ERP SoftCo Brings AI Finance Agents to Real Customers

Thank you, Tom. Special guest, I feel honoured. I do attend these every week and I do listening and they are fantastic.

So this I think will be a little bit different. So I'm not a developer, I'm not particularly technical, but what I'm doing at the moment is spending a lot of time talking to customers, showing the new platform, and there's a lot of enthusiasm for that. It's going on really, really well.

We're making some great progress with Navigator. But what I've also started to do is think about, so within the Core ELP, well, we have financials and dimensions and a large customer base. And one of the questions we keep getting asked is, How do I become more efficient as a team?

So these are finance teams. They have fairly painful processes in terms of every month I have to do a close the box and there's a load of preparation around that. But generally when we talk to customers, they are genuinely asking this question, I want to become more efficient.

I know AI is a thing, but I don't know how to use it. and I don't know what to do. So we've got a perfect opportunity to start to solve those problems for customers if we listen to them and we come up with solutions. And I genuinely think, I'm really excited about the Evo platform because I genuinely think we have something amazing.

And when you start to play around with the components and start to build things and it starts to hang together and you get results back from the agents and the teamwork, etc. I think we can easily wow customers and I think we can do a lot of that now. So what I'm going to do is I'm just going to talk through some of the things we've been doing, some of the things that we are showing to customers, we're working on them with customers, getting feedback on them.

And I'm just going to give you kind of a brain dump, I guess. And this is probably something along the lines similar to how I take a customer through. what our platform is. And I'm doing lots of sales sessions, lots of customer success sessions.

We've probably had about 100 customers in on customer success days where we've demoed Evo Builder, we've demoed Navigator, and the feedback's been amazing. So I'm going to, so we have a great. Story around what we've got with Navigator, so we've probably got around 50 MFEs at the moment.

So we've got a load of quick actions so I can quickly find a customer. Then this obviously layers on top of dimensions, which has got a traditional desktop front end and financials, which is was refreshed, but it was refreshed five years ago. I can't quite believe that, but five years ago.

So this is a whole new user experience. And then we start to look at kind of slightly more clever of things in terms of, well, can I look at, you know, a quick section that gives me a nice view of my P& L and can I have philtres on it, etc. And that, when you show that to customers, just because it looks fantastic.

It's nice and shiny. It gives them lots of information. That goes down really, really well as well.

So that's where we're on kind of navigator. But the thing that is really starting to get exciting is the combination of teamwork, connectors, agent creator, whatever. I don't know what these products are going to be called, but I'm calling them this for now. and then workflow automation.

So teamwork, we'll start there. And I think this is probably how we'll explain its customers as well. So teamwork is a great starting point, I think, to have a think about some of the challenges that we can address for customers.

So I only got my head around this. probably 3 weeks ago. And started off with, I wonder if this teamwork capability can do something like a P&L analysis, obviously within financials and dimensions and all finance products, actually. You know, the important thing is, it's a view of data, it's a view of how your business is performing. and the profit and loss is an essential part of that.

Within all of our products, we have reports that present that data, but what the reports don't do is have the kind of AI analysis or the narrative around them that gives you that kind of deep insight. A while ago, I tried taking the data from financials, importing it into Claude and running a report. And I was amazed and then we all go through this kind of wow moment I think when we're using AI of it did a really deep analysis of this P& L data and gave me a nice report.

So that was kind of that was something we've always wanted to try and put in the product. I don't think we need to now and I think we have a really fantastic approach to to do that. So, and I'm not going to run these because they take a while to run, they probably take 10 minutes to run, but if you think it's taking 10 minutes to run, it would probably take a person half a day, a day to go through this data, write some slides around it.

So already we can kind of sell that or position that as a massive benefit to customers in terms of Again, this is a wow moment for them. And you think about where we are on the journey, we're way, way ahead of what customers are doing now. Some of them are playing around with things like Microsoft Copilot, a few will have like a Claudia license, some of them will have personal stuff, but there is nobody that is actively using it, certainly not in the ERP core base.

So what this has done is I've asked it to connect to a financials instance and create a P& L summary for the periods in 2023. And 2023 because that's where I know we've got data. And what it's come back with is a nice report.

It's come up with some insights and anomalies. So it's got the commentary around it. So the first time I did that, I was like, wow, that is amazing.

Customers will love that. And then I kind of went, well, I wonder how far can we push it? Could we get it to do a whole board pack?

So again, all businesses or certainly mid-market businesses will have to do a presentation or a set of slides or a deck or a report at the end of the month in terms of how the business is performing. So I, not that one, that's not the right one. So I had to play around with this a little bit.

You see I've had a lot of experimentation. We've got lots of projects in here and I think that's probably a bit of feedback as well. I think using these tools, sometimes they work, sometimes they don't work.

And I think it's just important to to learn what they can do, what they can't do, because it's amazing. So if I go into here, there's this one. And what I created, what I asked it to do was, the instructions were effectively, create me a ball pack.

I want these things, I want P&L, balance, cash. balance sheet, cash flow, et cetera, budget and actuals. I gave it some instructions. I uploaded, I mean, this didn't work great, but it did do something.

I uploaded our corporate template. Obviously, I've got a connector, I've got collaborators in there as well. And what it created, and this actually, when I did it, it was using the the Opus model.

So I think we're not probably not going to use that at the moment, but it kind of gave us a feel for what it could do. So it created this amazing board pack, right? And this is the sort of thing you could hit distribute on, send it out to key stakeholders, and it looks professional, it's saved. an awful lot of time.

I think there's something like 20 slides in here. And again, anyone that's done any sort of management reporting or even reporting up to, you know, the management team here, when you've got to fill in PowerPoints and things, it takes time, right? And you've got to get it correct and you've got to write a narrative around it.

This is a fantastic capability that we can start to template and ship out as part of the teamwork package, I think, that we put in front of our customers. So that was one thing. So again, two things there that I've kind of gone, okay, that's interesting.

Those things are amazing. Another thing, again, every finance team has a month-end close process. What I can do here is I can prompt it and go, how How ready are we?

But what it's doing is it's looking to financials, looking at the state of the data behind the API, and it's given me a view of how ready is the team or the business to close the books. Really, really useful and not something we've ever really done before because we've just not had the time or the... the bandwidth, the builder, anything like that. So then the next thing was, can I stretch it a little bit further?

What else can I do with it? So again, in financial dimensions, obviously we store customers, we store things like payment terms. So each customer, and we do this ourselves against our customers, each customer will have to pay us or the business within a certain period of time and they'll have a credit limit, they'll have to pay within 30, 60, 90 days, whatever it may be.

And what we can do with the AI capabilities that we've now got, we can take that data, we can look at the transaction data that's coming into the system and go, which customers are paying on time and not paying on time, a little bit more proactively than kind of an age debt. And what the report here has done, so I've asked it to analyse the base, look at the data and review the payment against the account terms and then make some recommendations of how you improve that customer payment performance. So some of it will be no improvement needed.

So if I go down to Um... Here, look, what it said is, you know, these are the recommendations for this customer. So it's it needs it's falling behind payment, it needs an immediate action.

So there's someone stop slow payers, right? So here we're going, this is this is what they're on. And I might want to go in and And. click on a find customer app and then change the terms of business against that customer.

So what I want to do, I've not got this working yet, I want to be able to click on this find customer and then I want to be able to, and I've got the artefact in here, which would be my MFE, it wasn't there, MFE, but what it will do is it will then go into Um... my quick action, that will be able to find the customer, that would pop up over the top of the report. It would automatically philtre on the customer, and then I can go into this section and change the terms of business, right, based on the recommendation. So again, customers will be able to start to do those sorts of things, which would have saved them a lot of time just analysing the data, knowing what to do.

And it almost becomes a recommendation engine in that scenario. So that was another one we got quite excited about. And I think there's lots we can do like that.

So the other thing we start to look at is there's an age debt analysis, but again, we do that as a standard report in financial dimensions. We can put an analytical AI insight layer over it with teamwork. Churn prediction, so who are my good payers, who are my bad payers, who's not spending with me, who is spending with me?

Those sorts of things become, you know, template. reports that we can start to put into our offering that we're going to release to customers. Who's propensity to buy, who are the high spenders, who are my good customers, what should I sell them, when should I sell them, etc. So I think there's, you know, and that's just kind of scratching the surface on that.

I think as we start to talk to more and more customers, sit down with them and go, or what do you want to do? What are your problems in your business? We can run teamwork projects with them, for them, or they can do it themselves.

It connects to lots of data, obviously. It connects to financials and dimensions, which has lots of really high value data about the business and gives some great insight. Then the other, the final thing I just want to touch on was maybe there's a different way of using this as well.

So I was kind of like, well, could we have the concept of almost like a digital consultant? So if I was going out to see a customer as a consultant, could I look at their usage of financials and what we, there's a couple of ways we do this, we can look at the what's the customer using through the API? So this is what I've done here.

What I've done is connect to the API, look to the modules and the areas of the system that they're using based on the data that's in the product. And then it's kind of making some assumptions and assessment of the quality of the data, which parts of the system they're using, or the or the errors with the system or the things they're not doing, or the mismatched invoices, etc. But if you imagine you could then add in things like telemetry data, or you could add in some research from the internet about a customer you're going to go and see.

You could add in a catalogue of products that we could then sell them. then we can start to use it ourselves to go, well, how do I deal with this customer? What are the products that they might buy? And what could I give them guidance on in terms of how to get more out of the product they've already got?

And maybe there's a way of automating that over time through agents and upselling, etc. But again, it was kind of another wow, that's amazing. And I know this is kind of reporting base, but that's within the world of ERP, that's really, really important because it's all about cash flow, maximising your business profitability and having successful business.

So that was the first step on the journey for me. And I think, again, a few wow moments, and I think everybody should just have a play around with it. So what we're doing in ERP Core is we're doing these type of sessions and going, look what it can do.

It's amazing. And I will show this to customers. I have shown it to customers.

And again, the feedback is fantastic because you suddenly see the light bulb moment of I don't have to spend all that time creating those reports or doing that analysis. I can, I can, and nobody likes doing that, right? So, so this is fantastic and I think it's we're only just scratching the start.

So what we're going to do here is then, I don't know, I'm sure everybody's aware, but what you can do is you can package your templates. So we're going to have a board pack, a debt analysis pack. We're going to have a load of templates that it will have kind of pre-configured instructions to so that when a customer knows that comes into the product, they know what to do, they know how to run it. because some of the feedback we've got is one, I don't know how to prompt an AI.

Whereas if you can just go click in here, run, give it a month, away you go. And again, that's an easy kind of training and learning process for our customer base. So again, exciting stuff.

The next kind of area I wanted to just touch on is how we're starting to think about agents. So what we, again, if you think of the problem within many businesses, certainly mid-market, it's about efficiency. So that's the challenge we're getting from the customers is how do I become more efficient?

What can I do? And again, one of the big things for customers is how do I close the books in a more efficient way? So, we've started to think through this concept of kind of ongoing processes that can help organisations and finance teams look for problems and errors in the data, in the data that's coming to the systems or the behaviours of the people within the teams and the organisation, and look at how do we how do we automate some of those things.

So, we've got, I've created a couple here, and again, we'll package these up and release them. So, there are, you know, let's check, let's check, are we matching invoices correctly? Is my trial balance out balance?

Have I got an age debt problem? So, what I can do here is I've checked, click on this chap here, looks very happy. And I'm not going to run these because it'll take a bit of time, but I can ask it what?

You can spell, you do. And what it will do, it will tell you what it can do. I'm not necessarily going to run it, but it gives me a, and the way we built these, so I built these using agent builder and I love agent builder, I think it's fantastic, or whatever it's called, co-worker creator.

But I like the wizard where you go through the steps and it may feel like It's quite a long-winded way of, you know, you go through the step, but the outcome you get is you get agents that are very focused and very, they seem to be very reliable. And what we can then do is we can guide end users and customers in terms in terms of the purpose, and we can focus it down on it will do this for you. Whereas I think some of the challenges we get at the moment is or AI is very broad and can do all sorts of things, by having these packaged suite of agents that will do certain things for them, it changes the mindset, gives them a bit of comfort, and we can take them on that journey of, you can do it manually, you can ultimately automate it, you can get feedback, or you can run it as a teamwork project.

And what I'd really like as well is if I could go from a teamwork project, so you can see I've got a board pack rater here, Well, I actually created some of the instructions for the board pack creator agent from the teamwork project. I just prompted it, copied the instructions in, and it gave me an agent to create board packs. The next step I want to get to is how do I then start to automate these?

I've not cracked this yet, but is using workflow automation so that I can set up, again, the agents and maybe they run every day or maybe they run once a month or maybe they, you know, they we get them running overnight and what they'll do, they'll look for errors in the system and in the workflow automation process, I could get it to, I could have it a continuous close, I could get it to analysis. And maybe I can get it to, well, I know I can, when it works, I can get it to trigger alerts through Evo feeds to go, the board pack's ready, or you have a problem with your trial balance misaligning, and this is the reason. So again, I think once we start to package these up as content and ship them on top of the products, we've got something utterly amazing.

And when I start to think of other areas within ERP, so we have kind of a projects area. So businesses that are project focus, we can take the whole approach again through those types of organisations where they're looking at, you know, the project efficiency, project reporting, how do I close it down on my invoices being paid? Have I got the right resource?

Can I schedule really quickly? So all of those things we can think about, how do we, you know, teamwork, convert to an agent and then automate is kind of the way I'm thinking it through. And then that's how I'll explain it to customers.

I think it will work really, really well. So that was, I think that's probably all I've got to cover, Tom. I'm pretty sure that's probably.

I wasn't given a specific time, so I wasn't really aware of how long I was going to go on for, but I think that's all I want to cover. Hopefully that's been interesting, useful. I'm really, really excited and I think the whole of the, well, I think we're all excited, right?

But when you start to see these things come together and when you put them through customers and you get that, okay, this is a different access type of feedback, It's amazing, and it's a lot of the work. It's the work that everybody on this course that right is definitely paying off. It's brilliant.

That's fine.