What the AI adoption numbers actually mean for your practice

[00:00:00] Hello and welcome to another episode of Digi-Tools In Accrual World, a podcast watched by the Digital Disruptors. Today I'm joined by Kevin Fitzgerald and Ryan Pearcy and we had a really interesting discussion, some key points just before I throw it to the guys that we cover that you might want to stick around for. So Bright launches AI payroll assistant, which is Oscar, and we cover up some of the merits of that and also their AI rollout plan.

[00:00:24] We cover Mimo and how it's extended into Associate and we had a bit of a discussion about their history and how they've then managed to evolve upstream a little bit more to cover other functions. And we cover things from Free Agent and from Sweet Files and Trove, which are, Trove is slightly newer in the space as well as Ojo's. Ryan, tell me, how are you going and tell me what you were most excited about covering this week?

[00:00:50] Well, I love covering all the news. I love the banter we have and excited to kind of dive into all of that. But I mean, I'm keen to see what Zoho are going to do because they're trying to just penetrate the accounting space now and we'll talk a little bit about more detail on that. But, you know, I love the fact that Kevin's here because he can just give us an insight from the payroll aspect. So thanks, Kevin, for joining us again. Always happy to be here and give that type of perspective.

[00:01:16] I think the biggest one for me, as you may expect, is the AI skill gap in accounting. I look at the industry all the time and I wanted to keep, you know, up with the times is probably the way that I'd describe it right now. It's great to have you on, Kevin, because like we said, you've got a bit of a perspective, double perspective. You've got the accountant perspective, you've got the payroll perspective. And also just your warm, lovely presence is much more comforting than me.

[00:01:47] They had to cuddle up against the wire, the wire hanger or they got a cuddly, a cuddly soft teddy. And the ones that were more securely attached were the ones that had the cuddly soft teddy rather than the wire hanger. I have no idea what you're talking about. Yeah, it was a social experiment that was done probably in the 70s or 60s or something. Anyway, that to say, Kevin is the cuddly teddy. Why do you kick off this week's episode? Thank you for joining us.

[00:02:14] Listen, if your clients are still drowning in manual HR and payroll admin, you need to tell them about Employment Hero. It's an AI-powered platform that actually takes action. Automated candidate screening, smart payroll, real-time compliance updates, the works. HR, recruitment, payroll and benefits all in one place. And here's the bit you'll love. It integrates with your accounting software. So no more chasing down payroll data. Your clients and their team get time back. You get clean data.

[00:02:44] Everyone wins. Head to employmenthero.co.uk to see why thousands of UK businesses and bureaus have already made the switch. The big one from FreeAgent is they've shipped a new VAT return improvements feature and capability, which is all obviously linked to making tax digital, which hopefully all of our listeners know plenty about. The update is obviously part of the FreeAgent push to deliver making tax digital for compliance for simple for their users.

[00:03:13] Obviously relevant for sole traders and landlords where revenue kicks in above £50,000 from April 2026. So a couple of thoughts on this, like making tax digital is obviously, I'd say the compounding rollout is happening very quickly. It's great to see it. It's live for the first wave users from April 2026 for accountants and obviously servicing their clients. Using FreeAgent gives us a much smoother VAT workflow from what we're seeing from the release.

[00:03:43] And I also call out, like it's worth noting that FreeAgent is part of obviously NatWest Group, which is a interesting play for financial services business. And probably looking right through to that, I guess, compliance ownership of the SMEs that they serve. So really, really interesting, I guess, regular feature improvement from FreeAgent. Yeah, I think it's good because they do something that just allows you to see it at a glance.

[00:04:08] So there's something that, OK, is there anything that you want to run a VAT check on? So it's a VAT checker. And it's quite, you know, it's quite nifty. Like you said, some of these things are coming to the fore because of MTD. But in general, they've showed that continuous improvement week to week. And so, yeah, it's quite slick. Ryan, have you had a look at this?

[00:04:31] Yeah, I think this is emulating what Sage did a while ago, where they have a, well, let's call it a smart VAT system now, where it's prompting things that might be wrong. And we talked about this a couple of episodes ago, I think, where it feels like Xero and Intuit have kind of moved upwards. And Sage and FreeAgent have kind of moved downwards in the space, in that kind of small business space.

[00:04:57] And it's interesting to see that FreeAgent and Sage seem to be competing a bit on functionality, just the way that Intuit and Xero have been. But yeah, great release. I've got to say, probably not as MTD focused, because on the VAT side, I guess we've had that for a little while. But anything at the moment that helps with the reporting, the regular reporting, on a quarterly basis, be that VAT or MTD frequent tax, FreeAgent are all over it.

[00:05:24] So yeah, looks like a great trio of releases in the VAT section from FreeAgent. Hmm, very nice. And Dext is turning their AI assist from a configurable tool into something that firms can use straight away. So the nub of it is that setup friction seems to be the enemy of AI adoption. And Dext is standardizing bookkeeping judgments.

[00:05:48] So one thing that is quite interesting about this is that Dext is creating and maintaining core guidance items that practices choose which ones they get to activate and which clients they apply to. Core guidance is something that it gives every partner a ready-made library of compliance-aligned bookkeeping rules for the common workflows included in the existing software.

[00:06:18] But AI tools like this tend to fail at the setup stages because they don't always have like maybe the promise of confidence in how to configure the logic properly. So I think this is where Dext is doing something slightly different. They're owning the compliance-aligned baseline and the firm owns a client-level judgment. So the 83% automated edits claim that they're making is strong.

[00:06:48] But then again, there's some stuff that Dext is kind of moving beyond the capture into the decision layer of the bookkeeping. And what else is there about this? I guess AI assist was about applying the kind of firm-specific logic. And core guidance adds a shared baseline for the common bookkeeping treatment. So there's some things that they've tried to build into there to say not just, okay, trust the AI, please trust the AI.

[00:07:13] It's giving things like practice activation controls and client assignment or auto-apply versus review. And that's something that we've seen Dext moving more and more towards in previous years where they've had something that can actually like, okay, do a kind of whizzy thing in one go. And then it stops and says, hey, do you want to review and X, Y, and Z this.

[00:07:36] So they're just being a bit clever about how rules are now being applied and how you can get into a state that you're just much more confident with the trust that you can give it. Yeah, I'd say Dave Selleck has been talking very candidly about the fact that standard LLMs out of the box cannot do accounting. They try and give you answers, but they don't have core guardrails to hold, you know, to limit them going beyond.

[00:08:03] And this is exactly what I believe Dext will bring in, those guardrails. I, you know, apply this in this circumstance. Now, as Damon Anson has commented and critiqued one of our posts on Dext AI Assist, is this AI or is this automation? Because is it following, you know, a strict rule? Well, and it's not really AI.

[00:08:25] So I think this is maybe the way it's been built, a blend of like a core AI that's got underlying automation. But yeah, this is where we're all going. The one thing that I think is going to be interesting about this, I don't know if you've got details on it, is how did they come up with it? Did they come up with it based on what people have put through AI Assist already? And therefore, the accountants that are using it are training the Dext engine to learn and evolve and provide kind of the structure.

[00:08:55] Because then all we're doing as accountants is we're teaching the software vendor how to be accountants, right? And that's getting a bit controversial. It is, isn't it? Yeah. So it'd be good to know how they're doing that. They may have, you know, experts internally. I know they've got lots of accountants and they may be building that out. But if it's coming from what we're doing and how we're interacting with the system, I don't know if that makes me feel a little uncomfortable. I don't know where I sit on that. I don't know if it's, but usage, like tech platform usage is always something that technology firms look at.

[00:09:25] Like how do people interact with every page? I don't know if you would have seen it, but like, you know, even ourselves, like every page we have like a customer effort score. Right. Like, so we track like, and that's not a new thing. That's not like AI related. That's just good product management practice. But it's interesting the way you say it. Like is the cushion closer to, hey, we're watching what you guys are doing to learn.

[00:09:52] But, you know, to me, AI or automation, like it's a great improvement. Right. Like that's, that should be built into the platform. So I think for Dex users, they're going to be excited by this. Yeah. I agree. I think it's just good, healthy discipline to have product engagement measured and then use that as something that then is the next step within your roadmap or your pipeline.

[00:10:17] And not just sort of build for the sake of building, but instead have something that says, okay, this is how our users are implementing this type of automation and learning, learning heuristic from the AI properly. So that it works. Yeah. So that it actually goes in and does what it's meant to do. Because otherwise, like it said, the setup of that seems to be a friction point.

[00:10:46] And if you can reduce that friction by saying, okay, now we've, we've got this ability over here that just, it's like, you know, okay, we see that getting to step five, you end up like not wanting to do step one or two because they're just ugly. So we've made something pre-configured for you as a step one or two. It just gets you going a little bit quicker. Yes. It's nice. Yeah.

[00:11:10] And after Indie labeled me the cynic of the group in the last episode, I should, I should state that actually, I think this is a great release. I do. I think that we need this to be in more systems and that it's prompting things you should consider and may apply, but you have the control to pick that. We've talked a lot about marketplaces and AI. That's where you think the journey will go, that there'll be hyper-focused in certain areas. You can pick and choose and kind of bolt them on. And I think that's essentially what this is to some extent while being free.

[00:11:39] So I think that it is a great release, even though I probably own the whole cynic thing. All right. Well, look, we could talk about AI assist for Dex forever, but I've got something I think, sir. A pivot. So Zoho Books, which we had not talked about on this show for years and then came out with Corporation Tax and Accounts Compliance early this year. They're now doing a roadshow for practices. I don't think they've ever done a roadshow for practices.

[00:12:09] They're never at events. And now they seem to have gone, oh, we've got accounts and tax. We've got to talk to accountants. And I just feel that it's kind of come out of nowhere. Now, I think they've been doing one for businesses because they imply that it's now crafted for practices, not just businesses. And it's coming through the summer. They're going to have, I think, in six cities over the course of the summer. And they're going to be focusing on MTT for income tax, CT600 and final accounts, i.e. the new area of their system.

[00:12:37] But also talking then more generally about how AI is simplifying accounting and bookkeeping. And they'll then demo, obviously, the Zoho Books and Zoho Practice. So I think they're trying to penetrate that accounting market, which they've really essentially ignored for quite a long period of time. And that's probably why I've just not talked about them because they've always seen Zoho as that bit where they get a client in one area of their system. And then they upsell all other areas until the client's kind of on that whole ledger or the whole ecosystem.

[00:13:06] If they want to move, it's hard to do so because you kind of have to move everything because everything just talks to itself. And so maybe this is them going, well, we've kind of hit what we think is our critical mass of that route. And now we need to talk to accountants. It'll be interesting to know if anyone's going to go to this. So if you are going to go to the Zoho roadshows, please do reach out. And if you do go to them, let us know how they are because I want to benchmark them against others. And I'm not planning to go this year.

[00:13:35] Yeah, I think you're right. I've spent a lot of time in that space and I never really saw them at events or they popped up at events that were quite, I'd call it like the generic business shows where you could go and see a range of different vendors and suppliers. But they have been around quite a long time and they like to have a presence in the UK and it appears to do well. So I think this is going to be quite interesting and yet another competitor to reel off as we're going through the long list. I think they're a red colour.

[00:14:05] So maybe they have a bit of advantage of separating from blue and green. Yeah, red and white, I think. And just I guess for full clarity, they're supposedly in six cities, but they only list five on the list of cities are going to be in London, Bristol, Manchester, Lees and Sheffield all through June. So by the time you listen to this, you possibly have missed the chance to attend some. Okay, Bright has launched Oscar, much to my dismay.

[00:14:33] And also, you know, it's just a thing that I will always pick on. Why can't we pick a more diverse name? But anyway, Bright has launched Oscar, its payroll assistant. Oscar, good boy, is a good example of AI being useful because the workflow is fairly narrow and painful. And luckily we have Kevin on, so it's good to get a perspective on this. So new starter onboarding is a classic payroll bureau headache.

[00:15:03] So the chasing of the P45s, the bank details, etc. Missing information, the handwritten forms. Bright Pay is using Oscar to automate that process through WhatsApp, collecting and validating information before it reaches the payroll dashboards. The claimed improvement is fairly significant. A process that can take up to seven days in somebody chasing is reduced to around 1.5 hours. Again, assuming that you've got everything there in one place.

[00:15:32] With the bureau spending less than five minutes per new starter. That is huge. It's not something that I'd say is super glam, super practical. Yes, it's the stuff that we said and that we picked up on our last podcast with a podcast about Lucy Cohen's article and the sort of hidden drawer full of receipts. This is like, okay, you know, putting it into something that is easy.

[00:15:58] It's a general purpose assistant that can do the cleaning up and making sure that the data is there, fewer chases, fewer errors and overall better onboarding experience. So the interesting commercial angle is that some practices are already packaging Oscar as a chargeable onboarding service.

[00:16:20] So little old Oscar has got a price tag on his head and it turns that AI from an internal efficiency into a revenue line, which is a service. Obviously, the caution is, you know, the payroll data is sensitive. So while WhatsApp is convenient and I'm super aware of this because I'm not a very, I think John had asked us to share some sort of ID data on WhatsApp. And I remember saying, I'm not giving anything to Meta.

[00:16:49] And he was quite shocked about that. So I made him download signal. I'm not even sure about that was that manifested, but I will say that, you know, we want clarity on the consent here on the document retention and making sure that the data is secure. And then what happens when the AI kind of misreads any of that information? I guess that's something that is a question mark. I'm very, very intrigued, very intrigued about your perspective on this.

[00:17:14] Kevin, as someone that works in the space, can you tell me, like, is this efficiency as a paid service something that is scalable? And what's the deal with security? I'm surprised that it could be a paid service because, like, onboarding an employee doesn't feel chunky enough to go and say there's this additional value, right, to a client. And, in fact, it's a system improvement that truly benefits the user. So that kind of caught me off guard a little bit.

[00:17:44] But, you know, there's savvy businesses out there that see opportunity and everything. And maybe I'm a bit blind to it. So I think it's a clever release that they've made. Like, you know, usage of WhatsApp is absolutely enormous. I'm glad you didn't mention SMS because I think if I get one, I think it's definitely a scam rather than the other way around. But it's living in the world where the new candidate or new employee is, like, in their WhatsApp.

[00:18:13] Onboarding is a real pain point at scale. People usually, if they do one onboard a week or one a month, it can take hours. But if they're doing more than that, like, that's when it really gets, like, painful. So I think it's a clever release that they've made. And getting the information right from day one with onboarding is absolutely crucial, as you guys know, for pretty much every system that that employee file feeds into. Yeah.

[00:18:38] Yeah, what I would say is the seven days to 1.5 hours feels like it's ignored the whole most of the seven days is waiting for the person to reply in the first place. And obviously the 1.5 hours test, it's like, oh, they've replied straight away. But how I imagine this is being rolled out as a charitable service is that the business owner would be responsible for getting all of this information, collating it, and then sending it through to accountant or bureau.

[00:19:03] And the bureau is now going, well, you no longer need to do that job because all you need to do is send across or connect into the WhatsApp, however it's working. And then our AI will do that for you. So as we've saved you time, we are now charging for it. That would make sense to me. But going back to the security element, I don't understand what the difference would be of just sending an invite to a new system through WhatsApp. If we've got a phone number, et cetera, it could then do that and then it's inside that portal in a much more secure way.

[00:19:32] As an employee, I would much prefer to be putting my data directly into a system rather than into WhatsApp, as you say, Indy. So I'm hoping that maybe the way this word is that WhatsApp is the initial communication channel and then it goes into an interactive portal rather than going through WhatsApp. But yeah, it's an interesting one. But I want to kind of lay something on the top of this because I picked up an article where they were talking about how they've designed their AI rollout plan.

[00:19:58] And this kind of leads into different ways that different software vendors approach this. So some, if we take like the likes of Sage and Zero, early on they released an AI agent. We're doing AI. It's this. It does virtually nothing. And then they've kind of gone, well, that didn't really work. And they've kind of rebuilt their engine underneath. Right.

[00:20:20] When we're not releasing anything at first, we want to map how, because, you know, the big business across multiple different areas, map how every job that sits around compliance is working. Tracing, you know, how the journey is going through the software and then rebuilding, restructuring the system underneath to enable the AI to be more deeply embedded. And I think that's where most software vendors are going now, right? It's that deep embedding of AI directly inside of the system. And I think that is a smart way of doing it.

[00:20:49] It's hard to do it on legacy software. So what I would expect is that this is more sat alongside, but deeply interacting rather than a core element like core underneath. Maybe they have completely rebuilt the code to enable it. But if you take like a, from a finance ledger perspective and take digits, which is built as an AI LLM, sorry, an AI finance system, that's going to be very different to how like a zero is built. That's trying to now plug in AI.

[00:21:18] So I would assume it's more alongside, but I think it's still a great approach to them when you're a more legacy provider. What do you think on that, Kieran? Because I guess the, I assume, because the deployment here has been around for a certain amount of time. Yeah, it's 11 years. In fact, the way we're thinking about it is it does start relatively basic. You know, we have hero AI as we call ours, hero AI, not Oscar.

[00:21:47] So maybe a bit more vanilla indie. But it's... I like it. It's agnostic. A hero can be in any world, in any language. Yeah. And it also plays to the user, right? Like, you know, this is for you. And it's the hero AI is the layer that controls, I guess, all the touch points of AI in the system. And effectively, what are you doing?

[00:22:10] Like, you're pulling data and actually making it a sense, you know, dependent on the role in the business or asking questions that you need answers to pretty quickly based on the data that sits in the file. And then there's elements where there's much more specific capability in AI. And that's where we're thinking about, like, how do we deploy, I guess, the agents as such. But they will still be housed in hero AI.

[00:22:35] So it could be, you know, there'll be a payroll, pre-payroll specific agent that's really trained on looking for payroll anomalies, anything pre-pay run. And then we're looking at things like a compliance agent, which is very much like HR perspective, right? But you would still access and drive through one point. And the compliance piece is, hey, we're able to read all of your internal company policies. We're able to read all of your employment contracts. And we're able to read the Employment Rights Act.

[00:23:03] Like, and let's actually spin up, like, you know, a dashboard to show you where you have risk in the business. So this is all going to become table stakes, I think, pretty quickly for any cloud system that can deploy, like, agentic capability. For the advanced ones, right, like yourselves, where, and that's really going to create a much bigger divide from those traditional systems that most kind of bureaus have been stuck on for a long period of time to the modern ones, which they need to transition to.

[00:23:30] And we're still talking about the whole thing of getting onto cloud. You need to get onto cloud to get the most of this AI stuff. Now, one bit that I did pick up in this article actually builds on the stuff you were talking about, Indi, which is that Oscar element. So there's a few little bits I just want to pick out that I think accentuate and maybe elaborate on some of the stuff we were talking about. So firstly, Oscar does not commit anything directly to BrightPan on its own. It's gathering the data and it's passing it through for the team to review. So that's the first good step. It is not committing anything. It is purely doing that data gathering.

[00:24:00] And then it's saying about how it's interacting through WhatsApp. So it starts a conversation through WhatsApp, but the interaction sits inside a protected environment behind a BrightBoggin. So it does, I guess, resolve that security element. So exactly what we were hoping it would do, essentially, it drives the conversation to that point where the new employee is probably quite regularly interacting, apart from yourself, that will be in Signal. I'm not sure when they get into that.

[00:24:25] But it will then pull them into the system to enable that data capture, which is what you'd want to happen. So a good positive, and then admittedly, they're just tiny points at the bottom of this article that elaborate on the Oscar site. So yeah, very, very good elements and good consideration there on how BrightBoggin are. I don't know. I understand that. Sorry, that they have the conversation over WhatsApp.

[00:24:49] They collect everything they need for payroll and allow them to photograph documents. But what, the whole WhatsApp interaction sits behind a BrightBoggin or just the bit? I'm not sure on the photographing documents area, but I would assume that it's asking questions with prompts to follow in to a logged in secure area. Once again, assumption here. And then it will now allow them to add the documents in there.

[00:25:15] Now, if it is where they photograph and then they're sending it through WhatsApp and it's gathering it, well, I agree, then the protected environment is a misstatement on this article because it's not protected if you're sending it that way. They need to get to the point where it's asking for data and it's driving them into the platform to put that data in. That's the only way to keep that secure. So maybe it's something we can follow up with Brighton and get clarity on because I agree the way this is worded is it seems to be contradictory.

[00:25:42] And so I admittedly made an assumption and assumed that the tech would be smarter than the person that... You described what you did. Yeah, exactly. You described why I want to have it. Our expectation is that you can have a conversation, but you're not allowed to share any data and it has to be somewhere else. And actually in one slight public service announcement, PSA, put a passcode on your WhatsApp.

[00:26:09] From now on, people just highlight it, add the face ID recognition or passcode. It takes two seconds to do. But if someone yoinks the phone out of your hand and you haven't locked all this stuff, then if you are putting any sort of sensitive data in your WhatsApp, they've got access to it. It'll be in their photos. So don't think it makes a difference. Yeah. Well, put a password protect on your photos as well. That's also... That's what the whole... Don't take your own phone out of your pocket in public.

[00:26:39] Don't put your phone out of your phone. It's impossible not to do that, you know, Kevin. Isn't like... It's tricky, right? Because you're seeing these guys snatching phones and they're inside the phone in 20 seconds, right? Like, it's just... It is actually a little bit kind of concerning. It is. Because you run your life, right? Like, off that device in a lot of cases. So, yeah. It is indeed. Yeah, you've just been... I haven't password protect... Don't do it now. Yes, no, no.

[00:27:08] Because how old it is... I'm doing a PSA. You highlight the app. If you have an iPhone, if you have an Android, you're on your own. You highlight the app. Um, you, uh, if you press down on it, it says require face ID and then you just require face ID. Done. Now, that requires... So, are you telling us that you hadn't enabled this previously? I hadn't enabled it on my photos. I hadn't enabled it on everything else. LinkedIn, WhatsApp, messages, anything.

[00:27:38] You name it. Requires face ID. Got it. Cool. All right. Don't try and come after my phone. You don't have problems. Just don't turn around and show us what's on the screen, right? Um, shall we move on? I'll take it on. Memo, who we know... I've very much been focused on AORP cash flow, um, but are extending into the reconciliation territory, which feels like a really natural step, but it's, it's really interesting to see businesses in that space,

[00:28:08] you know, looking around for where else they could actually build value into their capability. So, you could call it a strategic shift. No longer just chasing AOR, um, running AP and, and, uh, processing payments, but it's kind of moving towards that month-end control layer, which is really interesting. And so, um, an interesting move. They're calling it detective work, not judgment work is the key line. So, matching, variance tracing, evidence scattering, painful, but all, um, I guess out there for automation.

[00:28:38] And there's also the approval layers, materiality checks, tax exposure, final judgment, stay human, um, which is the right frame and probably similar to something that we spoke about earlier. So, don't really have a, a big opinion on this one right now. Like, I've, I've seen businesses in the AOR and AP space be around for a long time. I think what I'd be wondering more is like, how do clients like receive that? Like, are they kind of saying, hey, why are you guys doing this? Or we have it already?

[00:29:06] Or that's nice to have, but I didn't expect you to do it. Um, and I think we're going to see quite a lot of that product scope creep again. I think we've already experienced it over the last 10 years, but I think we'll see it again and again. I've got a perspective on this one, actually, because obviously we, we saw, like you said, the evolution of Mimo.

[00:29:28] And a few podcasts ago, I think it was, I can't remember, I was way back, maybe 10 episodes, 20 episodes ago, we spoke about Mimo and how they've now started working with a handful of the top 20 firms in the UK. They had done some work with one of those top UK firms and they ran, they tested their balance sheet reconciliation.

[00:29:52] Um, and they ran it on a real client's payroll and they, um, they showed that the agent serviced, surfaced a variance on the PAYE balance and the GL showed 5,168 pounds. Um, but the agent said that it should be 3,606 pounds. And the reason was a missing employment allowance entry that the preparer had not posted.

[00:30:19] So the Mimo associate caught that within minutes and allowed, allowed, you know, the accountant to fix it in this case. So, um, what they're really doing here is, I mean, I know they've done a little bit of a pivot in terms of their focus. One of the things I think that they said when I spoke to Henrik about this was, um, he said, look, you know, this is like a, uh, you know, when you use the receivable side of what they do,

[00:30:48] it's more downstream and there's value to be added in the financing of how that could work. But in order to get that downstream activity right, you need to have the data layer very, very tight. And in terms of having that data, it needs to be as real time as possible to build that picture. And what they found is that that just did, that was just not there.

[00:31:09] So what they've ended up doing is, is sort of, it feels less like a pivot, more like an upstream, upstream move for them to then really make sure that they have the receivables on the tail end in a state that could be financed more easily. And this is showing, okay, you know, we're a tech company, so we're going to, we're going to then try and improve how that, that, you know, that process of data is handled.

[00:31:33] And while we're at it, let's, you know, build something that helps you reconcile more easily, handle the clothes more easily. Um, and by doing that, we've then built ourselves a whole new, um, service line, a whole new stream of business. Yeah. And, uh, name dropping Henrik, Henrik Grimm, main CEO, CEO of, uh, MIMO? CEO. Yeah. And so the only thing I would say on this is that it's, it feels like everyone's going after period clothes at the moment.

[00:32:00] That is the, the new topic we've had. We've had kind of payments at one point. We've had like debt chasing another point. We've, we now get, we've got, we've got the, the big thing of period clothes that everyone wants to go after. And I don't know who's going to win on this yet. And I don't know if it should sit in the ledger. I don't know if it should sit in a separate app and sit in, you know, part of something else, such as what MIMO have done. What I think will define this is how well it's built and how easy is it to use. And what I would do is I would shout out MIMO that I've seen this and it did feel quite slick in what they're built and how it was doing it.

[00:32:29] I haven't played with it, so I can't say how accurate it is, but it was feeling very slick from a navigation point of view. And that's, you know, that's quite key for an area where you're trying to redefine how accountants work, um, which is changing period clothes. So yeah, I think it's a, it's a good evolution and makes much more sense now that you have explained it. Indeed, I think I was with Kevin going, uh, I don't really understand why they've gone down this, but if they want to go in that financing area, the data has to be reliable. They needed to resolve it, right?

[00:32:58] I've got something which I've not really done for a while. I used to talk about the new releases from the Xero app store just because I love talking about new technology. And I've not actually seen any blogs about it, but what Xero are doing now is talking about specific apps and then diving into examples. And they've highlighted one called Trove. And I've never heard of Trove. And I was like, what is Trove? Well, essentially it is a new credit controller.

[00:33:24] So it is similar to, I guess, what Chaser would, uh, would be in that, in that space. And what they are advertising is that it can reduce overdue invoices by 60%. Now I remember Chase used to say it was 80%. So I don't know if, if Trove has just been more realistic or if it's just less effective. It's impossible to tell from just this stat that's been released, but it is showing a slightly different approach to how they built this tech product out.

[00:33:51] It's been around since November last year, has been kind of growing in, in utilization. And has the standard stuff of automated emails that look and sound like you, kind of the standard bits of the credit control. You've got full control over when they go out. So you can align that with your, uh, when you're reconciling the bank to make sure that that is completely up to date and live. You can integrate this into email, which is standard, but into Slack, MS Teams and CRM. So you can get updates throughout multiple different channels.

[00:34:20] But the main thing they also say is they're bootstrapped. They do not, uh, they're not taking on P-backed business, like P-backed money. So they are trying to build this up with, uh, just the, I guess, the funds they had at launch. And I think that's nice to see that people are still doing that because I know there's so much competition out there. So I just want to shout out Trove, no, no massive, uh, I guess, releases or functionality beyond stuff that's out there.

[00:34:46] But I do think that it's an interesting app that's hit the store at the end of last year. Yeah. It's great to see like newer businesses coming in. You know, Chase has been around maybe 15 years. Long time. Yeah. You know, like you can never say like the problem is solved. Like I think the technology world. So it's, it's cool to hear that this business is saying, Hey, I think we can actually do this better. Right. Like and come in and haven't seen the system.

[00:35:12] I can't say it competes directly with, with Chase, but it sounds like at the outset that's the, the, the structure of it. Um, but yeah, anything that can help as we were reeling off, like all the things that connects to us, like, does it connect to the person's WhatsApp? So you could chase them there as well. Well, you have to do signal, I think for Indy. So it's, uh, it's where they go first. It needs to be. Ironically, you message me on WhatsApp a lot, Indy. It's not that, I mean, obviously it's just you.

[00:35:42] It's not like, you know, we have some chit chat. But I just mean that, you know, one of the things I'm most aware of is you, you're giving a lot of data away. And in this day and age, what we have learned is data is very, very golden for us all. And I think it's a risk, you know, in the future that we empower more of these tech companies with our data. So just be careful about what you put out there. So I've got something from Suite Files.

[00:36:08] So Suite Files has long supported the templates that auto-populate placeholder fields with client data. And they've done a really good job of managing those, you know, those examples of the forms and the data that needs to be populated. But they've come out with something a little bit different. And they've launched something called Smart Templates, which it turns the existing document into a working template.

[00:36:34] Meaning that you can manually, well, instead of manually identifying every field, it automatically recognizes which fields need to become a placeholder and configures each one correctly. So it knows enough about how the system handles and merges that logic to make it all behave. So in the past, that's been something that's been fairly cumbersome.

[00:36:57] But Smart Templates has removed that bottleneck and allows the system to identify the fields that should become everything that is part of a working document and turns that static document into something that's a reusable template without anyone needing to understand all of the mechanics behind or underneath it. So what used to be like a half day project for a power user has become something that any team member can do in minutes. And the result is more than just a faster template creation.

[00:37:26] It allows the firm to do the template work that they usually do instead of doing the setup and doing all of the keying in of that data. So it strengthens a core workspace experience for them. And we're really proud that Suite Files has just come on to support the podcast as well. So it's great to talk about them. And Ryan, did you talk about it? Yes, well, I want to shout this out because I actually saw this at Context pretty much when it had just been launched.

[00:37:54] And it was very good. I kept asking, oh, does it do that? And it's like, yeah, just show me it. What I would dispute is that any team member could do it because you still have to understand the context of what you're doing. But still, it does mean that the barrier to entry, you do not need to have no complex word coding. You can kind of go straight in and just drag. Well, first it will prompt. So something really, really close to my heart, the capability of accountants.

[00:38:22] And Accounting Web has published a piece called The AI Skills Gap. And I'm just going to focus in a little bit more because my eyesight is getting worse as I get older. But talking about the gap in the accounting profession, like the data, 58% of finance departments report skills gaps. Yet only 19% of accounting professionals use AI tools daily. And 70% have never used AI at work at all. So this is a bit of a tough read for me.

[00:38:49] And it's kind of like, I think there's both sides to it. Okay, it's the finance department or the firm enabling and allowing and educating their teams on how best to use AI. But I kind of have to look at the individuals as well and say, what is the actual barriers to you learning AI?

[00:39:10] Now, you could split that and say, you could be in your own time, you know, using AI tools and using a data set that you've built or you can find online to actually practice AI yourself. Rather than being, I guess, waiting to be educated by your employer.

[00:39:28] And I think that is actually where I'd be kind of calling out accountants and saying, well, if you're one in five people who are using AI, 80% of people just are like, they're kind of like a little bit of what are they waiting for? Is how I feel about it. Because I feel strongly that I don't want the profession to fall behind.

[00:39:48] I'm seeing the accountants in our business use AI heavily and turn really heavy Excel sheets and databases into beautiful dashboards in 10, 15 minutes. And I'm talking about heavy data sets. And they're using Cloud if anyone's interested. But it does bother me to see such a low uptake in the industry again. And I'm like, you know, we know that there's such a key relationship point between accountants and SMEs.

[00:40:17] So there is a knock on effect that I do get concerned about. So, yeah, speaking from the heart there, I really want our industry to step up. Yeah. Well, I can probably give you four reasons that I think this has occurred. I don't know how accurate they're going to be, but I'm going to give you four. First, productivity targets. It's how we are in the accountancy sector very stringent in that this is the target you need to hit. And we're not going to be leaning on it. We need to hit our margins. You need to hit that.

[00:40:44] And when people don't hit that, even if they're trying to do things that are going to benefit the business, they will be dragged back to, you know, you not hitting your productivity target. And because of that, it's, I guess, incentivized people to focus on doing charge of work as much as they can, rather than being as efficient in the long term. You've then got the fact that it costs money to get the paid versions of this. And if you don't do the paid versions, then we don't want to put, you know, client sensitive data in,

[00:41:14] but you probably need to use client sensitive data to properly learn how to use the AI effectively or for your role. So you've kind of got that chicken and egg of, are we going to let people go out and, I guess, find a tool and then pay them to do that? And are they going to do it effectively or not without monitoring them? We probably do need to monitor it. If we're going to monitor it, we need to do it centrally. And therefore comes to the third thing, which is we always want to do our training centrally in accounting. We want to have control over what people are being trained on. We know the quality of it. And we can, I guess, ensure that people are ticking the box.

[00:41:42] We don't normally just go, here is a set amount of time. Go, be free, learn, which is not great. We probably should do that now at the stage we're at, but that's not what we're good at. And then fourthly, I think it is a wild west of how good AI training is out there at the moment. It's very limited for specific accounting training. Something that I know that we are and have been looking at for a while is how we can improve that for accountancy firms.

[00:42:08] Because you can have generic AI training, but that doesn't necessarily fit in with your day-to-day work that you're doing. So there's a big combination of factors. Some of the big firms have just gone, have a license, have some capacity, go and do it. But we've also not seen a lot of that come back into increased productivity. And there's always a lag, right? We're not necessarily seeing it come back into, we can now take on more work and do more.

[00:42:32] I'm really, like, I was really happy to hear, like when we caught up last month and you guys were talking about, like, the AI training. Because I do think that structure is needed and I think it's so new that, like, a lot of the firms are saying, well, how do we get started here? Like, so it feels like the handbrake is on, like, and in a really hard way, which is probably, like, your first three reasons. Like, lots of reasons why it's not happening. And maybe there's just not enough cases where people are seeing huge improvements to productivity and they can't put their finger on it and they can't prove it.

[00:43:02] And therefore it becomes a case of, well, we'll kind of wait and see. I get it with the client data, like, really, really important. I think we kind of have to back ourselves to kind of come up with solutions, all right? Like, to say, well, we want to learn this, we need to learn it, and therefore go and learn it. And maybe I'm speaking too much from a bias from our perspective internally, but we got the training, which was, I would say, relatively basic introduction, how to, here are some prompts. And then it was like, go learn it, like, yourself.

[00:43:32] Like, you know, really go and try and see what you can come up with and then share and then learn from the sharing. And it's kind of that step, right? Like, we're just like, I just need to, like, take off the learner plates, which I still have mine on, and I think everybody does. But at that such low percentage of uptake in the industry, I'm like, oof, that could hurt in a year or two. No, I agree. And we've done exactly the same thing as you, Kevin. We've kind of rolled out a couple of licenses.

[00:44:02] People then come up with use cases have done it, including myself, and then we've fed that back. And then that's kind of increased the uptake. And I can state that it does improve productivity way beyond the software cost. So in the same way you would adopt a SaaS system, adopting an AI system, if used in the right way in the right areas, it can make a big improvement. And that brings us to a wrap on this week's Digitals in a Crawl World podcast episode. Thanks for joining us. Thanks for tuning in.

[00:44:31] If you haven't already done, so please do follow our LinkedIn page because what we try to do is cut this episode up into all of the handy little five-minute sections. It gives you a little bit of time in the week and not so actively on one day when the podcast comes out in one go. You can actually listen to it and join in with the discussion because obviously there's some things that we know that you'll have strong perspectives on. And it's easier to do that in something like a LinkedIn or a YouTube.

[00:44:58] So if you're not already following us or subscribed on there, then we'd appreciate it if you can. If you haven't left us a podcast review, please jump on and give us a five-star review. It takes two minutes. If you want to leave us some nice words, then we always read through those. And we will catch you on the next one.