Industry trailblazers reveal the GenAI tools already reshaping the HR and Payroll functions.

Our expert speakers Barry Phillips and Johnathan Trainor share their top five favourite tools, along with practical use cases, showing how GenAI is transforming everyday work, improving productivity and opening up new possibilities for HR and Payroll teams like never before.

Transcript:

Gosia: Good morning, everyone, and thank you for joining us this morning. My name is Gosia, and I'm from Legal-Island. It's my great pleasure to welcome you here to this webinar entitled "The AI Revolution in HR and Payroll: Five AI Tools Already Reshaping Both".

The two speakers for today are, first of all, Barry Phillips, who is founder of Legal-Island, one of the UK and Ireland's leading employment compliance firms. He has trained hundreds of HR professionals on how to get the best out of GenAI. He's also author of "ChatGPT in HR: A Practical Guide for Employers and HR Professionals", as well as a regular speaker on AI for HR professionals, both online and conference throughout the UK and Ireland.

Barry will be followed by Johnathan Trainor, who is Technical Support Lead at PayEscape. He began his career processing payrolls before progressing to his current role, where he leads technical support and delivers exceptional client service. He also oversaw the development of PayEscape's AI chatbot, helping to enhance support capabilities and provide faster, more accessible assistance for clients.

This webinar is timed to last about 30 minutes with a Q&A at the end. You may like to drop questions into the chat as we go along.

So over to our first speaker, Barry Phillips.

Barry:  Thank you, Gosia, for that very kind introduction, and hello to everybody. I'm delighted to have this opportunity to share what I think are five great tools which are going to be of immediate use to you in HR. It could be that you've seen a few of these tools before. It could be that one or two are new to you. And if that is the case, that I think is really exciting and gives you an opportunity to have a play and test a few of them after this webinar.

I just want to say from the outset that I don't have any commercial interest in these AI tools at all. I'm not on any commission if I manage to sell a subscription to any of them, so I think I'm quite well placed to share the tools with you but also to highlight where I think they don't quite perhaps come up to speed and they still need a bit of improvement.

Before I start on my top five, I just wanted to ask a question of everybody in the format of a poll here. And it's simply this: Do you suspect or know that employees in your organisation are using GenAI tools without authorisation or even when strictly prohibited? It's just a yes or no answer, and we'll reveal the answers to that poll towards the end of this webinar. So thank you very much for participating in that.

It does give me an opportunity to say the tools that I'm going to show you are really great, but please do make sure you have permission from your employer to use them before you start doing just that.

So my top five. In at number five, is image generation from ChatGPT. And to explain why this is in at number five and why I think it's so important, we need to go back to October of 2023, which is when ChatGPT, or OpenAI, first made public the image generation facility that went along with ChatGPT subscription.

And in October, I put in a prompt. I said simply to ChatGPT, "Show me an image of four HR professionals around a table talking about working time". And this is what it came up with. It's not bad, but I think you would agree it's not really usable.

The humans around the table don't really look very human. It's struggling to represent working time and even to spell it properly. But what you probably noticed, too, is it really did struggle representing these things, digits, fingers, toes, feet, etc., at all well.

If you look at the poor gentleman on the right, his hands are merged together. The poor gentleman on the left, he's missing his ring finger, and it looks for some reason wild, but his middle finger is just about to be chopped off as well. So altogether entertaining, amusing, but not really usable.

Since October, this facility has developed and improved significantly. But a month ago, something happened which, for me, was a game changer, and that was that OpenAI developed Version 2 of its image generator, which comes bundled into subscription whether you're Basic or Premium subscription. And it's really good.

I dropped in the same prompt, "four professionals around a table discussing working time", and this is what it came up with. Significantly better and different. And not only are the humans featured looking like real humans, but it comes with a fabulous editing facility. So if you're not happy with your first draft, first version, you can edit, you can tell it to change things, and so on.

So for the second version, I asked it to change two things, and I'm just going to jump between these two images. And so I'm going to ask you, can you see the two changes? I think one is quite obvious. I asked it to change the gentleman on the right for a woman of an older age. But the other thing that you may not have noticed is the working time PowerPoint slide presentation at the back there. I asked it to change the third bullet point, "rest breaks", to "annual leave", and it managed to do it no problems at all.

To really test it, I uploaded an image of myself, the same one that you've just seen, and said, "Drop me into this meeting, please". And not only did it manage to do it, but it actually turned my aspect by 10% to 15% to show me actually looking like I was engaged in the meeting and listening to the gentleman on my right-hand side. So very, very effective and very much improved in just the last month.

So, what's the application of this to HR? I think it's going to really help improve the visibility of HR teams, it offers realistic representations of your workforce now, and images bespoke to your organisation. It may just be this is the end of iStock Photos. Why would you go to someone like that, where they've got hundreds of photographs that are good but not quite right, when you can develop your own thing?

That's first one. Next is Synthesia. For anyone that's not familiar with Synthesia, what this allows you to do is to generate avatars that look like human beings, and you can use them to read out scripts that you also upload to it.

This is beginning to be used by employers for lots of different things, for personalised onboardings, for compliance training, internal comms. And it offers 100 languages. So if you're in an organisation which has employees in another country, Italy, Japan, Korea, it has got a language that you can use to deliver your message in a local language.

They are very realistic. This is an avatar that we use. Unfortunately, I can't show you the video because GoToWebinar, the platform that we use for these webinars, isn't really designed to convey a Synthesia video. But I can tell you they're very realistic now. Give it a few more months, and I think we'll be struggling to differentiate between an avatar and a human being. They're so convincing.

Why I think this is useful to us in HR is that they can be used to personalise onboarding, for health and safety announcements, in training, and just to mix things up, in internal comms.

But I think this is significant because of two reasons. Cost. It's just £22 a month to use this for a reasonable use of these avatars. And it's also very quick to do.

Let's imagine the days prior to this when you had to, in HR, make an announcement. It meant finding somebody that was happy to go in front of the camera, possibly getting a proper camera in, in front of a green screen, autocue, all sorts of complications. It might have taken you even half a day just to produce three or four minutes of a message. This will do it almost instantaneously, and it's very cost-effective. Old days, probably a few hundred quid to do it. Now it's just £22 a month and is being used by quite a few employers that we have been talking to.

Next one in at number three is NotebookLM. So just as Synthesia gives you an opportunity to re-present your HR team to your organisation, this gives you an opportunity to re-present HR information. And I'm thinking particularly of standard documents, static information, the likes of policies that you might find in staff handbooks, etc.

If you go to this website, you'll see that you are given nine options, nine different ways to present something. So let's assume you have an AI use policy. It's just out, and you want to make sure everybody is going to understand its terms. You've got lots of options. You can ask NotebookLM to turn it into a quiz, a mind map, an infographic.

If you chose an audio overview, what would be produced is basically an audio of two people in a kind of podcast environment discussing your AI use policy. Not AI use policies generally, but your particular policy. So if your policy said you can only use Copilot, it would be talking specifically about that.

We did a training video recently for our staff for using LLMs like ChatGPT, Copilot, etc., and we asked it to present it in a video overview. It's what we call the "ABC rule" of using ChatGPT. So I'm just going to show you it very quickly just to give you an indication of what it can do.

Video:  Are you sure you're not accidentally putting your company's sensitive information at risk? Look, we get it. The speed is incredible. But every time you paste something into a public AI tool, you could be opening a door you really don't want to. We're talking about exposing private client data, compromising company security, and even just getting plain wrong information back.

So, how do you get all the benefits without taking on all that risk? Well, the good news is there's a really simple framework to help us out. It comes straight from the Legal-Island AI Use Policy, and it's all about helping us work smarter and safer. It's as easy as ABC.

First up, A is for avoid. Just avoid putting any personal or sensitive commercial data into a public AI. Simple as that. Because once it's in there, you're not getting it back.

Then B is for be sure. Be sure you've gone into the settings and turned off the training function. If you don't, your private queries could end up training the model for everyone else.

And finally, C is for check. Always check the output. AI can get things wrong.

Barry: I think that's probably enough to give you an indication of the sort of quality of the videos that it produces.

This is not expensive. It's about £7 a month for a medium subscription to that, which allows you to use it to reposition and to re-present your information in a different way that is, I think, more accessible and certainly more digestible. So that, I think, is the importance of this particular tool for us all in HR.

On to the next one, and it's in at number two, Wispr Flow. Wispr Flow is like a Dragon for anyone that remembers the first sort of dictation software that came with PCs in the '90s and the noughties.

The difference between Wispr Flow and Dragon, however, is that, quite frankly, it works when Dragon didn't. Dragon was about 70% accurate, but this, in tests that we've done at Legal-Island, is pretty much there. It's 99% dictating what it is that I'm saying to the screen.

The best thing about Wispr Flow is it's platform agnostic. So in other words, if you're in WhatsApp, it will put text into WhatsApp from your speech. If you're in Outlook, if you're in PowerPoint, it doesn't matter where you are, it will still work.

So what are the implications of this, or the uses of this, to you in HR? I think if you've got employees with accessibility issues . . . I'm thinking perhaps people who are dyslexic, or people perhaps that don't type very well, or perhaps employees that have RSI problems, then you can cut out the keyboards by using this speech-to-text facility.

Some organisations, I think, are beginning to look at this and say, "Can we implement a wholesale efficiency improvement here?" Because the stats say that you can speak 20% faster, at least, than you can type. So who knows? Maybe that's an application not only for your HR team, but also for your whole organisation.

In tests, we found that staff here . . . it's a bit of a Marmite tool. They either love it or you don't. I love it for general emails. It's much quicker than I can type. But for other work that requires a little bit more thought and consideration, I prefer to go at the speed that I can type.

So, that's number two. Before we do the big reveal on number one, I just wanted to share this with you. This is about our Certificate in AI for HR. It's a CPD-certified course, the only one that exists specifically for HR. It starts on 11 June. It's online, it's every Thursday between 1:00 and 3:00, and it covers every key component of AI relevant to the HR profession.

For you, for attending this webinar, we're doing a very special feature of 20% off. And to qualify for that 20%, you need to use the code that Gosia has shared with you in the chat.

It's the third time we've run this, and we're getting really, really good reviews from it. It includes a lawyer who will be looking at the legal implications, employment law issues around use of AI and in particular misuse of AI.

Places are going quickly, so if you'd like one, please do apply today, or certainly by the end of the week.

So, to number one, the big reveal, what is it? I think it's simply this, it's workplace LLMs. They are so good, it was impossible not to put them there at number one. I was going to choose between ChatGPT, Claude, and Copilot, but they all have their strengths and weaknesses. So I thought they all belonged on the podium there together.

And what is the application of this? Well, we're seeing wholesale benefits of efficiency improvements across the HR team in drafting job descriptions, interview questions, letters, policies, reviewing contracts, analysing data, etc.

I would say to you, if you haven't really got familiar with an LLM like ChatGPT, Claude, or Copilot, just put it in your diary, 10 minutes every day, just to investigate and try to understand the power of these machines and the benefits that they can bring to you, your HR team, but also to the rest of your workforce.

Before I hand over to Johnathan, I just wanted to ask you for your views. If we were we do another webinar like this, what would you like us to cover in it? So perhaps you could just indicate. You're allowed to choose more than one of the options that are provided for you, and that would inform us of what we would do for you on a future occasion. So thank you very much for your feedback on that.

We're going to do a Q&A at the end, but that's my email address if you wanted to email me either during this webinar or also afterwards. I have seen quite a few questions coming in, so I'll attend to those as I pass over to Johnathan and leave it in Johnathan's good hands. Thank you. Johnathan?

Johnathan: Thanks, Barry. Hi, everyone. Now we're going to take a look at five AI tools that are reshaping the payroll side and how you can use them after today.

So, at number five, I have Claude, which is an advanced AI assistant designed for analysing large amounts of information, writing content, summarising policies, and accelerate software development using natural language.

I just want to highlight at this point that AI is here to help us. It's not here to replace us. The human element is always very important. And our company uses Claude to enhance the development of our payroll software. So our developers will use it to check their code, test payroll calculations, debug any issues, and help with generating APIs.

UK payroll software does change with HMRC updates, so changes in tax legislation, pensions, statutory payments, compliance requirements. Claude can be used to interpretate all that legislation and documents, summarise any changes, and convert the requirements into technical specifications, which significantly speeds up payroll software delivery.

So, for Claude and application to payroll, it can help developers write and test payroll software code faster, assist with interpreting the payroll legislation and converting it into logic, review any payroll compliance logic, speed up debugging, and it can generate technical documentation automatically. So your key benefit there is reducing the payroll software development time and also improving the payroll accuracy.

At number four, I have Scribe, and Scribe is an AI tool that automatically turns workplace processes into professional step-by-step guides and SOP documents. So its features include training documentation and SOPs, creating those, project implementation, onboarding new starts.

And how it works is it's as simple as just starting the Scribe recorder. Scribe will capture every click and action that you do, and the AI will automatically create a guide with screenshots and instructions.

The basic level of Scribe is free to use. You would have to pay for a subscription if you want your own company branding on that.

But why does that matter in payroll? In payroll, processes can be repetitive, compliance-heavy, and dependent on experienced staff knowledge. So Scribe helps payroll teams document these processes and keeps them consistent.

So, for our application, it increases efficiencies through automated payroll run procedures and creating SOPs. It can produce pension processing guides instantly by just performing the action, training new payroll administrators, and improving audit readiness with instant compliance documentation creation. Your key benefit there is just transforming manual process documentation into automated knowledge-sharing within your payroll department.

At number three, I've got Articulate 360, and this is an AI-powered eLearning and training content creation platform. It helps organisations to rapidly create online training courses, programmes, guides. You can create payroll quizzes and compliance learning materials.

So the AI assistant can generate the course content, summarise the learning objectives, and create interactive training materials within minutes. You can upload your source material or enter prompts into it, and ask the AI to then generate the course based off this. You can then come along and edit and refine your course before publishing the interactive eLearning module.

Because of that, then, payroll teams frequently need training on HMRC updates with the changes in legislation, pensions, statutory payments, and compliance processes. So Articulate AI can create those learning content within minutes instead of sitting manually to create them, which could take days of work.

So it's improving your knowledge-sharing. Payroll knowledge is often held by experienced payroll professionals, so the AI-generated training can help organisations to standardise their processes and then reduce dependency on key payroll professionals for that information.

The interactive side of it is also fun. You're able to create the interactive courses, scenario-based learning, compliance assessments, quizzes, and also create videos as well.

At number two, something you'll probably be more familiar with, is Copilot. So Microsoft Copilot is Microsoft's AI assistant which is integrated into their Microsoft 365 applications. That includes Word, Excel, PowerPoint, Teams, and Outlook.

An example of how it can be used within those packages, I have taken a standard pre-report from our payroll software that breaks down a monthly payroll for April 2026. So you can see on that report it's got all the employees, their gross payments, statutory and PAYE payments, and their associated cost centre branches and departments.

I've included a video demonstration. So what I've done is I've taken that pre-report, and I've used the Microsoft Copilot function and asked it to analyse the data within that worksheet.

So if you just click on the wee Copilot icon, it gives you little hints of what Copilot can do. I've just clicked "Analyse Data", so what it's doing now is it's summarising and creating a report.

Once it's done analysing, the report layout is created, and it's now just finalising the formulas that are needed for the report based on the data.

And then once Copilot has finished creating the report, it will give me a summary of all the actions that it has taken in the Copilot column on the right-hand side.

If we take a look at the report that it created, I've got key metrics, I've got departmental and branch breakdowns, a summary of insights. So I can see that Operations is the largest department, whereas Ballymoney is the biggest branch in terms of costs. And it also created charts to illustrate the findings from the report.

That took within two minutes to basically summarise all that data that was on that report. So it's huge time savings and efficiencies in analysing your payroll data by using this free Microsoft tool.

Another example, I have another video here, is using Copilot within Microsoft Word. And this time, I asked Copilot, "I am a payroll manager, and I need to draft a letter template to be provided to employees informing them that we are currently in a re-enrolment period for UK pension auto-enrolment".

Copilot's now creating the letter. You can see on the panel it says, "Describe what you would like to edit". Once it produces the final letter, you can put in your own changes that you want. You can attach your company branding and incorporate that into the letter.

Again, a letter template for re-enrolment has been created in under a minute compared to the time that would be spent to actually manually create the letter from scratch.

So, for application, it can analyse payroll data and identify employees with unusual payments and variances in payments, create reports and templates instantly, identify any anomalies or trends within your payroll data. You can ask it for a summary of insights to provide to management, forecast payroll trends, and as you saw in the videos, automate repetitive tasks.

And at number one . . . Barry, I think we're on to the same thing with these ones. So at number one, I've got OpenAI, which many will recognise their product name of ChatGPT.

OpenAI can be used in payroll to explain tax, NI, and pension calculations, drafting responses to employee queries, creating payroll policies and frequently asked questions, summarising HMRC updates, and assisting support teams.

But just as Barry said earlier, too, it's important to consider the information you're putting into these AI tools because you don't want to be putting any personal information. And so just take that into consideration.

So the key benefit for this one is it acts like an intelligent payroll assistant that is available 24/7.

And speaking of payroll assistant, one of the ways that PayEscape has adapted OpenAI into our proprietary payroll software is connecting it to our self-developed payslip assistant called PayAssist. So PayAssist feature is available to both system payroll administrators and their employees within their employee self-service portal. With PayAssist, payroll admins and employees get instant answers to all their pay-related questions directly powered by the calculations that are produced within the system and OpenAI.

All personal information is scrubbed before it's sent to OpenAI. So it's literally just the calculations and the payroll legislation that's been built into it.

The first picture shows the payroll admin view. So I've asked it why no pension deductions were calculated for an employee, and the Payslip Assistant has responded to say they're a non-eligible jobholder due to their age being below their auto-enrolment age of 22.

And the second and third pictures show the employee's view in their ESS app. So the employee has asked how their tax has been calculated, and it literally gives them a breakdown of that for the employee to understand.

The payroll administrator is not getting questions fired at them for payroll because the employees can ask them themselves. They can then spend more time on doing other payroll activities.

I'll drop a link into the chat if you would like to find out more information about the PayAssist feature and any other automated features that are available on the payroll software, PayEscape. So just click on the link provided to book a demo with a member of our team. And now we're going to take a look at the poll results.

Barry:  So do you suspect or know that employees in your organisations are using GenAI tools without authorisation or when prohibited? And 55% said yes and 45% said no. That's interesting.

It does allow me, I think, just to link into one question that we have that came to me by email as Johnathan was speaking there. And the question was simply this: Can LLMs train on the information that you put into the prompt bar? That was something that Johnathan touched on. And the answer to that is yes. What you need to do is to turn off the training data permission to make sure that that doesn't actually happen.

And if you don't know how to do that in ChatGPT or in Claude or any of the other LLMs, I would encourage you just to have a look or to ask somebody who can show you how to do that. From a data security point of view, it really is essential for you to do that before you start using your preferred LLM.

I don't see any other questions in, so that's great, because we promised half an hour and it's 11:30. So I just want to say thank you to Johnathan very much for his presentation and his five on payroll. That's me from Legal-Island, Barry Phillips at Legal-Island, signing off and thanking you very much for your time.

I really hope that you found that last half an hour useful and that you came away with some good tips to help you improve your work in HR or your work in payroll.

We look forward to seeing you again soon. Thank you very much. Bye-bye.

Johnathan: Thanks, everyone.