The Future of ATO Data + AI: Securely Connecting with ChatGPT, Copilot and Claude

Trent McLaren: Hello, ladies and gentlemen. Welcome to this morning's session of the AI in Practice Summit. How are you all going? Have you had a fantastic time so far? Strap yourselves in. This has been the most requested session on the entire summit, so I'm very excited for you to get stuck in.

Trent McLaren: On to the main event. I mentioned that this was the most requested session, the most registrations. More than a thousand people registered for this talk. I am joined by Stan Corner, who is the founder and CEO at Business Automation Works. They have built incredible products like ATO Mate and now Omble, Omble being the product that you will see front and centre today. This session is The Future of ATO Data + AI: Securely Connecting with ChatGPT, Copilot and Claude. Look forward to getting your feedback. Stan, I'm gonna hand over to you. Thank you for your participation, and I'll let you take it away. Thank you so much.

Stan Corner: Hey Trent, thank you for a warm welcome, and thank you for having us here today on the AI Summit. Super excited to share our vision of the future, and looking forward to jumping through some live demos showing how AI can securely connect to the ATO data. So I think without any further ado, let's get on with it.

Stan Corner: So let's start with my view of AI. If you'd asked me this time last year, I'd have said something like, AI is definitely part of our future. It's something we need to start thinking about, maybe start investigating. The world changed really rapidly early this year with the release of all the new models from our large language partners: Claude, ChatGPT and of course Copilot. So while not that long ago I was suggesting AI was part of all of our future, what I'm saying today is it's here. It is part of everyday life as we see it today. How do we relate that to what we all do and love, which is accounting and working with clients and the ATO? I'm gonna share a lot of that today.

Stan Corner: On the call, we will have many of you. I think there's pretty close to a thousand participants today, which is super exciting. Some of you will be just dipping the toe in the water, some of you will be superheroes or super champions using AI. I sit somewhere in the middle of that. So let's jump in and talk about how we got to where we're at today. So I am the CEO and founder, a very fancy title for a pretty simple bloke, of Business Automation Works. Business Automation Works was founded some six or seven years ago. It started with our flagship product called ATO Mate, but let's go back a little bit from there. Business Automation Works: something I'm super passionate about is solving problems, and specifically solving problems in the

Stan Corner: tax and accounting space. Wholly Australian owned, and we now power two products, ATO Mate and Omble. Let's start with ATO Mate. So many of you will know and love ATO Mate. More than 2,000 practices trust ATO Mate today to process ATO documents and communications for their practice and their clients. Where did ATO Mate come from? Well, it's super interesting. It started with a conversation with a senior partner in a large accounting practice. And it was a casual conversation around time, efficiency, workload, repetitive tasks, and all sorts of things that today might be seen as being pretty much normal to attack. So we built ATO Mate about five years ago. As I've suggested, it's trusted by a couple of thousand practices.

Stan Corner: Now, why is this important in today's discussion? Well, it's how we got to where we're at today, talking about AI. So ATO Mate is an automation tool, using robotic process automation and some AI, to actually reduce the time it takes for you and your team to do things like capture the documents from the ATO, do validation, cross-checking, build communication, send it to a client and then file. Super importantly, ATO Mate drives highly repetitive tasks via automation. So, wildly successful, used by a couple of thousand practices. So I started by sharing with you that what I'm super passionate about is solving problems.

Stan Corner: After we launched ATO Mate, and many were using it, we started getting a bunch more questions around some challenges. The challenge that we kept hearing is that logging into the ATO portal is a pain. You have to multi-factor authenticate every time you log in. It throws you out when you're not active in the portal for more than 20 minutes. So a bit of research tells us that on average the accounting professional, across all

Stan Corner: roles, so admin, junior accountant, accountant, senior accountant, managers and partners, is logging in an average of 5.5 times a day. When we did some further digging, what we also discovered is that not only did you have to log in, but once you logged in, while there's loads and loads of information that's available to you, it was slow and repetitive to get this information. So you couldn't get a whole view of an entity or a whole view of a family group. You couldn't look at clients in the ATO portal based upon the relationship manager or the partner that is responsible.

Stan Corner: So out of these problems came Omble, which we launched earlier this year. It is effectively a set of dashboards, folks, that brings all of the data from your practice management system, so XPM, APS, Access Cloud, Reckon, HandiSoft, LOGiT, CloudConnect, Xero and MYOB. If I've missed someone, I apologise. So it brings the data via our integration into Omble. The sorts of things we're bringing in are the family group, the relationship manager, the managing partner, type of entity, a bunch of other things. We then, via our direct integration with the ATO, bring all the data that hides in behind that login out into Omble from the ATO. So things like lodgements, lodgement dates, transactional history, account balances, payment plans, and a whole lot more.

Stan Corner: So we built Omble with this view to improving the visibility and accessibility of that data to an accounting practice. Now we've got about 150 practices using Omble today. What they're reporting is it reduces the login to the ATO portal super significantly. It also means that there's now accessibility to information that either you didn't have time to get or you didn't know that you needed. And the result of this is far more proactivity in managing client affairs, but also better client communication.

Stan Corner: So this has all been about setting the scene to get to here, which is the topic we're here to talk about today. The big question on everybody's lips is: can you securely use your AI tools, Claude, Copilot, ChatGPT, et cetera, to analyse your clients' ATO obligations? Now, frequently I hear the answer to this as being no, it's not safe. No, it's not secure. Certainly, the way you could do this today is you could log into the ATO portal, you could download all of this information, build it into a bunch of reports, and drop that into Claude or ChatGPT or Copilot. But that's pretty tedious, and you also then have to know what you're looking for. So until recently, I think we'd safely all suggest the answer to this big hairy question is a simple no. Is it safe and secure?

Stan Corner: So we at BAW had already built Omble. The question is, can we securely connect your large language model AI tools into Omble and allow you to access that? So the simple answer, folks, is yes. It's so exciting. I'm pretty confident in saying this is a unique opportunity, it's a unique tool, and BAW are the first to make this available to you. So how does it work? Let's talk a bit about this, because security and privacy is critical to you, the practice, your clients, and very importantly, it's critical to Business Automation Works.

Stan Corner: So let's talk about this. I've already said to you, what we do with Omble is we bring all the data out of the ATO portal. Now, the data you see in Omble is the same data you'll see if you log into the portal. So it's live, real time. We also bring your practice management data out into Omble. So Omble sits there holding all of this data. Now, it sits in the BAW platform. It is secure. It is all held in Australia, but it gives you this capability of accessing this data, either via an interface in Omble or, what I'm about to show you, via these large language models. We've built this secure connector in between, called an MCP connector or MCP server. Many of you will be familiar with this terminology.

Stan Corner: But in very simple terms, what this does is it controls the connectivity. So it controls the accessibility. So as a practice, you'll need to authorise your users in Omble to be able to connect to the MCP server. So ultimately, you are in control. The MCP connector then controls your AI tool to make sure that it can't go rogue or start giving you delusional thoughts. That's not the right terminology, but I think everyone gets where I'm going here. So it's really about building this structure to ensure that you can get to this data.

Stan Corner: Now, super importantly, what Omble gives you is structured data. Using AI tools, if it's really super unstructured, you're gonna get a lot of variability. Omble builds this data into a very structured set of connectors, or API-to-API connectors, via this MCP. That then allows you, folks, to choose whichever AI tool you use today. Now, I'm not going to give you advice on how you should be managing this, but we've built the infrastructure allowing you to access this. As a practice, you need to implement the governance that's required. So when I talk about governance, the first couple of things: many of you have already chosen your tool of choice for the practice. Might be Copilot, ChatGPT or Claude, it might be a mix of all of these. So you get to make the decision on which tool you want to use.

Stan Corner: My strong advice is make sure that you've made the decision, implemented the governance, rolled it out practice-wide to all your users, and make sure it's a paid version of these tools. Why is the paid version so important? Pretty simply, the paid version guarantees you that the data that you use in these tools will not be shared outside your subscription of the tool, and it cannot be used for training. So super, super important. Another red-hot tip that I'm hearing from many practices is they're rolling out, say, Claude, and then they are blocking the free versions of these tools. What it means is it takes out the human error of allowing one of your team members, with the best intention, to go out and use secure data. Now, only the tool that you authorise in your practice will be able to access this connector. All right, little sip of water.

Stan Corner: Let's move forward and talk about why this is possible. So Business Automation Works is the team behind Omble. We are partnered up and integrated with the ATO. Both Omble and ATO Mate are approved solutions with the ATO. What does this mean? Well, the ATO understands what we're doing. They audit us, well, you could probably suggest to a painful degree, every year, to ensure that we're meeting all of the security obligations that are outlined by both the ATO and TPB and other privacy laws. Business Automation Works is also ISO 27001 certified. This is the highest certification for security in the land. It is again a process of being audited on what we do and what we say, what we write and how we implement our security. It gives you ultimate peace of mind. Adding all of this together, it is a process of ongoing due diligence and audits, to ensure that in fact, when we say safe and secure, there is some backing behind this.

Stan Corner: So let's get to the topic at hand. What are the opportunities that having AI connected to ATO data mean for you, your team, your practice, and very importantly, your clients? So I'm going to cover a few areas, and I'm going to give some real live examples in a little bit. We're going to jump over into Claude, which is my tool of choice, and we're going to run some prompts. Understanding your practice data: Omble holds, and so does the ATO, this massive information that is available to you. The question is, how do you know what you want to know when you want to know it? And then how do you get that data out? It also allows you to very quickly and succinctly uncover what actually matters. We're going to do a couple of prompts that will show you the power here.

Stan Corner: I've heard this multiple times: one of the things AI will do for you is it will give you back time. So what may be taking hours or days can now be completed in minutes. And you're going to see that when we go to the live demo. One of the examples I'm going to give today, and I know Trent, who introduced me today, is passionate about this, is preparation for meetings. The power of what you can do with AI in preparing for a client meeting is quite astounding when we get there. So let's jump in and have a look.

Stan Corner: So the first thing we're going to do is we're going to show you how you securely connect Omble and your tool of choice. Now, today I am going to be using Claude. So what we're going to do to start with is we're going to show you connecting to Claude securely. Really important part of the process. To be able to connect to Omble, the user, so me in this case, must be an authorised and logged-in user in Omble. So what I'm gonna do now is I'm gonna do a connection, and you're gonna see that happening live.

Stan Corner: So Claude is now open. So I'm gonna do this via a voice prompt. "Connect to Claude." Try that again. "Connect to Omble."

Stan Corner: So what's happening here is I've done a voice prompt. I've said, let's connect to Omble. What you're going to see here is that Claude goes out, finds the options. Now, very importantly, folks, this isn't just searchable. This is set up at a practice level. I'm an authorised user, and Omble is authorised as a connector within Claude. So we simply hit connect. What you're going to see here is that it's asking me to log in to ATO Mate and Omble, which is an important part, and then I'm going to be forced to multi-factor authenticate. So again, it's all around finding the right connection.

Stan Corner: It's all about ensuring that there's a secure connection and everyone connecting is authorised. So it's now going, would you like to connect? And I'm going to accept that. Now again, I'm going to repeat: this is not possible without authorisation. So I've got two connections, one to my dev and one to my production environment. So I'm going to select the one I want to use. And Claude's now going off, it's doing the connection. And the first thing you're going to see is it's going to confirm this connection by giving me a little bit of information. So my Omble practice is a copy of real practice data that has been masked. Now, super importantly, it's masked, so you can't identify who the client is or who the practice is. This practice contains 1,550 clients, and it covers, as suggested on screen, lodgements, obligations, transactional history, et cetera.

Stan Corner: Now, I will put a little caveat in here. When you look at the data I'm gonna start producing: we copy this, we mask it. It's not up to date as of today. So please ignore the dates. Okay, so what I'm gonna do now is I'm going to start doing some prompting. I'll just need to find my mouse to get it down here. So I've got some prompts I've prepared earlier, and I copy and paste these in. I'm gonna do a bit of a prompt that is an overall health check of the practice. Now remember, this is going and looking at the data. So what I've asked for is: focus on outstanding items for my managing partners, relationship managers, list how many entities, family groups, et cetera. Give me a total amount of debt, payment plan information. I just want an overview of my staff workload and our clients' overall ATO health. So let's click go on that.

Stan Corner: So what we can see is Omble is now going off and doing some checking. Sorry,

Stan Corner: Claude is going off and doing some checking. It's going and looking at the specific data that I've asked for, but it's using the rails and the API connections via that MCP connector to ensure that, A, we're searching the right data, it's in a controlled environment, and most importantly, it is secure.

Stan Corner: Okay, so from this search, you can see Claude is telling us what it's going to do, but here is an overall balance-of-health check. So total number of clients, number of accounts, interestingly, the total debt as at today for your clients with the ATO, overdue lodgements, interesting information. So this took just a few minutes. Let's imagine you had to do this manually. This would take hours, if not days, to collate. And we've been able to achieve that in just a couple of minutes. So scrolling through, we've got the managing partners, a breakdown of the number of clients assigned, et cetera, down to relationship manager, giving us some overdue lodgement details, total number of clients. So all very useful information, some of which you probably know, but definitely wouldn't have to hand in a couple of minutes of prompting.

Stan Corner: So as you can see, Claude's given us a detailed list. It's also given us some details of some information that may be critical at this point in time. For instance, which entities have got director penalty accounts. So you may or may not know, but the ability to search on this. So, quite a detailed prompt. So I think everyone can agree that this information is readily available to you, but to get it in this sort of format... So let's move on to the next prompt, which is a little less detailed, but just as powerful. So this is about looking at lodgements just for individuals that are due over the coming year, up till the 15th of May. Again, I've popped a prompt in and I've set Claude off to do its work. So Claude, again, is connecting securely to Omble, which holds all of the data for the ATO obligations, et cetera, and your practice management data.

Stan Corner: Again, Claude is going to go through using the request that I've given, in this safe and secure environment, to produce this outcome.

Stan Corner: We'll let that run, and you'll see this over the next minute or two. Bring this up and produce it. And as you can see, there's a question, and the question goes: the latest due date, or before that date?

Stan Corner: I've answered Omble's question, because it wasn't quite sure, so my prompt may not have been clear enough. And Claude has now set off to capture that data and start presenting it. So as we can see, Claude's gone off to Omble, gathered all of the data, and then created this visual chart for us. So as you can see, these ten thousand lodgements, not yet lodged, et cetera. So it gives us all of the data in a very easy-to-read view. Now, it's also showing us that there's a whole bunch of returns to be lodged that are not assigned to a partner or relationship manager in the practice management system, therefore not in Omble. So again, this is information you could put together, but it would take hours of someone's time to create. So today we've been focusing on the practice-type data. So I'm going to go and pop this particular prompt in.

Stan Corner: This one is really about the health of your clients. So I'm going to let this start. So it is: review all accounts with balances greater than $100,000 and list entities that have rising debt balances, general interest charges and/or penalties. Also show any overdue lodgements that these entities have. So I'm going to set that working and talk about my thinking about why I want to create this. So if we think about all of the activities from the ATO, and rising debt, and the ATO taking much more aggressive or firm action to collect that debt, one of the secrets here is that certainly once a DPN is issued, there's action to be taken.

Stan Corner: Accessibility of this data, and this is why I've created this particular prompt, is understanding which of my clients, or which of your clients, are at risk of that ATO action. So some of the key indicators or personas of a client that may need some assistance might be the rising debt. They've got penalties, general interest charges, and very importantly, for the purpose of looking at who might get a director penalty notice, critical: if they're not lodging their activity statements, for instance, with overdue debt, the ATO is very likely to issue a binding director penalty notice, meaning that the director is liable for all those company debts. So, really interesting. When you see this come out, I think you'll be surprised at the level of information that is available.

Stan Corner: Okay, so you can see that the results are starting to appear on screen. There's some basic findings. One thing you may not have noticed, but intentionally I listed general interest charges as GDI. Claude was smart enough to understand that what I actually meant was GIC. So the beauty of these AI tools is you don't have to be precise. So let's run through a bit of the data. It lists out, I think it's 19 entities that have got growing debt that also have outstanding balances of more than $100,000. They've got penalties, and it also went off and showed us, are there payment plans in place? So really important information. Again, information you can get to, or you may know. We've then asked it to show overdue lodgements. This is pretty important to the action that the ATO might undertake, et cetera. So loads and loads of information. Now, I'm gonna go one step further.

Stan Corner: And I'm gonna ask Claude to put this into a report ready for sharing with the partners. So one of the things that is truly amazing about these AI tools: it takes someone like me, who is not particularly proficient in some of these tools, PowerPoint for example, et cetera, and it actually makes me into a bit of a... well, I'm not gonna go expert, but it does allow people in your team that may not be experts in a particular arena to go off and actually start doing some of this work. So you can see this happening. It's titled it, it's put it into a format that is truly readable and usable.

Stan Corner: And on the journey goes. So again, I'm going to reiterate: the power of connecting your AI tool into the ATO data is super, super significant and can have a very significant impact. All right. So I'm going to go on to our last prompt of the day, and this last prompt is something that I know is near and dear to all our hearts, and it's preparing for our client engagement.

Stan Corner: So I'll get this prompt in and then I'll give you context. So let's pop our prompt in. So this is in preparation for a meeting with the head of a family group. So it's the Harris and Olson group. I've told Claude I have a meeting with them today. I need to know the following: all entities, all account balances, all money upcoming, any overdue lodgements, all upcoming lodgements due this financial year, and any other relevant information. So I've asked a really open question into the whole group. Results to be returned in a PowerPoint ready for the client meeting. So let's press go on that.

Stan Corner: So I think, again, this is probably a task that's undertaken maybe by yourself as the partner or the manager, but frequently the gathering of this information will be completed by someone else, logging into the portal, collating this into spreadsheets, or then into PowerPoint presentations. So again, while this takes a few minutes to run, it is significantly faster,

Stan Corner: with much more detail than you are likely to prepare in readiness for a client meeting. So while this is running, one of the things that I'm often asked is, Stan, what do you think AI is going to do to this industry? So while there's many, many things getting around, is it going to reduce people? No. AI still needs the human impact, cross-checking the data, putting our humanistic, our ability to cognitively understand things, in place. But the most important thing I think AI and automation has done, and will continue to do, is it's a bit of a back to the future moment. What do I mean by that? I think if we wind the clock back a number of years, this industry was all about client engagement, spending time with the clients, becoming their trusted advisor.

Stan Corner: I think over time, with the compliance and the amount of work we've needed to do, we've turned that into a lot of closed-door work and less working with clients. So back to the future is: these tools, and what we're showing you today, have the potential, or the ability, to get you back in front of the people and do that all-important thing, which is have that relationship with the other human, in this discussion being your client. So let this run. I'm pretty amazed by the results. I think you will be too. So I'll let that run and then we'll come back and do a quick wrap-up.

Stan Corner: Okay, and here is that presentation. So you can see on the left-hand side of the screen all of the information that's been used. But let's open the PowerPoint. Download that. So this is now in PowerPoint, ready for a discussion with the client. So let's run through that. So at a glance, 19 entities. The formatting might need a little bit of work. Lodgements, lodgements overdue, listing all of the entities by type of entity, super helpful. Amounts owing, activity statement and account balances, lodgements that are overdue, lodgements that are due before the 31st of March, and then other points. For instance, one of the accounts has got a credit sitting on the income tax account. So this might be an action that you didn't know about, and you can go and get those funds returned. And next agreed steps. So, formatted ready to go for your client meeting.

Stan Corner: Now, that took a few minutes, but how much time would that have taken you or your team to create and put together? More importantly, would you have actually gone to this level of detail? I hear in many cases the answer is simply no. So the empowerment of AI connected to ATO data via Omble gives you the capability to do things that you've not been doing, enhancing that client relationship. So hopefully what I've been able to do with some live demo here is give you a really good taste of what is possible, which is what we're looking to do. So the areas we covered: understanding your workloads, the internal view; uncovering what might need attention, so clients that have got rising debt, outstanding balances; and preparation for a client meeting. These are just three of the many, many opportunities that this new tool presents.

Stan Corner: We started with the question, can you securely and safely use your AI tools, Claude, Copilot and ChatGPT, on ATO data? Well, the answer, hopefully, as you've just discovered using Omble and our MCP connector, is simply yes. And this is the outcome. The challenge before today: many hours logging in, searching, downloading, reviewing, collating, and creating outcomes. With Omble and AI, it now turns to minutes. Super exciting what this will present as an opportunity to you, your team, practice, and industry as a whole. So what's next? Well, really simply, let's do a quick recap. Omble as a platform is available to you today. You can search, filter, view all of the data we've just talked about. We will release the MCP server over the next week to ten days to our existing users. So if you're here today and you're already an Omble user, sit tight; you'll get access and some training.

Stan Corner: So Trent and the team at The Firm, thank you for having us join the AI Summit. As you can tell, I'm pretty excited about where we're at today. I'm even more excited about the future, and I hope I've shared enough today to ensure that our participants are as excited and optimistic as I am about this wonderful industry that we participate in. Thanks again, and have a brilliant day.

Trent McLaren: Thank you, Stan. That was awesome. Well done. Thank you for participating and showing us everything about Omble and how to connect it using the MCP with Claude and ChatGPT. Thank you for joining this session. Thank you, Stan, and we'll see you later. Bye bye.