliance cycle in mind. And what are the big things that actually trigger us when we’re reviewing data? And instead of me having to source it as a human tech, is now doing this, and it’s doing it at a far, far more granular level than I ever can. So the technology is not just looking at the transaction in the accounting file, it’s actually also looking. Looking at the attachment to the transaction in the accounting file.

Cassandra Scott [00:21:54]:
And it may be going out to other external resources. And ABR, the australian business register is one of those resources, and it’s looking at the client record or the supplier record in those resources and saying, hang on a sec. By the way, did you know that this particular supplier is no longer registered for GST, or has had their ABN cancelled or should be registered for GST? And the transaction that you’ve got in your accounting software actually contradicts with what we’re seeing from a legal perspective. Now, that mightn’t seem important to a lot of people as bookkeepers. It’s something we’re aware of. But the consequences of that for somebody like grassroots are if the ATO comes knocking and they’re audited, the ATO doesn’t say, we don’t care that you made a mistake. They’re going, we’ll actually like to have that GST that you’ve claimed back. So it’s about saving our clients money, not necessarily today, but potentially down the track.

Cassandra Scott [00:22:53]:
And some of the things that will surface is, you know, duplicate payments. One of the clients that we worked with, I think we found about $4,000 worth of duplicate payments to suppliers. And they were incredibly grateful for that. You know, in the scheme of the size of their business, was it a lot of money? Probably not. But in terms of, did they want to be overpaying people? Absolutely not. And smart business people aren’t. So data quality in any business, in financial, financial information, is paramount, because what you need to be able to do as business owners is rely on that information without having to question the integrity of it before you need to rely on it. And what I mean by that is, if, you know, you’re perhaps looking at financing for new equipment, a new project expansion, whatever you want to be able to pull your financials today, get them to whoever needs to review them to make those decisions that are real time decisions.

Cassandra Scott [00:23:49]:
And often there’s some timelines attached to that sort of response that you need. And having to wait for another month or, you know, even a week for your bookkeeper and accountant to just go through and check that the data’s right may actually preclude you from stepping into whatever it is that you need to do. So data quality is an enhancement against an enhancement alongside of real time high quality when you need it, financial data. And that’s why we use those tools with our clients, and you know, where they’re absolutely invaluable from our perspective.

David Mitchell [00:24:24]:
My feedback is that we find out about the issue very soon and we can fix it quickly with a tweak or a phone call or an email, and we don’t find out about it months down the track and then have to back date and what have you. So many people. Many people in the call today are internal bookkeeping or an accounting function is dext, I think. Did you say dext or expert? Expert.

Cassandra Scott [00:24:50]:
So expert is one for data analytics. It’s more aimed at the bookkeeping practices more so than the bookkeeping, and accounting practices more so than individual businesses. But if you’re working with an external accountant or an external bookkeeper, ask what tools they’re using to look at the data quality of your financial information, because there’s more than one out there. Dexd precision is another one. And you’ll see that we’ve got dext on there. And under sort of two headings. One is Dex prepare, and the other there is dext precision. It’s an expansion on what Dext historically has done.

Cassandra Scott [00:25:25]:
And historically, Dext has been a document extraction tool. So your supplier invoices would go into dext. Dexd would read the data on those, extract that data, and move that transaction pre coded with the attachment into your accounting file. They’ve stepped into sort of more of the data quality side as well. Mindbridge is another one that looks at data quality. But, you know, ask the question of your internal bookkeeping team or your external bookkeeping service provider. How are you qualifying the quality of the data that is in our accounting system? Because if they’re not starting to do that now, particularly your external providers, they will start to be challenged by people, people that are using these tools. And, you know, one of the comments I often hear is, you know, AI’s going to replace humans.

Cassandra Scott [00:26:17]:
And my response to that, and it’s out there in the public domain from other people, is AI is not going to replace humans, but humans who use AI are going to replace humans who don’t. So even internally and often, one of the biggest pieces of resistance that you will find in adoption of any technology, and I’m not just talking about AI technology, but any sort of technology can be internal mindsets. And I sit here, and as I said earlier, I started working when I was still using ledger books. But I’ve made it a purpose of mine to try and keep up with technology. And I appreciate that there’s often factors that preclude people from stepping into that space. Some of it is just sort of hands up in the air, I don’t want to know about it. Others, because it’s unfamiliar, they’re uncomfortable with it. And what I would do is challenge people to start thinking about what is next and look at how these products can be adopted within your businesses to enhance your data quality.

Cassandra Scott [00:27:20]:
You know, there’s stuff out there for practices, there’s stuff out there for accounting and bookkeeping practices, but there is stuff out there already for businesses to use. Now, interestingly, the story around expert is an interesting one, because the guys that set it up were actually business owners. They had small businesses themselves and actually got screwed over as a consequence of edited data, poor quality data, data that had been removed from their accounting ledgers. And the trigger for setting up a product like expert was to protect other business owners from going through those same challenges. And, you know, they lost huge amounts of money because other people within their firms were tampering with their accounting records. And that info, that story is actually on their website. So, you know, this is what we’re talking about. And, you know, every, and I know grassroots is very, very focused on, you know, cyber awareness with their clients and the prevalence of cyber intercepts.

Cassandra Scott [00:28:25]:
Having data quality tools can actually be a foil against some of that as well, because you’re going to be able to see if somebody is perhaps doing something they shouldn’t be or if data is being amended and you’re not expecting it to. So it’s the whole cycle of work, not just the financial and the bookkeeping cycle, that these sorts of tools actually support and enhance.

David Mitchell [00:28:47]:
Right. So can I just quickly circle back to the internal it team? Is there a name that you have at the top of your head that they could go and look at, just in terms of the data quality tools?

Cassandra Scott [00:28:58]:
Look, I’d start with Dex precision. Ada might be another one. Even though we’ve got that under financial reporting, you could potentially look at expert. It is more focused on bookkeeping and accounting practices, but that doesn’t mean to say that a business couldn’t use it. The other thing that expert has in it is great workflow management tools. So that might be an advantage if you’ve got a larger business and a larger financial management team. And the work that’s being undertaken is done by different people within that team. But I would look at, you know, whilst I say experts, not aimed at small businesses, there’s nothing that precludes a small business from actually using it.

Cassandra Scott [00:29:36]:
But any of those tools I would be more than comfortable in suggesting are worth looking at from a data quality perspective. Even if you’re using Xero these days, they’re starting to build some data quality, the aspects into it as well. Not sort of quite as front and center as what some of those other tools do. I can’t speak more specifically about NYOB or QBO, but with all of the accounting products, it’s like the race to the finish. They’re all doing something. Somebody catches up and goes ahead, the next person does it, they’re continuously iterating. And data quality is a huge focus for everybody out there at the moment.

David Mitchell [00:30:18]:
It’s a point I’ve got here, actually. The vendors that are rushing to get their AI products out the door and into the hands of clients just because they’re there doesn’t mean to say we need to use them or how do we choose. I mean, this is a very broad question and it’s something that you probably can’t answer, but do you have any insights on which tool and will that change over time? This is a journey, and as you said, they’re all catching up with each other. No one wants to change accounting systems.

Cassandra Scott [00:30:46]:
But, yeah, look, I think strategically, when you’re looking at selecting any tools to support your business, one of the key things I’m always looking at, you know, for my own business, but when I’m talking with clients, is integration. So what you don’t want is technology and tools within your business that don’t integrate with other things that you use. And, you know, in bookkeeper accounting world, we call it our app stack. So what is our app application stack? What do we use and how do we use it, and how does it all integrate? You don’t want to be redoing the same thing in three systems. And that’s quite the antithesis of the whole idea behind artificial intelligence. It’s about simplifying workflows, not building complexity. So if you’re looking at tools to support the bookkeeping cycle within your businesses or practices, if there’s any bookkeepers on the call, look at what’s integrating with the existing tools that you are using using. And when I talk about integration, I actually look at the depth of the integration because I have seen a number of tools out there that say we integrate with Xero, but that integration is not necessarily a live integration.

Cassandra Scott [00:31:55]:
It’s I need to push this button to integrate, or it actually only integrates at 10% of the data linkages that I would expect. So if you’re looking and assessing any tools out, what type of integration is it? Is it a push of data or is it a live integration? And does it talk back and forth? And it’s usually through APIs, what specifically integrates. So are you looking at only a small portion of the data set moving between the systems, or are you looking at all portions of the data set? And is that data set what’s needed, or are you still going to have to top and tail it in another way? Is it changing and will the vendors change? Absolutely. I think, David, and you know, 15 years ago, the incumbents in the market from accounting software were the myobs and the QBOs or the Quickbooks. Back then, zeros came in and did some steamrolling. That’s the market. And I suspect that there are products absolutely being developed out there at the moment that are likely to hit the market any time from tomorrow onwards that may disrupt some of the incumbents that are in there at the moment. So, you know, AI is, is tech faster.

Cassandra Scott [00:33:15]:
We’ve seen the evolution of technology even over the last 50 years. And that chart of how it’s being implemented and how fast it’s moving is only just going to become steeper and steeper rather than flattening out. So I think from an internal business perspective, the best strategy I would look at to keep an eye on the market is, is talk to your service providers, your external service providers. So if you’ve got an accountant or a bookkeeper, it’s amazing how close to technology those people actually are at the moment. Perhaps look at somebody within your business as being a tech champion. Look at places, you know, LinkedIn, funnily enough, is a great way to sort of keep on top of, you know, what tech is out there. There’s probably some publications out there that you can sort of subscribe to that sort of keep you up to date on what emerging technology is there. I’m not a big fan of chopping and changing just for the sake of chopping and changing.

Cassandra Scott [00:34:16]:
And even sometimes if you’re getting 80% of what you need out of it, it’s probably worth sticking to if you’ve got all of the integration and sort of ticking those boxes as well. But I think the important thing is having an awareness of what’s out there and how it’s impacting not necessarily just your business, but potentially also your industry. And how is it going to allow you to step into the jobs to be done type concept, which, if you’re a business owner doing your own bookkeeping, if you want to be moving that off to the side and not doing that as strongly as you have previously, that frees you up to do the things that you want to do in your business. You know, you might be an awesome consultant or an awesome plumber or an awesome engineer, but you’re losing time now because you’re dealing with the day to day minutiae of financial management. So what is going to free up your time to allow you to do what you’re passionate about and doing best within your business?

David Mitchell [00:35:13]:
Great. Speaking of jobs to be done, one of a large job in medium sized organizations is approvals of expenses and things like that. How do you see AI assisting with that at the moment, and where do you see that going?

Cassandra Scott [00:35:28]:
Yeah, so again, David, there are already tools in the market that actually step into approvals processes. You can see that there with the likes of Dex prepare, Hub doc and approval Max. So all three of them work on the concept of you’ve got your supplier document that comes into the system. There could be rules and algorithms. And again, this is sort of the fine line between what is automation and what is AI, and automation is probably the first step with AI. These documents are looking, sorry, these tools are looking at those documents and extracting relevant information. You know, Dext prepare, as an example, has just bought out an approvals matrix. If this is from this supplier and it’s over this dollar value, then it needs to be approved by this particular person in the practice or business.

Cassandra Scott [00:36:19]:
Hub Doc’s got very similar features around it. Approval Max is an awesome tool for doing this as well because you can actually set up multi tiered approval matrices. It could be dollar value driven, it could be supplier driven, could be budget driven. So, you know, all of these tools are now out there where instead of the business owner perhaps getting a pile of paperwork in a Manila folder landing on their desk every day or every two days, where they’ve literally got to go through and check everything and hit an approval on it that can actually now be automated and simplified through technology. And the business owner might not care about a $50 bill from officeworks, but they might care about a $5,000 bill from officeworks and have oversight of that. So the $50.01 might be routed through to their office manager, who’s got the authority to approve that $50 office work spend. But the business owner might want to see that $5,000 spend, or do they require purchase order approvals before the expenditures actually.

David Mitchell [00:37:22]:
Well, that’s what my next question was going to be, whether we could do that earlier in the process.

Cassandra Scott [00:37:28]:
Yeah. Yeah. And again, this is where the human factor comes in, David, because these are just tools. This is everything that we’re looking at here is tools. You’ve got to be able to define how you want these tools to operate within your business. So if you’re a larger business that’s got, you know, maybe a management team, team leaders, you might have, you know, more than one person sitting in the finance roles, you know, how do you want the approvals process to run throughout your business? Because what happens in my business might be completely different to what needs to happen in your business. It could be because it’s the industry. You know, think about the trades and constructions and going to Bunnings and spending money at Bunnings.

Cassandra Scott [00:38:09]:
You know, I’ve worked with clients who have given Bunnings cards to their employees, an example, and we know that what’s been spent on the cards has not actually ended up on site. So, you know, how do you want those workflows to operate within your business? What are the risk assessments that need to be undertaken around your existing workflows? Where are the risk points for, you know, financial fraud? And the reality is that fraud occurs more often than not by somebody who is within your business. And there’s, you know, hundreds and hundreds of examples on the Internet. You’ve just got to google them to talk about fraud being undertaken by people within the business. So how do you, where are your pain points? And this is where you humans come into it, because they’re the ones that understand deeply what you do and how you operate and where those gaps in your process are and how technology can fill those and support those. One of the interesting conversations that we often have with clients when we’re talking about these types of tech solutions for those sorts of challenges is the cost of it. And one of the things that I would throw out to the audience is the whole concept of cost versus price. So yes, you know, every tool and, you know, we’re working in a, as a service, a software, as a service environment.

Cassandra Scott [00:39:24]:
And David, I know, you know, what grassroots does is very much involved in that space as well, or, you know, manage service agreements and that, but consider some of it in investment because if you’re, you know, spending 40 or $50 a month now, what is that potentially saving you in making sure money’s being spent on the right things at the right time, that there is a reduced risk of fraudulent activity being undertaken, that the data that you’re utilising is of the highest quality. You could be working with organisations that require audits for whatever purposes. If you’re sitting in the not for profit space, you know, have there been correct authorisations for that expenditure against budgets? If you’re working with grants, there might be particular grant reconciliations that you need to do where you can demonstrate those approvals processes. So think about the tools that are out there in the context of your business operations and get the humans that understand what you do involved in reviewing and assessing and understanding how those workflows would be implemented. I’ll take a breath now.

David Mitchell [00:40:33]:
Yeah, well, I mean, you’ve launched into a space that we are in all the time that you can implement technology, but is the organization, how do you implement that in an organization? There’s a lot of change management there, there’s a lot of planning, there’s a lot of thinking through the processes. Sometimes you don’t want to automate a broken process that you’ve got now. You want to identify a better process process and automate or enhance that process.

Cassandra Scott [00:41:03]:
Yeah, how?

David Mitchell [00:41:05]:
I mean, I’m sitting here and know a little bit about it. How would you recommend our audience? They go, this is great. What’s my next step? Like, how do I start this process?

Cassandra Scott [00:41:17]:
Yeah, look, you know, find the person that can do that and has the knowledge to do it. So I’ve actually done this for a couple of different businesses and literally what, what I do is I go into their business and stand by them side by side and follow the flow of information. You know, where does it start, where does it end up? And then ask 100 questions along the way to understand why things are happening. Why do you have to send this and photocopy it three times and send it to the guy in the room upstairs? And the answer to that can be, I don’t know, because that’s what we’ve always done and I think that’s probably the legacy question in a lot of practices and businesses, are we doing things because that’s the way they’ve always been done. But what is the benefit we get out of doing something? So I sort of quite often say to people, I’m inherently lazy. I like to look for the shortest way between a and z, and I don’t want to have to double back and do things twice. And, you know, we all hear about it in terms of workflow management. You only want to handle something once and it should move on to the next important point in that workflow.

Cassandra Scott [00:42:23]:
So I think sometimes looking at each of the things that you’re doing within your business and stepping through them and understanding what you’re doing, but more importantly, why you’re doing it, you know, is it being done because you had a business owner that had been stung 15 years ago by an internal staff member that ripped them off for $20,000, so they’ve just become hyper vigilant about every single penny that’s going out the door. And is it time now that, you know, that person perhaps takes a bit of a breath, understands that there’s technology solutions that can streamline what’s being put in as very mandraulic processes and, you know, have that change management conversation? It could be, David, at the simplest level, that you don’t have the right person in your business to be driving this forward. So if you’ve got a business owner who is quite strategic and looking at leveraging off of technology across their business, and they’ve got somebody in there that is absolutely resistant, then the question might be whether they’re the right person in the business. And I know that sounds like a terrible thing. I was on a webinar the other day talking about AI in bookkeeping practices, and one of the questions that floated up is, what if I employ a new staff member and they don’t want to use this technology? My response to that is, why are you employing them in the first place? The people in your business aren’t driving the way it needs to run. You as a business owner actually needs to do that and drive the culture and the leadership, the protocols around it. And, you know, as with anything, and I know grassroots has got an awesome leadership structure within there, and David and I have had some great conversations, but the leaders in the business need to drive it. It’s not necessarily somebody else within your practice that will drive it if the leader is resistant as well.

Cassandra Scott [00:44:15]:
So you might have a champion in there, but if the leader’s resistant, then you’re going to sort of come across roadblocks. So I think it’s assessing, you know, the humanity, humans within your business and, you know, who’s the right fit to be doing the right thing. Are they the right people to have with it? But also as a business owner, you need to assume a leadership position on this and not just do it from a tokenistic perspective either. Not just say, yeah, yeah, yeah, I want this, you go ahead and do it. You’ve got to be actively involved, because if the people that are working with you in your business don’t see you being authentic about it, then the chances of implementation and change and adaptation and adoption are going to be challenged. Stepping forward. So find the right people to help drive this is probably where I’m at and assess it from a workflow perspective, step through and understand why things are happening and why they’re being done.

David Mitchell [00:45:09]:
I think from my experience, the business leader needs to either understand what’s possible by their own investigation, or reach out to someone who does understand what’s possible and then pick the leverage points in the business and work out in which order to implement. So rather than doing a massive great big change, work out what the best order of change is. And so it’s not too onerous on the business.

Cassandra Scott [00:45:37]:
Yeah. And sometimes, David, even the tools that are currently utilised within businesses may still be the perfect solutions. And the challenge might be that the people utilizing them have just never been trained properly on how to use them.

David Mitchell [00:45:48]:
Absolutely.

Cassandra Scott [00:45:48]:
And I see that regularly, even with accounting software, I come across really weird things that are in play simply because there’s never been training. And I can’t overstate the importance of learning how to use any tool to its maximum potential. You know, whether you’re a chippy, learning how to use your new bandsaw or whatever, you want to be able to use it in the way it’s meant to be used and you get the training to do so. So, yeah, absolutely about that.

David Mitchell [00:46:22]:
And all the enhancements that are coming every month and every quarter.

Cassandra Scott [00:46:25]:
Yeah, that’s why my hubby has such a big Bunnings bill, because he keeps upgrading on his saws and I don’t know, but yeah, I understand that.

David Mitchell [00:46:35]:
One last question about tools before we move on, is we’ve spoken about the inputs now there’s outputs as well, and generally that’s in reports or other forms of communication. What are you seeing happening happening beyond the profit and loss, beyond the standard bookkeeping communications?

Cassandra Scott [00:46:56]:
That’s actually really interesting. And this is where I’m seeing some interesting AI outputs around financial reporting and analytics. So I think historically, as business owners and bookkeepers, what we’ve been used to doing is going into our accounting software and say, running the p and l, running our eye over the p and l or the balance sheet and going, hang on, hang on a sec, that looks a bit weird. And then perhaps exporting out data in very linear format. So export an account transactions list out to excel and then you sort of run through and have a look at those transactions to look at for any anomalies. What tools are now doing is saying, well, you can actually ask this piece of software to do an analysis on your balance sheet or an analysis on your p and L. And I’m not talking about just saying you’ve got, you know, assets exceed liabilities and you’ve got equity, or you’ve got liabilities exceed assets and you’ve got negative equity, all that. I’m talking about diving deeper and going, well, hang on a sec, do you realise, and if I flip back to profit and loss, do you realise your cost of goods sold has increased significantly over the last two months, but your revenue’s decreased? Do you think you might want to have a look at that? And then being able to interrogate that data further by saying, well, can you tell me if there’s a particular trend that is causing that cogs to have increased? That’s not usual in our pattern of spend.

David Mitchell [00:48:22]:
And is this a conversational, like, chat GPT conversational questioning of your accounting data?

Cassandra Scott [00:48:28]:
Yes. So expert’s currently doing that. You’ve got the likes of Ada and fathom doing it in different ways as well. But I can literally jump into expert, run up my business’s p and l or ask it to do an analysis of my p and l just by a simple analyze my p and l for the last six months and then I get their analysis and it says something and I go, oh, I might explore that. Can you give me more information? So this is where the way we’re going to be reviewing our financial data is going to change. And I think the other consideration there too, David, is as a business owner, the way we look at our data is going to change. But who else is looking at our business data? You’ve got the ATO looking at our business data. So when you submit your payroll via STP, or whether you’re submitting superannuation, or whether you’re submitting tax returns or taxable payment returns, you know, from a personal perspective, your data is with Centrelink and you know, all of those agencies, you’ve got data with, you know, real estate, when you purchase property, you’ve got data.

Cassandra Scott [00:49:39]:
When you purchase motor vehicles or boats or, you know, put money in and out of your bank account. So AI and financial reporting and analytics isn’t just happening within our businesses, it’s being used by other organizations and agencies to analyze your data against a greater subset of data as well, to look for any anomalies and float that to the top. So you just have to, you know, look at some of the ATO reviews that are out there at the moment. They’re using massive, massive volumes of data to try and identify the people that are buying boats that don’t have the income that matches it. So, you know, we can use it internally in our businesses, but it is already significantly being used by other organisations and agencies.

David Mitchell [00:50:27]:
Very good, very good. All right, we shall move on to leave a little bit of time at the end for questions and answers. The next slide is about what other tools, general business tools in here that use AI. Now this is not the purpose of this webinar, but there are lots and they’re evolving all the time. So my question to you is, what have you used? What are you seeing your clients using? We don’t need to go into a deep dive, but what do you think?

Cassandra Scott [00:51:00]:
I think we’re remiss if we don’t talk about these tools in the context of bookkeeping. So as a business owner and a practice, I’ve used some copilot stuff, I’ve used Grammarly, I’ve used chatbots. Absolutely. Probably haven’t stepped into sales and CRM so much because I just don’t run that. Employment hero. Yes. Zapier otter AI. Love my otter AI.

Cassandra Scott [00:51:26]:
Fireflies for note taking. So now I don’t have to sit in a meeting and take copious notes. I can actually just get a note, could do that for me. And surface the summary that I can then pick up in 2 seconds flat and email out to whoever I’m communicating with. You’ve got them. Canva is a really, really cool one that’s out there and I think most people have even got access to that as free chat. GPT is the obvious one that sits there. So yes, they’re all enhancers of productivity within your business, and some of them are.

Cassandra Scott [00:52:00]:
It depends on what you’re wanting to do. So there are some tools that are sort of more appropriate, depending on what it is. So canva, for instance, is great at your design and creativity, and other things are project management, note taking. If you’re working in meetings, note taking is probably the simplest one to step into. I now no longer have paper notebooks with doctors writing in them. It’s all nicely typed out and searchable as well. So. Yeah, absolutely.

David Mitchell [00:52:29]:
Yeah. I have to give a shout out to Microsoft Copilot. We were talking about integrating earlier and if you are already using Microsoft Office rather than using also a multitude of other tools, copilot, over the top of word, Excel, PowerPoint, Outlook and Teams does a fantastic job of utilizing AI and integrations there. I can join a meeting five or ten minutes late. That rarely happens, but I can and ask it to give me a quick summary and it just gives me a very quick summary of what’s happened. So I can participate in the rest of the meeting. Yep, I can ask it questions about. I cannot participate in the meeting and ask it questions about what are the key points that were discussed.

David Mitchell [00:53:18]:
Could be a project meeting that I couldn’t become, couldn’t be a part of. So most of us have Microsoft office. So that copilot would absolutely be something that I would be looking at for that. Yep, very good. And challenges and considerations for implementing AI. This is something that you wanted to specifically discuss. So we’ve probably thought about a few of them and discussed a few of them, but other things that we’ve missed.

Cassandra Scott [00:53:48]:
Yeah, and I think it’s really important. David, there’s a couple of questions that have come through chat about data security and data privacy, and they are paramount in anything that you do. So the examples that I gave before with the P and L, we’re not putting those into chat GPT, we’re actually putting those into a secure, or we’re actually using a secure data set. So our client’s data is actually not being exposed in an open way. What’s happening is there are things called large language models. I would suggest Googling to sort of get the finer definition, but the large language models are often what sits across data sets, and it’s those models that you use to ask the questions, and it uses the intelligence that it’s taking from the broader open AI type environment, but then applying those questions and interpretations to the closed data sets. So data privacy has to be absolutely critical. And if you are wanting to dip your toe into the likes of the chat GPTs, anonymize your data.

Cassandra Scott [00:55:00]:
Don’t put anything in there that you wouldn’t want to share with your best friend on Facebook or out in the wild web. An interesting thing if you’re using chat GPT is to go in and ask it what it knows about you and what it can tell you. And that gave me some really interesting insights a couple of days ago when I did that, and it extracted information I didn’t know was actually publicly available, right down to members of my team, which I thought was pretty interesting. So yeah, please be conscious of data. Data sovereignty is another thing. So when you are looking at these solutions, where is data being held? Ask those questions. And any vendor who is not prepared to answer that information would be one I would be crossing off my list. That should be information that is readily available to you.

David Mitchell [00:55:56]:
Yeah, data sovereignty is a good one. And from Microsoft, which is the majority of the tools we use, our clients can select, or we do it on their behalf, where their data is housed. And it’s for many legal reasons, privacy reasons, certainly Europe, GDRP is very strong on that, but that is a very important one. Like, for example, mailchimp, for example, you put a lot of data in that that is not domiciled in Australia, or it wasn’t, unless they’ve changed. Recently, Stuart has raised the question that you were talking about with putting financials into chat GPT. Now, I don’t know as much about chat GTP, can’t even say it as I do copilot, but certainly with copilot you can. If you have the copilot and you’re selecting your workspace as opposed to web, you can query your own data because as you said, it’s using the large language model to communicate. We use them in the communication tools, but it’s only communicating and not training the large language model on yours.

David Mitchell [00:57:08]:
And I believe the paid subscription of chat GPT is similar to that, but I don’t know for sure. So as Ben says there, we will get back to Stuart on that particular one.

Cassandra Scott [00:57:19]:
Yeah, yeah, I’m not sure about that, David. And the comment I would make is that if there’s anything out there that’s free. Why? Because you’re the commodity. You are the product. Yep. You are the product if it’s free. And this goes back to what we were talking about previously about investing and the concept of price versus cost as well. So free is not always cost effective.

David Mitchell [00:57:42]:
Very good. Anything further in challenges and considerations?

Cassandra Scott [00:57:47]:
Excuse. Yeah, look, I think some of the questions is about the data accuracy as well, particularly when you’re using the open models like chat GPT. And you can see I’ve got some screenshots on there and you know, you can go out and do this yourself, but it’s a bit funny when you ask chat GPT how many r’s are there in strawberry, and it comes back, you know, very definitively that there are two r’s. But then when you ask it to explain itself, the explanation appears to be, you know, very, very logical from a chat GPT perspective, but we know it’s absolutely incorrect from a human perspective. So this is where it comes back to. The humans are not going to disappear, because all AI and chat GPT, particularly at the moment, is doing, is looking at masses and masses of data and drawing its conclusions. And this is why something like chat GPT may not be the best solution for perhaps doing data analytics, because you don’t know how compromised the data is. That’s actually sitting in chat GPT, and it’s a training tool, it’s a learning tool.

Cassandra Scott [00:58:56]:
They’re feeding stuff into it, learn from. So consider, you know, if you’re obviously concerned about accuracy, we’re talking about financial data in businesses. That’s critical. Think about the tools that, you know, have higher levels of accuracy because the data set is a contained data set, not a broad data set. At chatGpt, they call them hallucinations, and that’s a classic hallucination from chat GPT as well. The other thing that comes up, too is, you know, the ethics around using AI. Is it cheating? And, you know, if you’re talking about a uni student who’s completing their exam using AI, then, yeah, look, my ethics radar says that is cheating. But if you’re somebody in a business and you need to do, I don’t know, develop an Excel spreadsheet, and I’ve done this myself.

Cassandra Scott [00:59:46]:
And you know what, you want cell a and cell z to do with five other cells, but you don’t have the skills in Excel to develop that formula. Use chat GPT had some fabulous results out of their developing formulas just by putting prompts into chat GPT. So there are times when it is cheating, absolutely. But there are times where it’s about enhancing my productivity. I could have gone down a rabbit hole for 3 hours trying to find the perfect formula, and yet I was able to do that in five minutes, get what I needed to have done done, and then move on to the next thing as well. So it’s one of those things that I think everybody needs to step into with their own paradigms and expectations around it. One of the things I would like to share, if anybody’s interested in sort of reading more about AI, is I was very fortunate I can’t share it into the chat. David, you might be able to.

David Mitchell [01:00:45]:
We will send it to everyone with the recording.

Cassandra Scott [01:00:48]:
Yeah. There’s a guy in the US called Zac Cass, no relation whatsoever, who’s a futurist who steps very heavily into the AI space and his background is in particularly chat GPT working for them and OpenAI. His perspective on AI from a humanity perspective is actually really, really interesting. So if you’re wanting to dive deeper into this, I actually recommend having a look at some of the stuff he’s putting out. It’s very challenging thinking and often does answer some of the questions that we’re talking about here in terms of, you know, data security, impact on humanity, accuracy, you know, what, what can we do with it? And particularly with our kids and, you know, grandkids in my instance, coming into this world, you know, how do we work with that as, as well?

David Mitchell [01:01:37]:
Very good, Cass. We’re going to be respectful of everyone’s time. We have 30 seconds to go, so I’d just like to thank you and like to thank everyone for joining us here today. I hope everyone’s picked up one or two useful tips and we will see you at the next webinar.

Cassandra Scott [01:01:52]:
Thanks for having me, David, really appreciated.

David Mitchell [01:01:54]:
Bye.