WEBVTT - GenAI for SMPs (Part 2) 0:00:07.920 --> 0:00:11.940 Hello, my name is Gillian Bowen and this is Small Firm, 0:00:11.940 --> 0:00:12.900 Big Impact. 0:00:20.210 --> 0:00:23.239 If you can help every single person in your business, 0:00:23.239 --> 0:00:25.130 say half an hour a day or 45 minutes a 0:00:25.130 --> 0:00:27.530 day or an hour a day, it might seem like 0:00:27.530 --> 0:00:29.720 a small number, but when you scale that across your 0:00:29.720 --> 0:00:33.470 entire business or your entire team, that dramatically increases the 0:00:33.470 --> 0:00:37.670 efficiency of the organization. Treat your AI model or your 0:00:37.670 --> 0:00:41.240 tool like you would treat an intern or a grad. 0:00:41.240 --> 0:00:43.970 When I'm saying you can do a particular use case 0:00:43.970 --> 0:00:46.580 when you're explaining it to the model, explain it how 0:00:46.580 --> 0:00:48.530 you would explain it to an intern or grad, you're 0:00:48.530 --> 0:00:49.790 going to get a really good result. 0:00:55.570 --> 0:00:58.570 It's the podcast giving Chartered Accountants the up to date 0:00:58.570 --> 0:01:01.960 information they need to do their jobs. Each episode, I 0:01:01.960 --> 0:01:06.280 share resources, tools and expert advice provided by Chartered Accountants 0:01:06.280 --> 0:01:09.130 Australia and New Zealand and a range of people across 0:01:09.130 --> 0:01:12.880 our profession. So get following the podcast in your favourite 0:01:12.880 --> 0:01:21.010 podcast app. Let's start a conversation. Today. It's a big 0:01:21.010 --> 0:01:24.790 welcome back to Kayur Patel CA for part two of our 0:01:24.790 --> 0:01:29.140 discussion on generative AI. Now welcome back. First of all, 0:01:29.140 --> 0:01:30.640 Kayur thanks for coming back. 0:01:30.760 --> 0:01:32.170 It's good to be back. Thank you. 0:01:32.410 --> 0:01:36.910 Part two, as we promised, is focused on use cases 0:01:36.910 --> 0:01:39.970 for accountants. And why you might be asking. Well, all 0:01:39.970 --> 0:01:41.740 of that is included in part one. So go back 0:01:41.740 --> 0:01:45.670 and listen to that first. But in summary, GenAI and 0:01:45.670 --> 0:01:49.750 getting across it is all about efficiency and productivity and 0:01:49.750 --> 0:01:53.530 adding value to your individual work life, but also your 0:01:53.530 --> 0:01:57.190 business life and the life of your clients. And what 0:01:57.190 --> 0:02:00.670 you'll learn in this discussion is that starting small now 0:02:00.670 --> 0:02:04.000 is the key to bigger things down the track. So Kayur, 0:02:04.120 --> 0:02:07.030 just quickly, before we dive into the specifics of the 0:02:07.030 --> 0:02:09.790 act or, you know, action stations, as I like to 0:02:09.790 --> 0:02:13.330 call it, um, why do that? Why focus on small 0:02:13.330 --> 0:02:16.119 things now rather than waiting for the next big shiny 0:02:16.120 --> 0:02:17.169 AI tool? 0:02:17.810 --> 0:02:20.960 Yeah, it's a good question. And, you know, I've been 0:02:20.960 --> 0:02:27.019 helping clients, um, understand GenAI and then implement GenAI, a safe AI 0:02:27.020 --> 0:02:29.990 in their, in their environments for the last few months 0:02:29.990 --> 0:02:32.900 and by far and away the biggest piece when it 0:02:32.900 --> 0:02:38.330 comes to helping educate organizations around AI is how do 0:02:38.330 --> 0:02:41.030 I make sure that I get efficiency now? Because everyone's 0:02:41.030 --> 0:02:42.800 waiting for the big shiny thing, the thing that's going 0:02:42.800 --> 0:02:47.510 to say one silver bullet. Um, but what it's really easy, 0:02:47.690 --> 0:02:50.720 is really easy to miss is the fact that if 0:02:50.720 --> 0:02:53.720 you can help every single person in your business, say, 0:02:53.720 --> 0:02:55.490 half an hour a day or 45 minutes a day 0:02:55.490 --> 0:02:57.830 or an hour a day, it might seem like a 0:02:57.830 --> 0:03:00.320 small number, but when you scale that across your entire 0:03:00.320 --> 0:03:04.970 business or your entire team, that dramatically increases the efficiency 0:03:04.970 --> 0:03:07.940 of the organization. And that is possible right now with 0:03:07.940 --> 0:03:12.109 tools available right now, as long as you use a safe, 0:03:12.110 --> 0:03:16.489 enterprise grade, secure what I like to call safe AI option. 0:03:16.820 --> 0:03:18.590 Mhm mhm. Yeah that is a key part. And we'll 0:03:18.830 --> 0:03:21.649 explore that as well. Um and if we don't explore 0:03:21.650 --> 0:03:23.989 it we'll put information about that in the show notes. 0:03:24.110 --> 0:03:25.880 Um so what I was thinking and you and I 0:03:25.880 --> 0:03:27.590 have had a chat prior to this because, you know, 0:03:27.590 --> 0:03:30.890 planning is key to a quick and good discussion. We're 0:03:30.889 --> 0:03:34.970 going to cover three areas. So area one or part 0:03:34.970 --> 0:03:38.120 one is going to be text based use cases. Then 0:03:38.120 --> 0:03:41.300 we're going to have a look at data based use cases. 0:03:41.300 --> 0:03:43.430 And then we're going to wrap up by having a 0:03:43.430 --> 0:03:46.910 look at a couple of bespoke use cases. And as 0:03:46.910 --> 0:03:48.410 I said we're going to try and do that in 0:03:48.410 --> 0:03:51.800 20 minutes or less. So first of all Kayur, how can 0:03:51.800 --> 0:03:56.510 I use text based generative AI right now in my 0:03:56.510 --> 0:03:57.470 accounting firm? 0:03:58.140 --> 0:04:01.350 Yeah. Okay. So the first category is a text and 0:04:01.350 --> 0:04:03.450 then some data stuff as well. And for each of 0:04:03.450 --> 0:04:08.040 these categories, um, I might just reinforce one comment from 0:04:08.400 --> 0:04:13.200 the previous episode, which is remember that these AI models, 0:04:13.200 --> 0:04:16.200 these GenAI models, you have to treat them differently to 0:04:16.200 --> 0:04:19.260 like a search engine. Right? Because they're built they're built 0:04:19.260 --> 0:04:21.570 based on neural networks. And the example we used last 0:04:21.570 --> 0:04:24.450 week was, you know, when a baby is learning how 0:04:24.450 --> 0:04:28.620 to speak, they might look at their dad and say, um, dada, 0:04:28.620 --> 0:04:30.750 and then everyone will make a huge deal about it. 0:04:30.750 --> 0:04:34.230 And so that will get reinforced. That data is that 0:04:34.230 --> 0:04:37.950 person if they say, mama, um, nobody will make a 0:04:37.950 --> 0:04:39.600 big deal about it. And so then they'll know that 0:04:39.600 --> 0:04:44.580 that's not reinforced. And so that's how these GenAI models learn. 0:04:44.910 --> 0:04:47.969 And so as I'm going through these examples, uh, with 0:04:47.970 --> 0:04:51.210 your listeners, I would encourage everyone if they take nothing 0:04:51.210 --> 0:04:56.279 away from today's podcast, but this, um, treat your AI 0:04:56.279 --> 0:04:59.609 model or your tool like you would treat an intern 0:04:59.610 --> 0:05:02.220 or a grad, right? So when I'm saying you can 0:05:02.220 --> 0:05:05.580 do a particular use case, when you're explaining it to 0:05:05.580 --> 0:05:07.919 the model, explain it how you would explain it to 0:05:07.920 --> 0:05:09.570 an intern or grad, you're going to get a really 0:05:09.570 --> 0:05:12.450 good result rather than just typing it into a search engine. 0:05:12.450 --> 0:05:15.510 And equally, you know, if you gave a piece of 0:05:15.510 --> 0:05:17.850 work to an intern or a grad, you wouldn't expect 0:05:17.850 --> 0:05:19.440 that they would give you a piece of work that was, 0:05:19.740 --> 0:05:20.969 you don't have to look at it, you just send it 0:05:20.970 --> 0:05:23.219 straight out to the client. Um, you would want to 0:05:23.220 --> 0:05:26.550 review and check because that's how that's how the process 0:05:26.550 --> 0:05:29.370 goes to make sure that we tick the boxes and 0:05:29.370 --> 0:05:32.279 that we're providing good work for our clients. The same 0:05:32.279 --> 0:05:35.460 thing with these AI models. Treat it like an intern or grad. 0:05:35.490 --> 0:05:39.089 Such good tips. I love that such good tips. Okay, so. 0:05:39.089 --> 0:05:42.720 So with that in mind, pretty much any text based 0:05:42.720 --> 0:05:44.729 thing that you get an intern or a grade to do, 0:05:44.760 --> 0:05:47.520 you can get an AI model to do so. Here's 0:05:47.520 --> 0:05:50.160 what I is really good at from a text based thing. 0:05:50.160 --> 0:05:54.480 First category you can do is very good at creating documentation, 0:05:54.480 --> 0:05:56.700 creating text. Right. So I use it right now to 0:05:56.700 --> 0:06:01.650 draft articles for me to draft documents, to draft emails, communications. Um, 0:06:01.980 --> 0:06:05.310 I even get it to convert text that I write 0:06:05.310 --> 0:06:09.780 into my voice, like my actual voice and my video. Very, 0:06:09.779 --> 0:06:13.980 very good at doing those text based things, provided you're 0:06:13.980 --> 0:06:18.719 using specific prompts that, um, are engineered in a way, 0:06:18.720 --> 0:06:21.000 as if I was having a conversation with an intern. 0:06:21.000 --> 0:06:24.300 So I'm giving it the right background. I'm giving it, um, 0:06:24.300 --> 0:06:27.210 a conversation rather than just a one liner. Do this 0:06:27.210 --> 0:06:30.779 for me. I am providing it with other examples of 0:06:30.779 --> 0:06:33.089 documentation or email that I've done in the past, so 0:06:33.089 --> 0:06:34.950 it learns how to sound like me or how I 0:06:34.950 --> 0:06:37.980 want it to sound. Those are all things that I would, um, 0:06:37.980 --> 0:06:39.990 provide to an intern or a grad when I get 0:06:39.990 --> 0:06:44.160 it to write my emails or do text based documentation. 0:06:45.240 --> 0:06:47.910 So my brain's ticking away already. So just quickly to 0:06:47.910 --> 0:06:53.490 confirm, in my understanding say, for example, I paste into ChatGPT, 0:06:53.490 --> 0:06:57.540 for example, a version that is that I've paid a 0:06:57.540 --> 0:07:01.679 subscription for or that is just that that's legally protected. Um, 0:07:01.680 --> 0:07:05.279 I put in some emails that I write and then say, 0:07:05.279 --> 0:07:08.580 can you create an email that is written as if 0:07:08.580 --> 0:07:12.600 I wrote it to blah blah about said topic, and 0:07:12.600 --> 0:07:14.040 it would bring it up? 0:07:14.790 --> 0:07:17.429 Absolutely. And it would. If you've given it content like 0:07:17.430 --> 0:07:21.150 your previous emails, it's going to sound much more like you. 0:07:21.150 --> 0:07:23.580 And then if you were to provide it with more context, 0:07:23.580 --> 0:07:26.550 so explaining some background about who the recipient is and 0:07:26.550 --> 0:07:29.280 how much background knowledge they have, and whether you want 0:07:29.280 --> 0:07:32.040 the email to be formal or more casual or how 0:07:32.040 --> 0:07:34.290 detailed you want it to be, it's going to do 0:07:34.290 --> 0:07:37.020 even better. Just like it would be if if, um, 0:07:37.350 --> 0:07:39.060 if it was, if it was a human. 0:07:39.300 --> 0:07:41.490 And so what you would do is you would write, 0:07:41.490 --> 0:07:44.310 you're engaging with it, aren't you? You would say, okay, 0:07:44.310 --> 0:07:47.460 that's a good first attempt. But, um, if I give 0:07:47.460 --> 0:07:50.670 you this additional information, how would you rewrite that? And 0:07:50.670 --> 0:07:52.260 then it'll produce another version. 0:07:52.350 --> 0:07:54.720 It's very much an iterative process that you hit the 0:07:54.720 --> 0:07:57.450 nail on the head. It's very much an iterative process. 0:07:58.080 --> 0:07:59.370 Great. What else? 0:07:59.490 --> 0:08:03.840 So also in text. So um, outside of the creation bucket, 0:08:03.840 --> 0:08:08.250 it's very good at improving and transforming previously written text. 0:08:08.250 --> 0:08:12.030 So for example, you might have a piece of advice, um, 0:08:12.030 --> 0:08:13.530 to go out to a client. I think in the 0:08:13.530 --> 0:08:15.450 last episode we talked about fringe benefit tax. So maybe 0:08:15.450 --> 0:08:18.420 it's on how fringe benefit tax relates to their motor 0:08:18.420 --> 0:08:21.210 vehicle pool, for example. And you might think that the 0:08:21.210 --> 0:08:24.270 technical information is correct, but you want to craft it 0:08:24.270 --> 0:08:28.080 into a nice response. You can copy and paste in 0:08:28.620 --> 0:08:31.470 what you've written, some bullet points of the technical information, 0:08:31.470 --> 0:08:34.050 making sure you've got the right you know, the correct answer, 0:08:34.230 --> 0:08:37.050 and then get the model to be able to convert 0:08:37.050 --> 0:08:40.890 that into a nice email or notes to file or 0:08:40.890 --> 0:08:43.380 document or whatever it is that you need to send 0:08:43.380 --> 0:08:45.540 out the door to the client. I do this all 0:08:45.540 --> 0:08:48.660 the time, and it means now my emails to clients 0:08:48.660 --> 0:08:52.470 or whoever else it might be. I'm literally writing bullet points, 0:08:52.470 --> 0:08:54.510 and then I'm using AI to convert it into a 0:08:54.510 --> 0:08:57.240 nice email. And you do that for every email that 0:08:57.240 --> 0:09:00.620 you send that needs to be, you know, professionally worded, that 0:09:00.620 --> 0:09:03.650 starts to stack up into some serious time saving. Also 0:09:03.650 --> 0:09:06.410 very good at transforming text from a perspective of, you know, 0:09:06.410 --> 0:09:09.590 if you've got some generic text, get AI to convert 0:09:09.590 --> 0:09:12.079 that to a first person text. So it reads as 0:09:12.080 --> 0:09:16.100 if the recipient is it's talking to the recipient rather 0:09:16.100 --> 0:09:19.160 than talking about something um in general, it's very good at 0:09:19.160 --> 0:09:23.480 that and then also very good at taking some specific facts 0:09:23.480 --> 0:09:25.489 and then applying them to a template that you might 0:09:25.490 --> 0:09:27.559 have in your organization. If you've got a template that 0:09:27.559 --> 0:09:30.709 you use for a specific type of report for clients 0:09:30.710 --> 0:09:33.500 or whatever it might be very good at, then taking 0:09:33.500 --> 0:09:36.950 some specific details and putting them into your templates in 0:09:36.950 --> 0:09:38.480 the right way so that it reads well. 0:09:38.960 --> 0:09:41.600 All right. So are there any other text based use 0:09:41.600 --> 0:09:43.400 cases that you want to tell us about before we 0:09:43.400 --> 0:09:44.060 move on? 0:09:44.600 --> 0:09:48.140 Yeah. So the other category I think is summarization. So 0:09:48.140 --> 0:09:50.870 AI is very, very good at being able to summarize 0:09:50.870 --> 0:09:54.890 massive amounts of text or documentation into an easy to 0:09:54.890 --> 0:09:58.790 understand summary. So where do I use this? Anytime I'm 0:09:58.790 --> 0:10:02.300 going to talk to a client, um, I will get 0:10:02.300 --> 0:10:06.050 documentation that is relevant to them and summarize it down 0:10:06.050 --> 0:10:08.450 so that I understand it so I can be proactive 0:10:08.450 --> 0:10:10.670 when I talk to them. So that might be recent 0:10:10.670 --> 0:10:13.520 tax alerts from Inland Revenue. That might be latest industry 0:10:13.520 --> 0:10:16.640 information or documentation. That might be documents that I found 0:10:16.640 --> 0:10:19.610 from the client's own website. But I will use it 0:10:19.610 --> 0:10:21.829 every time before I go and have a conversation with 0:10:21.830 --> 0:10:25.429 a client. And that's just one example. But its ability 0:10:25.429 --> 0:10:29.569 to summarize large documentation into something that's easy to understand 0:10:29.570 --> 0:10:31.910 is brilliant. Everyone should be doing it. 0:10:32.120 --> 0:10:37.130 So it's teaching you, it's upskilling you before any sort 0:10:37.130 --> 0:10:39.500 of meetings or interactions that you're going to have with 0:10:39.500 --> 0:10:40.429 the client. 0:10:41.150 --> 0:10:43.220 Absolutely. And to the extent that I recommend to our 0:10:43.220 --> 0:10:47.120 own graduates and interns before they look at a new 0:10:47.120 --> 0:10:50.689 technical piece or information or topic or client industry for 0:10:50.690 --> 0:10:53.929 the first time, they should be doing that for the 0:10:53.929 --> 0:10:57.500 relevant documentation, for whatever that technical thing is, or the 0:10:57.500 --> 0:11:00.559 client or the industry every single time. It will just 0:11:00.559 --> 0:11:03.290 help them be a) more proactive, but b) just understand 0:11:03.290 --> 0:11:06.290 that base level of information to help get them up 0:11:06.290 --> 0:11:09.050 the curve quicker before they get in. And then do 0:11:09.050 --> 0:11:09.680 the work. 0:11:09.830 --> 0:11:13.250 How much time do you reckon that alone would save, um, 0:11:13.250 --> 0:11:16.550 a person or yourself? Instead of having to go and 0:11:16.550 --> 0:11:20.390 find everything, read it all and then write your own stuff. 0:11:21.290 --> 0:11:24.800 Oh, I mean, for me, on a weekly basis, I 0:11:24.800 --> 0:11:28.370 save hours and hours just using that one set of 0:11:28.370 --> 0:11:30.080 use cases, the summarization. 0:11:30.320 --> 0:11:30.920 Um, wow. 0:11:30.920 --> 0:11:34.520 You know, like a recent example, um, BEPs pillar two, 0:11:34.520 --> 0:11:37.910 the big documentation. I know that I didn't need to 0:11:37.910 --> 0:11:40.489 understand every single thing, but I needed to get a 0:11:40.490 --> 0:11:43.970 general idea of what it was about and when it came 0:11:43.970 --> 0:11:47.600 into force for various balance dates. That was something that 0:11:47.600 --> 0:11:49.190 I would have had to read through and kind of 0:11:49.190 --> 0:11:53.000 sift through and take quite a bit of time on. But, um, 30 seconds, 0:11:53.210 --> 0:11:54.589 genuinely 30 seconds. 0:11:55.429 --> 0:11:58.160 And is it safe? Are you informed enough? Like is it able 0:11:58.160 --> 0:12:00.770 to trust? And your argument is that you can trust 0:12:00.770 --> 0:12:03.500 what it spits out because you've put the source document 0:12:04.190 --> 0:12:05.360 in the AI. 0:12:06.309 --> 0:12:09.910 Absolutely. And and also I'm using this to give me 0:12:09.910 --> 0:12:12.880 that base level of knowledge and understanding. I'm not using 0:12:12.880 --> 0:12:16.060 it as the final authority on the topic. Important. And 0:12:16.059 --> 0:12:17.890 so when I'm using it to give me that base knowledge, 0:12:17.890 --> 0:12:21.280 I'm far more comfortable with being able to, um, to 0:12:21.280 --> 0:12:23.710 trust what it says. The other piece, though, is I 0:12:23.710 --> 0:12:26.440 do make sure that my prompts are specific. The way 0:12:26.440 --> 0:12:30.190 you prompt for these things will heavily influence the the 0:12:30.190 --> 0:12:30.939 stuff you get out. 0:12:31.390 --> 0:12:33.670 And that's the difference. That's the key note there between 0:12:33.670 --> 0:12:36.400 you and I and between our listeners and yourself. You 0:12:36.400 --> 0:12:39.370 live and breathe GenAI every single day. You lecture in it, 0:12:39.370 --> 0:12:41.380 you teach in it, you work in it. It is 0:12:41.380 --> 0:12:44.679 your job title. Um, that I would argue you are 0:12:44.679 --> 0:12:48.250 very qualified to understand what prompts one should put into 0:12:48.250 --> 0:12:50.740 the AI to get back the best result. So before 0:12:50.740 --> 0:12:53.500 we move on then to, um, data based, what are 0:12:53.500 --> 0:12:56.410 the risks here, um, with what you've proposed and how 0:12:56.410 --> 0:12:58.120 do you respond to those risks? 0:12:58.660 --> 0:13:02.740 Yeah. So look like everything there are risks. Um, and 0:13:02.740 --> 0:13:05.469 those risks are a) that the AI tool you're using 0:13:05.470 --> 0:13:07.300 and this would be for all of the examples that 0:13:07.300 --> 0:13:10.300 we give in this session, um, is not safe and secure. 0:13:10.300 --> 0:13:14.530 So absolutely advocate for nobody should be put in client 0:13:14.530 --> 0:13:18.700 or their own company information in a publicly available free 0:13:18.700 --> 0:13:21.850 AI tool. And also not even in a paid AI 0:13:21.880 --> 0:13:25.450 tool unless it is, um, what we call enterprise grade. 0:13:25.450 --> 0:13:29.590 So make sure that you have safe enterprise AI. Um, 0:13:29.650 --> 0:13:32.110 the second thing is you got to make sure that 0:13:32.110 --> 0:13:36.370 the information you get out is accurate and reads correctly 0:13:36.370 --> 0:13:40.360 and well. Now, I actually think accountants are best placed 0:13:40.360 --> 0:13:42.910 to make sure that that happens because we've already got 0:13:42.910 --> 0:13:45.610 the review structures in place. Right? So if a grad 0:13:45.610 --> 0:13:48.160 does a piece of work, normally they're not just going 0:13:48.160 --> 0:13:49.720 to send it out to the client without a review. Right? 0:13:49.720 --> 0:13:52.030 It'll get reviewed by the manager or the partner. We've 0:13:52.030 --> 0:13:54.760 already got a review structure. So keep the same review 0:13:54.760 --> 0:13:57.699 structure that you've got in place in place. It's just 0:13:57.700 --> 0:14:01.510 that the first cut is being assisted by these models. Um, 0:14:01.510 --> 0:14:03.910 and then the other piece is you can massively reduce 0:14:03.910 --> 0:14:09.430 the risk if you understand how to prompt well. And a 0:14:09.429 --> 0:14:11.590 big part of the training we provide to clients is 0:14:11.590 --> 0:14:16.000 helping them understand how to curate good prompts to get 0:14:16.000 --> 0:14:18.309 the right answer and the right tone of voice and 0:14:18.309 --> 0:14:19.630 all of those types of things. Yeah. 0:14:20.050 --> 0:14:22.840 That sounds like a key piece of learning for everyone 0:14:22.840 --> 0:14:27.640 that's listening. How to prompt well, how to engage effectively 0:14:27.640 --> 0:14:30.460 with the AI to make it produce the best result. 0:14:30.460 --> 0:14:32.890 We'll see if we can find some suitable reading or 0:14:32.890 --> 0:14:36.850 information for that. Or heck, we may even do an 0:14:36.850 --> 0:14:39.550 episode literally just on that. But a little bit further 0:14:39.550 --> 0:14:41.290 down the track. There's so much to talk about in 0:14:41.290 --> 0:14:44.110 this space, and I know that our members really, really, 0:14:44.110 --> 0:14:47.530 really want the information. Let's move on to data. Um, 0:14:47.530 --> 0:14:50.770 it's key to the life of an accountant. Um, what 0:14:50.770 --> 0:14:53.380 can a small or medium sized practice use right now 0:14:53.380 --> 0:14:57.100 to be more efficient, uh, more productive to add value? 0:14:57.550 --> 0:15:01.120 Yeah. So there's probably a few examples that, um, that 0:15:01.120 --> 0:15:03.790 you can jump into straight away. So if you use 0:15:03.790 --> 0:15:07.870 Excel and I'm assuming that, that everyone listening to this does, um, 0:15:07.870 --> 0:15:10.930 you can get these large language models to do two 0:15:10.930 --> 0:15:13.300 things with Excel. One, you can get them to help 0:15:13.300 --> 0:15:16.630 you either create or troubleshoot your formulas. So if you've 0:15:16.630 --> 0:15:19.840 got a massive workbook for forecasting or budgeting and you're 0:15:19.840 --> 0:15:23.050 trying to get you're trying to figure out why something's breaking, 0:15:23.050 --> 0:15:26.320 it's very good at helping you to either troubleshoot or 0:15:26.320 --> 0:15:28.960 create the right formulas for you. So that's that's probably 0:15:28.960 --> 0:15:31.540 one use case. The second thing that some of these 0:15:31.540 --> 0:15:35.170 models can do, though, is you can upload Excel files 0:15:35.170 --> 0:15:37.810 to them and just say in its simplest form, just say, 0:15:37.810 --> 0:15:41.740 I've got, um, sales data for the first seven months 0:15:41.740 --> 0:15:45.220 of the year. Um, in an Excel file, you could 0:15:45.220 --> 0:15:49.210 then get the model to create columns for each month 0:15:49.210 --> 0:15:52.870 for the rest of the year, and then predict out, um, 0:15:52.870 --> 0:15:56.410 sales volumes for each month for the rest of the year, um, 0:15:56.410 --> 0:15:58.570 based on some parameters. So you could say a 10% 0:15:58.570 --> 0:16:01.630 uplift in this or a change in this variable or 0:16:01.630 --> 0:16:03.010 whatever it might be, and it will then go and 0:16:03.010 --> 0:16:06.160 create the file, sorry, create the extra columns, create the 0:16:06.160 --> 0:16:08.590 data in those columns and allow you to download that 0:16:08.590 --> 0:16:12.190 file as well. And, and some of these newer tools 0:16:12.190 --> 0:16:15.880 like Microsoft Copilot, um, you won't even need to upload 0:16:15.880 --> 0:16:18.190 them and it'll just it just already is able to 0:16:18.190 --> 0:16:20.890 access your Excel files and just, um, and provide you 0:16:20.890 --> 0:16:24.280 with those changes on the fly as well. Um. 0:16:25.000 --> 0:16:28.330 How many accountants do you think are across the use 0:16:28.330 --> 0:16:31.060 or the availability of of that to them? 0:16:32.020 --> 0:16:37.750 Um. To be honest, not many. So. Probably very few. Um, 0:16:37.750 --> 0:16:41.800 and look, that's not necessarily a bad thing, because the 0:16:41.800 --> 0:16:44.710 ability of these tools to be able to work well 0:16:44.710 --> 0:16:48.790 with Excel is a very recent thing. So if we were, 0:16:48.820 --> 0:16:52.720 if we were recording this 2 or 3 months ago, um, 0:16:53.200 --> 0:16:55.810 actually even more recent than that, that's not something I 0:16:55.810 --> 0:16:58.060 would be recommending people play with, because I just didn't 0:16:58.060 --> 0:17:00.430 think it was good enough at that stage. And it's 0:17:00.430 --> 0:17:03.220 still early days, don't get me wrong, but it's now 0:17:03.220 --> 0:17:04.959 got good enough for me to think, well, this is 0:17:04.960 --> 0:17:06.070 interesting and 0:17:06.609 --> 0:17:09.250 I should get it trying so that yes, yes. 0:17:09.250 --> 0:17:10.450 I'm improving with it. 0:17:10.570 --> 0:17:12.639 I just feel like this is good. I mean, I 0:17:12.640 --> 0:17:15.010 know that a podcast is hard to be, you know, 0:17:15.010 --> 0:17:17.860 breaking news, but I feel like that is a breaking 0:17:17.859 --> 0:17:21.910 news development that you can use AI that way and 0:17:21.910 --> 0:17:24.790 excel that way together. Um, and I just think, as 0:17:24.790 --> 0:17:27.220 you said at the top of the show, um, saving 0:17:27.220 --> 0:17:30.429 30 minutes a day scaling that across your business, that 0:17:30.430 --> 0:17:35.260 to me sounds like something that is, um, day changing, 0:17:35.260 --> 0:17:36.879 week changing, month changing. 0:17:37.390 --> 0:17:40.720 Yeah. And that's before you even get into, you know, 0:17:40.720 --> 0:17:45.370 like AWS, they have a data visualization app. If you're 0:17:45.369 --> 0:17:49.570 using that already, you can now use an AI module 0:17:49.570 --> 0:17:52.930 on top of that to be able to natural language, 0:17:52.930 --> 0:17:55.000 ask it to do things for you. So create me 0:17:55.000 --> 0:17:57.879 a graph that compares sales data for the last three 0:17:57.880 --> 0:18:00.850 months with accounts receivable and cash or whatever else it 0:18:00.850 --> 0:18:05.920 might be, and that's also extremely new. Um, and, and 0:18:05.920 --> 0:18:08.859 those are like kind of bespoke data visualization tools. Power 0:18:08.859 --> 0:18:12.970 BI is going to have that very shortly as well. So, um, 0:18:13.060 --> 0:18:16.000 the examples I've talked about is are literally using the 0:18:16.000 --> 0:18:19.929 same tools we've already got, but um, with or alongside 0:18:19.930 --> 0:18:23.170 an AI model. Um, that's before you even get to 0:18:23.170 --> 0:18:27.969 these like specifically built data visualization and manipulation tools that 0:18:27.970 --> 0:18:29.199 have got AI embedded. 0:18:29.710 --> 0:18:31.630 So we're talking about data. 0:18:31.630 --> 0:18:33.580 So there's two things there for data isn't there. There's 0:18:33.580 --> 0:18:36.820 a there's programs that have an AI data tool built 0:18:36.820 --> 0:18:39.639 on top of them, like you'd said with AWS, but 0:18:39.640 --> 0:18:42.250 in in the sense of Excel, you are putting the 0:18:42.250 --> 0:18:47.320 Excel into your text based AI, am I right? 0:18:47.890 --> 0:18:50.859 Yeah. This so well there's two options. So if you um. 0:18:51.609 --> 0:18:54.190 Just that I clarify that that's what we're talking about. Right. 0:18:54.190 --> 0:18:55.810 But you go ahead first. Yes. 0:18:55.810 --> 0:18:57.760 Well for most of the models that's exactly right. You'll 0:18:57.760 --> 0:19:00.129 upload an Excel file, um, and it will read the 0:19:00.130 --> 0:19:02.740 file and understand it. And then you'd ask it to, um, 0:19:02.740 --> 0:19:05.859 you know, in that last example, extrapolate out sales based 0:19:05.859 --> 0:19:08.619 on some variables and, um, update the Excel file so 0:19:08.619 --> 0:19:14.020 you can download it. Um, but Microsoft Copilot, as that improves, um, 0:19:14.020 --> 0:19:16.360 that will get better and better at doing that natively 0:19:16.359 --> 0:19:19.690 within Excel as well. Um, so there's there's a few 0:19:19.690 --> 0:19:23.350 different options depending on which route of safe AI you 0:19:23.350 --> 0:19:25.840 want to go down, which models you decide to choose. 0:19:26.700 --> 0:19:28.530 Yep, that makes sense. I just wanted to make it 0:19:28.530 --> 0:19:31.530 clear that we weren't talking about a data based AI. 0:19:31.560 --> 0:19:34.290 We were just talking about how to use AI with 0:19:34.290 --> 0:19:37.260 your data, and as a result, you are using the 0:19:37.260 --> 0:19:40.350 text based AI to manipulate your data. 0:19:40.500 --> 0:19:43.530 Yes, absolutely. And the great thing about that is it's 0:19:43.530 --> 0:19:44.850 available right now, right? 0:19:45.450 --> 0:19:49.379 Yes. Absolutely. Absolutely. Okay. All right. So we're almost out 0:19:49.380 --> 0:19:51.689 of time. I want to have a quick look at, um, 0:19:51.690 --> 0:19:54.330 1 or 2 or it might just be one bespoke 0:19:54.330 --> 0:19:58.770 option that's relevant to a particular practice, an example of 0:19:58.770 --> 0:20:00.210 what that might look like. 0:20:01.140 --> 0:20:03.180 Yeah. So this is when you start to get into 0:20:03.180 --> 0:20:09.419 a little bit more, um, I guess, um, specific models 0:20:09.420 --> 0:20:11.970 that have been altered in a way to help with 0:20:11.970 --> 0:20:15.030 a specific problem. And this is good if you've got 0:20:15.030 --> 0:20:17.730 repeatable tasks that you do in your practice all the time. 0:20:17.730 --> 0:20:21.750 So one pain point that, um, I'm assuming almost every 0:20:21.750 --> 0:20:24.420 accounting practice has is onboarding clients, right? Like we just 0:20:24.420 --> 0:20:26.040 want to get in and start doing the work. You've 0:20:26.040 --> 0:20:27.330 got to get the right information. You got to do 0:20:27.359 --> 0:20:30.330 AML checks, you got to do a whole bunch of stuff. Um, 0:20:30.330 --> 0:20:34.409 and so you could, um, build using a set of, uh, 0:20:34.410 --> 0:20:39.870 using an algorithm and, um, and a specific piece of, um, 0:20:39.869 --> 0:20:45.000 publicly available safe AI to build you an AI based 0:20:45.000 --> 0:20:48.360 automated workflow to, for example, onboard clients. So a client, 0:20:48.359 --> 0:20:51.240 you send them a link and it asks them for 0:20:51.240 --> 0:20:54.090