How to Upskill Your Workforce for AI in 2025
In this episode of the HR Leaders Podcast, we are joined by Jenny Griffiths MBE, VP of AI Innovation at Oracle, and Alejandro Modarelli, Partner at KPMG, to explore how AI is transforming HR and driving workforce innovation.
Jenny and Alejandro share insights on the adoption of AI in organizations, the importance of upskilling employees, and how to start small with AI use cases to achieve big results. They also discuss the role of HR in managing AI-driven workforce transformations, the rise of agents in automating repetitive tasks, and creating human-centric AI systems.
🎓 In this episode, Jenny and Alejandro discuss:
How AI agents are changing workflows and boosting productivity.
Why upskilling employees is critical for embracing AI-driven change.
The role of HR in workforce transformation and governance of AI systems.
Why starting small with focused AI use cases leads to successful adoption.
How AI is transforming HR by automating repetitive tasks and mapping skills.
Achieve efficiencies, boost effectiveness and enhance employee experience with AI!
Equip your prospective and current workforce with AI-powered tools to reduce time spent on straightforward tasks, and leverage AI to generate insights for strategic decision making on people practices and measuring the value of HR.
AI can help with a range of HR activity from creating personalized development plans and suggesting career options to answering queries on benefits and summarizing 360 feedback.
With a single user experience and secure data model, seamless HR processes and an AI-embedded infrastructure, Oracle Cloud HCM can help you redefine what best feels like for your people.
Chris Rainey 0:00
Alejandro, Jenny, welcome to the show. How are you both?
Jenny Griffiths 0:04
Very well. Thank you. How are you I'm good. That's
Chris Rainey 0:06
the moment when where you both go? Who says yeah, says hello first. After all the time we've been doing virtual no one solved that. Yet, when you say hi to everyone on the call, everyone's like, do I say something? Do I go first before we jump in? Because we've got so many great things to talk about. Could you both just give everyone a little bit of a background about your journey to where we are now, and your and your roles and then, and then we'll jump in. Jenny, you want to kick us off?
Jenny Griffiths 0:34
Sure. So I joined Oracle two years ago as their VP of data science, and I'm now their VP of AI innovation. So that's been a really exciting journey, and that's all due to the advent of generative AI kind of coming our way, and working out where the kind of touch points are, and user journeys where we can embed it, where we want that to go. And how I got into AI was I started an artificial intelligence company in 2012 Wow. So relatively early in the kind of AI commercialization story, I guess. So I ran a computer vision company for 12 years, working with the fashion and legal industries. And, yeah, that was my journey to get to Oracle today.
Chris Rainey 1:13
Cool, no pressure, right? In this day and age, being VP of AI innovation,
Jenny Griffiths 1:17
yeah, it's a title,
Chris Rainey 1:20
right? Yeah, people, everyone's like, well, everyone wants to have a chat of you. I'm
Jenny Griffiths 1:25
definitely more interesting than when I was a VP of data science. That's an
Chris Rainey 1:29
upgrade there. I'm sure your LinkedIn requests have gone up dramatically. Andrea, over to you, my friend.
Alejandro Modarelli 1:36
Yeah, hello, everyone. I'm Alejandro Modarelli, partner at KPMG, I work on the people consulting team. So what we do is everything from strategy to HR implementation. So technology and my journey, you know, start, I'm from Argentina, you know. And I started at Santander Bank, you know, I implemented for them a global HR system. And then from there that took me to travel all around the world running eight different HR transformation projects across Germany, Spain, the US and now here in the UK for the last 10 years.
Chris Rainey 2:15
Yeah, you forgot to mention the most important thing is that we're fellow skateboarders. I thought that
Alejandro Modarelli 2:20
was awesome. So
Chris Rainey 2:23
that's something we discovered on our last call, right? Do we both have a passion for skateboarding? And you one up me and said, you have a skateboard ramp in the house, so,
Alejandro Modarelli 2:32
and we are ready, you know, to to use it, right? Or we can go to skater ham. Yeah,
Chris Rainey 2:37
I'm with you. We're gonna do an episode everyone listening, where we skateboard on a skate ramp and we do an interview, I promise I'm gonna get practicing. Yeah, we don't injure ourselves on camera, although be good content, safety,
Alejandro Modarelli 2:53
health and safety. I'm really excited
Chris Rainey 2:55
for this conversation. And one of the cool things is that both of you in your roles. You're interacting with leaders and organizations from different industries, backgrounds, different stages of their journey. So I wanted to start off by talking about, what are you seeing is sort of the level of maturity of organizations adopting AI, and we can use that as a good starting point to jump in. Jenny, love to jump over to you for this.
Jenny Griffiths 3:19
Yeah, sure. It's really interesting time at the moment, because I'd say the interest in AI has never been higher, at least in my opinion. And I think the reason for that is that generative AI has kind of hit the market in such a form that everyone can interact with it, and everyone can talk with it. So it's this very kind of human way of interacting with artificial intelligence we've never seen before. And that means that people's adoption cycles have accelerated hugely. So this is a market more evolved than I've ever seen before. It's an interesting one in the sense of because it's easy to interact with the systems. We've got people trying to build systems that haven't done that in the past, and there's been some research into that around you know, how successful are people going to be in building things? So apparently, 75% of people who are going to try and build AI agents are going to fail with my previous startup founder hat on. That feels very low to me, 95% failure. But you know, when you kind of look behind the scenes, you're like, okay, that feels about right, because not all these companies have data science expertise, software engineering expertise, you know, even legislation, compliance with the EU, AI act, etc, etc. So I think the appetite is really, really high. People are looking at building it themselves. It's easy to get 80% of the way there with off the shelf tools. And where we're focused is really kind of helping people reach that extra 20% through, you know, OCI, or actually, by taking on all of that journey for them through embedded AI and just getting them the tools out of the box.
Chris Rainey 4:49
Love that. Andrea,
Alejandro Modarelli 4:51
yeah, what we see in the market, you know, is that we have different levels of maturities within the different organizations. About two thirds of the con. Companies have done almost nothing, you know, and they are now having fear about the implementation of AI, and they are looking at what would be the government, governance and what would be the frameworks and the regulations before taking any action. Then in the middle, there is a group of around 20% of organizations that have some limited experience using AI, so they have both licenses, and they are starting to deploy, but they don't really know how to do it, you know, probably align to what Jen says you need multi disciplinary team in order to do it. And then finally, there is a leading agent, you know, group, you know, about 10% of the companies that have developed a strategic approach to AI, and they're starting to deploy it more broadly across the organization, right? So if we think about the first group, you know what they need to do is educate the employees and looking at how develop, how to develop this governance framework for the second group that started to experiment, I think probably what they need to look is to to adopt an agile test and learn approach, and start to capturing lessons learned and deploying into new new areas of of their companies. And then for those organizations that are really mature, I think it's mainly about looking now at what was the impact of their investment and determine what the focus of the next wave of AI deployments would be within their business, right? So that's that's what I see.
Chris Rainey 6:34
It's super interesting. And your point, Jenny, about people already having a playground to play around with the tools we didn't have that in the past. Definitely not right. So like the fact that everyone can have access to open AI or Gemini, or, you know, just new ones popping up every day. Now, whether you're using it all as an organization or not, your employees are they're already, they're already using it as well. And that's, and I think that's maybe some companies are in denial about that as well, but it's the first time I could think of a significant technology innovation where the general public have, have have access. Yeah, to that. How does that shift the way we think about how we adopt this in a company when we already know that our employees are playing around with these tools?
Jenny Griffiths 7:18
Yeah, it's a really interesting question. It's a bit of a blessing and a curse. So I'd say on the plus side, you know, it's, it's wonderful to be in a room with people who are genuinely excited about technology and where it's going. And, you know, I think that everyone having access to these kind of tools has removed some of the mystery around AI, which was always a bit of a barrier to adoption in the past. It always felt very inaccessible, a bit like a black box. You know, what's actually going on here? How is it going to affect me? And the fact now that these tools are so readily accessible, everyone can play with them in the day to day lives and see the benefits in their day to day lives is making that conversation a lot easier and a lot more open, which I only think could be a good thing. The curse is this kind of the gap between using AI in your personal life and then enterprise AI and the guard rails that you need for enter enterprise AI is pretty broad, and that's not particularly well understood at the moment. So I'd say that's probably the downside of the accessibility at the moment, is, you know, yes, you can use this personally, but to use it at scale across global organizations, with guard railing, making sure you're guarding against things like bias. You know, that's a huge conversation. That's a huge development effort, and sometimes it's quite hard to kind of quantify that and get that scene when it's just so easy to fire up on your laptop and and have a play around.
Chris Rainey 8:41
Jump in Alejandro that way. For me, I can see your brain. I can see your brain going.
Alejandro Modarelli 8:49
I can only agree with Jane, you know, it's a it's a very, yeah, it's not something you know that is easy to deploy enterprise. Why? Of course, you know, I know of organizations that are really giving the tools to the employees to experiment, right? But that is, yeah, within the governance, how those tools, you know can be, you know, limited to a certain team and then expanded it more broadly to more wider set of employees to use it, but those companies are on the are the exception, not really the norm, right? The norm is everybody is much more worried about the regulation and being very careful and how they deploy these tools.
Chris Rainey 9:38
Yeah, and I feel like that's now like your point earlier, companies are starting to set up those AI ethics councils starting to create more decision making frameworks about how they do this and be more thoughtful for the points that you made, Jenny as well. But that's such a new, a new you know, this is, there is no case studies out there for. This is kind of unexplored territory in many cases, you know, who are the decision makers that even need to be in the room? Is a question that I hear from HR leaders as well, to even make that happen. Jenny, given the fact you've done this as a, you know, a startup founder now, Oracle, what advice would you give to HR leaders, listening? Of like, where to start? Because that's kind of one of the biggest challenges for
Jenny Griffiths 10:22
them. Yeah, I'd say this sounds like very strange advice for work, but I'd say start at home and just use these tools as much as you possibly can and really get familiar with these tools. And the reason I say start at home is that if you're able to experiment with AI in bits of your life that are incredibly low risk. So we talked about, you know, enterprise. You know, start on something small, but start on something that you know incredibly well, and that will get you familiar with how you should write prompts, how you should be interacting with these systems, how you can get the most out of them, and spotting hallucinations so effectively when the model goes wrong and it goes off piste, you can catch it, if it's your area of expertise. So you know, Chris and Alejandro, you could do a Gen AI for skateboarding, and you would know if it was right or wrong straight away. So yeah, I'd say, when it comes to kind of knowledge and up skilling starts at home, really kind of hone in on how to use these tools properly. And then I'd say bringing that into the workplace, I would start relatively small and identify like, where are the pain points in either your jobs or your team's job, where you're you think that AI could have real value, and where your team will be really excited to embrace it. So what are the really repetitive tasks that your team are doing again and again that is going an absolute time sink that no one really gets any joy from. That's always a really great plus place to start innovating, because it instantly kind of wins the hearts and minds of your team. They really want AI to help them out there. And you can easily measure benefits and uplift and ROI and all of those things if you're looking at quite a constrained task that you know is a bit of a time sink at the moment,
Alejandro Modarelli 12:01
yeah. And the way, the way that we see it, yeah, it's exactly, you know, a three step approach, you know, you first as Okay, as HR, or as a business, you can identify, what are the pockets of opportunities that you have within your business to implement some AI use case, right? And then our recommendation is, then, you know, you give these tools to a limited set of employees that they will start using it and then validating that this works for them. Also rewrite, you know, with their job role description would be, you know, in terms of, I'm not doing these tasks, but you know, now I'm going to start doing these value added tasks that I am not doing, you know, have not been doing before. Then expand within that part of the organization, that specific use case, rewrite, you know, what the workforce would look like in the future for that area, and restart again, right? And move into another area. That's what we see. It's been successful for different companies in order to deploy in AI solutions? Yeah,
Chris Rainey 13:03
it's so important. Because the challenge, and even me personally, is there's this we're being inundated with new technologies, new shiny objects every day, and it's easy to get caught up and start chasing different things, right? Especially in HR, every single part of HR is being disrupted, every single every single layer and now, and there's a company out there saying, Hey, we're going to solve all of your problems with AI, but to your point, starting with what a problem we're solving, and focus on those specific challenges to the business, and just focus there, and then go for a small user group, you know, validate there is, you know, way more valuable, even if it's a small, quick win, you know, it could be an increase in productivity in a specific area that is the focus of, as opposed to otherwise, you're just going to get lost in everything going on as well.
Jenny Griffiths 13:52
Yeah, it's really true. I'd say Chris as well. When you're looking at experimenting, if you run one experiment and it goes, well, look at innovating, you know, in a kind of linear way as well, and building upon that progress. And the reason that I say that is we've got a tool called time to hire, and it's basically reduces the time between finding your right candidate and hiring them by 10% but then we realized we ran a kind of study with some customers, and if you actually changed some AI experience. So I think of his time tire and job descriptions automation, and that increased, sorry, reduced the time to hire by 20% you're like, Ah, okay. So this is where I'm getting excited by agents as well. Is that if you actually begin to change some of these technologies, you see a win in one section. Then you're like, right? What's the automatic extension of that, then actually you're compounding these benefits, rather than trying to look in kind of hiring and then over here and over here. You know, if you actually take a natural progression, you can see the benefit of chaining these experiences a bit like if you were to introduce a team, right of people? It's, How would you organize them? It's kind of good. Have that level of strategy when it comes to testing out. Ai, yeah,
Chris Rainey 15:04
talk a bit more about I think a lot of people have heard the word agent. It's the new buzz word. What does that mean? Just kind of break that down for people listening the types of agents. What are they? What is that? Yeah,
Jenny Griffiths 15:19
that's completely fair. I think it's going to be the word of 2025, I'm calling it now in January. But essentially agents, or agentic workflows. They're essentially AI workers that will go and do tasks for you. And it's taking it one step before generative AI, as we've seen it yet. So generative AI, where it started, was all around kind of generation and creation of content. Hence the name. So things like sum tasks, I don't know, writing cover letters for CV, that kind of content generation. And now agentic workflows are essentially chaining these workers together so that they can create content, but then they can also use tools and they can take actions for you. We're seeing early agents basically go away suggest actions, which I think is super powerful. So it's basically having an intern go away for you and create a kind of menu of options for you to pick from and take the right, right actions from there. But I think where agents are going to go is that we'll be able to automate whole work streams, and people are going to be able to take actions on your behalf. You can have reasoning agents, where they can talk to each other and work out if an answer is good or not. So yeah, it's going to be essentially chaining these generative AI workflows together into action, taking agents.
Chris Rainey 16:33
Wow, it's hard to get your head around honestly, and we're really just at the beginning. That's the interesting part of this Ale, and add anything to that?
Alejandro Modarelli 16:43
No, I think it cannot be better explained, you know. I only wanted to add that. Yeah, I only wanted to add that, you know. Like she said, now it's an intern, but somebody said, but by year end they would be PhDs, yeah, that's how fast these agents will be changing our way of working.
Chris Rainey 17:01
Yeah, I'm already seeing it. We've created in Atlas, our bi directional agents as well that are now conferring. So when you ask a question, it kind of goes through the different agents, and then it kind of chooses which agent can best then take on that task. And that just blows my mind, that we have the capability to do that, and it's all doing it within seconds, right? And something that would take hours, weeks, months in the past to achieve that we're doing now as well. So what is the role of HR in all of this? So when we talk about AI transformation and the journey that we're on, what do you see the main role of HR. What role is HR playing in this AI, transformation?
Alejandro Modarelli 17:47
Okay, let me, should I pick that one? Yeah. So, you know, I think HR will have two typical roles in this transformation journey. One is around guiding the wide workforce transformation of their organization. So HR should look at the workforce plan so re skill and Up skill employees look at new roles and reshape how that talent will look in the future, so that they can really plan where they want to get in two or three years time. So that means working really closely with the business leaders, identifying what are those AI opportunities in their own space, and determine what those future roles would be and what are the skills that they will need. And then also the role of HR in this area of work for transformation would be the change agents. So they will need to be in the middle, right? And I think, you know, related to what Jen said about learning step by step, the other role that HR will have is about transforming themselves, right? So they also need to look at how they will professionalize the HR functions through the use of AI. And how can they leverage, you know, these agents to take the most basic tax tasks and really look at doing the most value added tasks, you know, for, for HR, right? So, you know, so, yeah, basically fulfilling those two critical roles are the two things that HR role is in the in the in 2025, let's say,
Chris Rainey 19:27
Do you think that it's going to be these? I can also imagine, like a new role in HR of someone in the team who sort of manages agents,
Jenny Griffiths 19:36
an interesting concept. I'm just trying to
Chris Rainey 19:39
think right now because that maybe that's not made any sense at all, but I just thinking like, because who's going to be managing these different agents? I was speaking to a Sutro, uh, yesterday. I can't, can't mention their names. I might get in trouble. But they were saying to me, Chris, you know I can. I've got, right now across my HR tech stack, I can turn on seven agents. I. Every vendor that I have across all these areas are saying, use my agent right, right now. So now you're gonna have an agent for all these different parts of the of the business, who's managing that?
Jenny Griffiths 20:11
I mean, that's a great question, Chris, I think there will have to be a lot of kind of testing done up front, around, like any software buying process, right of competitive analysis, what's working well, what's not, whether it should be turned on. So I think that will still probably be a really essential part of, I'm guessing, kind of procurement cycles or it. I think the role of HR in all of this is fascinating with regards to working out how people's roles change, and how the kind of forms of those roles change around these agents that are being introduced. So as an example, I used to be VP of data science, and when we were hiring data scientists, we had to kind of amend that job description incredibly quickly from a typical skill set to actually talking about prompt engineering and LLM evacuation and LLM as a judge. And that was changing like lightning speed. And I think for HR to keep on top of those role changes, and I like to think AI is going to bring more humanity into our roles as it automates the things we don't enjoy doing. And, you know, it increases our ability to do a job in a more human way. I think that adapting those job roles around the agents that are coming in is going to be, you know, a full time job, hopefully managing the agents themselves should be simple. If you got a good one going, then, yeah, they should be relatively self managing.
Chris Rainey 21:34
It's going to be interesting now that we're going to have prompt engineers in the HR team. If you told me that if you told me that 20 years ago, obviously I won't even know what prompted agent meant 20 years ago. Actually, no one even what does that even mean, but it's incredible to think now that that's going to be a key skill for everyone, not just HR for everyone, but especially in HR, where you want to make sure that you have AI, but that is human centric still. So there still needs to be a human in the loop somewhere, otherwise we're going to get in
Jenny Griffiths 22:08
trouble. Yeah, we need to democratize prompt engineering as well. It's, you know, speaking as a data scientist and software engineer, it's so tempting just to be like, you know, I want this to be the thing that only I can do, and I'm an expert in but it's why it's so important, that we just get everyone kind of learning about how it's actually working, what's happening under the hood, how you can get the most out of these systems and basically teach people to talk to AI in the way that, you know, we learn how the correct way to communicate with our peers at work, we've got to build AI into those systems as well.
Chris Rainey 22:40
Yeah, I was making a joke to my wife the other day, saying, I feel like in the future in schools, when my daughter's six, when she grows up, that will be a class so you have an English, Math, Science prompt, engineering like I do think that that will be just as important as a new type of literacy. That you have to know, if that makes sense, the same way we're seeing what's interesting, because we're seeing coding come into the into the schools, but now that's being disrupted by AI, because you can do that there as well. I was listening to Mark Zuckerberg on a podcast the other day saying by the end of the year that he believes that the agents will be able to do 80% of the work of his engineering team. And I was like, whoa. Like, that is pretty crazy to think about as well. So what are you kind of hearing from customers? Are you seeing customers already make investment in their HR teams to upskill them to ensure that they're prepared? Or you feel like they're really behind on that?
Alejandro Modarelli 23:39
I think they're starting to experiment, you know. Okay, I'm going, you know, back to the question that you asked before. I think what we see today is probably not one master of all agents, but really different owners, right? And that has to do with what you explained before about I have seven agents, so it's the fragmentation of the HR, IT landscape where they would have someone using a recruitment solution that has its own agents, its own data, and then you would have someone that is being now trained in the HR department in order to develop that agent or AI solution that is specific to that product. And then another one for the next product, you know. And as we know, in average, companies have five different IT products to run HR. So I think there will be multiple HR agents. And I think what we see now is, yeah, people are starting to get trained on all these different capabilities in order to start using them and deploying
Chris Rainey 24:39
when you're chatting with clients, what's sort of the best practice approach that you your guide as much as you can, as opposed to them in terms of workforce transformation, what's the approach that you take? I'm sure you have a very clear, defined,
Alejandro Modarelli 24:55
Oh, yeah. Well, okay, yeah. We have a very clear, defined approach. You know, it's in line with. What I explained before you know, what we typically do is, okay, we use AI to what I'm going to explain right now. So what we typically do is look at the whole organization. So we look at it end to end. We load into an AI tool all their job descriptions, their skills, the tasks. And then, using AI, and looking at the AI capabilities such as translation or summarizing information, analysis of data, etc, etc, we identify, what are those tasks that could be augmented by AI, right? With that, we find what are the departments that can really benefit more, and specifically, what are the roles? And from there, it's very easy to say, okay, these are the use cases, right? So a good example, you know, we ran for one of the NHS Trust, one of these exercises and suppi, surprisingly, you know, a role that you would never think the midwife could be heavily affected by AI, really. So, because, yeah, because, you know, the midwife needs to summarize information for each one of the parents. You know, when they go out, you know, they also need to schedule things, etc, etc, there's a lot of opportunity within the midwife population to automate so that they can spend more time with their patients, right? So these are the surprising findings that you will find. You know, if you do an end to end analysis of your organization, and then what we what we recommend is, after you've done that, you know, you identify the use cases and you start deploying them organically through the organization. And finally, when you see what the impact is, rethink what your workforce would look like, right? And and start again, right?
Chris Rainey 26:48
Yes, fascinating example that you just shared there. And I remember when my wife was pregnant and we had the midwife meetings, and I remember asking, because she was writing down all of the notes. And I was like, why don't you just write in the computer? Because, and she's like, No, we were required to do both. And I and I remember being like, saying that that must take you so much time. Like, you have to do it twice. Like, because they have the folders of anyone who's having a kid, they give you, like, this big folder with all of your information about your child. I think it's red, red. And then they also have to enter all of that information manually after into the computer as well. So I can imagine the time savings on that. And to your point, in order for us to be more human, you can have much better interaction with with the with the parents, if you had an AI note taker, for example, so you could be truly present in the room as well. And I'm sure it would also decrease errors, because there's human error in remembering what that person's told you, right as well. Wow. What an interest and that. And that's NHS is the largest employer in the UK as well. So you can imagine the impact called example. Yeah, Jenny, any examples that come to mind for you?
Jenny Griffiths 28:07
No, I was just thinking, Chris, about the comment that you made then about performing better. And the way I always think about AI, and the way I'd love it to evolve is that we all know our strengths and weaknesses. And an example that springs to mind for me is I have a terrible short term memory, which I'm not sure is something I should be proudly
Chris Rainey 28:25
by the way, I'm with you on that. All right. So bad, tiniest
Jenny Griffiths 28:29
bit of stress and my memory is just completely shot, so I become a sieve. So something that I've been using recently is using AI to help with my kind of employee summaries at the end of the year, when I'm doing performance reviews, because my fear as a manager is that I'm going to judge someone based on the last week, two weeks, month, you know, what's been happening recently? But obviously, reviews are meant to go back a year, so I've been consciously kind of taking notes just as and when things happen around, you know, training courses they asked me to do, or, you know, a gap I noticed in a meeting, something they did really well. And then actually, I can feed all of that into AI, and it gives me a lovely summary of how I felt about that employee over 12 months, rather than the last week or two. So that's personally where I found it the most useful is to super interesting. Yeah, yeah. I'm hoping that I can actually increase my memory, rather than just relying on AI in the future. Yes,
Chris Rainey 29:22
I think with those things, I'm the same. By the way, my team's like, they hate me for it. They're like, Chris, we said this to you. My short term memory is really bad, but I'll remember a name from 10 years ago. It's really weird. Oh, yeah. So they're like, this makes no sense whatsoever. But what you're doing there is also, is like, kind of reducing bias Exactly, right? Yes, you have that short term. As you said, it could be like the last few weeks versus the incredible work day maybe did for the first six months of the year as well. And you're judging people based on that as well. And no human can remember all of those things. Yeah, exactly how good your memory? Yeah.
Jenny Griffiths 29:59
Its memory and its emotions as well. Like you say, with bias, you know, if you're particularly stressed, if you're in a bad head space, you don't want that influencing a whole year of reviews. So yeah, I found that really interesting to be using myself.
Chris Rainey 30:12
Yeah, I feel like managers could use all the help when they can get when it comes to that as well. And you mentioned earlier about like aI being able to nudge you and make suggestions and kind of pick up those things that you wouldn't Hey, Chris, you've been on 14 hours of zoom calls this week. Maybe you should cut that down a little bit, because it has the data right to be able to do that as well. Or, Hey, you haven't checked I've got a tool that I use that kind of reminds me to check in, because I have a remote team, and it just now and again, if I haven't checked in with a particular employee, it will give me a little nudge to say, hey, you haven't checked in with Adam. Maybe you want to schedule a call about, yeah, I haven't spoke to Adam in a week. I probably should check it and see how he is. It's like, because we're all just so busy. So that's kind of and that's a small thing, but it makes the world of difference because that person feels like they're valued. I've just not just forgot about doing because they're working remote, because they're remote, because there is that bias that I spend more time with the people in the office, that's a real thing as well, and it's very because they're present right in front of me. So of course, I'm going to check in and recognize them more than I recognize people remotely as well. So super interesting. I'd love to learn more about sort of KPMG point of view on AI architecture going deep now. So because that, especially if this sort of rapidly evolving landscape, is something that we need to pay attention to that many people, honestly, it's kind of above going over their head. Let's say it that way.
Alejandro Modarelli 31:46
Yeah, I think it's an evolution, you know, from from the previous architecture that we had, right? So we have an employee an experience level, right? And then, I would say an employee level, and then a core HR level, right? That when we log in every morning, you know, that's why we call the experience level they you will log into maybe your outlook, or maybe your teams, you know, and interact with your organization. Right? Then the second level was, yeah, well, you know the experience level that would be role in HR multiple different systems, right? So it could be that you have a service now integrating or an Oracle layer, integrating all your systems below, or the transactional systems that will allow you to query, interact and transaction, and then your transactional systems, right? So those are the three levels. Okay. Now what we see is the evolution of that where you know, the the top level, this experience level, where you log in every morning, will have its own agent that will guide you through. You know, what's new in our organization today, and how to be the agents of agents, and guide you through what you maybe you need to do today. And then it will go and contact the HR level, right? And the HR level, we have other agents that will also enable to think, oh, you know, it's a performance process this week. You know you need to be doing this and this and that, and reminding you and telling how the process would run, and then connecting to the transactional system as well. So that's how we see interacting the different agents interacting going forward.
Chris Rainey 33:37
How about you? Jenny,
Jenny Griffiths 33:40
yeah. I mean, I think from a kind of technical architecture perspective, Oracle is really interesting. And I say this, I know I'm going to sound really boring when I'm talking about this. I'm well aware, but I think, you know, we're interestingly positioned because we've got hosting via OCI, so Oracle Cloud infrastructure where, you know, GPUs, there's a real shortage of them globally at the moment. So it's great that we've kind of got all of them in the background, the NVIDIA h1 100, say, one hundreds. But basically it's like the firepower that you need to be able to run these large language models. They're huge and very complex. So you need that we've then got the area, which is around data and security. And it's worth saying actually, when it comes to customer data, we don't mix company customer data. We don't re share it with models to be trained. You know, very few companies can actually say that at the moment, that they are keeping customer data as customer data, and they're keeping it securely. And then you've got this apps part of the business, which is where I where I am, where we're looking at those kind of points in user journeys where AI will add value into user experiences. So can we identify places that we think AI will add the most value? What are we seeing from other customers, like, where are they using AI the most? So I think having those kind of three pillars together of the hosting infrastructure, the data and. Security side, and then the insights around applications and being able to build them into applications that makes it a really interesting place to be. So yeah, very much. Enjoying working on AI within a company that does all three of those things incredibly well. Yeah,
Chris Rainey 35:15
and you need all of her. Those are the essentials, just for the foundation, exactly to make it work. Share some of the use cases. Jenny, what are some of the Gen AI innovations, oracles, focusing on what lead someone use cases? Share that. It'd be super interesting to understand
Jenny Griffiths 35:31
more. Yeah, definitely. So one of them that I really enjoy is we've got a benefits agent that's just launched, or it's launching soon, but I'm allowed to talk about it. That's essentially around using rags. So retrieval, augmented generation, is a kind of functionality within generative AI that lets customers ground the response and the reality of their company, but without sharing that data back to the model provider. By the way,
Chris Rainey 36:00
making that sound so simple you describe that really simply, and I do such a terrible job when I try and explain that to
Jenny Griffiths 36:10
people. Oh man, I can geek out about this. Sorry, everyone who's listening. But essentially, you know, it means that if you went to chat GPT or Gemini or any of the providers at the moment, and you said, you know, are my son's braces covered in my dental plan? It would basically, in a very wordsmith way, tell you to go and read your dental plan. It will tell you very nicely, but that will be the answer. Whereas rag and the system either developed at Oracle, will let you ingest your company's data, so you'll have that data dental plan. You might have 12 dental plans across the world, and it will work out, okay, you're a UK employee. This is the plan you need. It will do search within that document, and then it will kind of give you that answer. And you can layer those questions, one on top of the other. So that kind of benefits, benefits analysts, it's coming soon, which will be really, really cool. Other things we've got going. We've got employee hiring advisors. So this is helping source candidates, looking at the optimal time to run campaigns, wording in campaigns, reducing time to hire. So you've got that kind of hiring piece of the puzzle. And then we've also got scheduling assistance as well. So instead of playing ping pong with people around, you know, when they're free for an interview and shift schedules is another big problem, you're able to kind of create those optimized workflows using scheduling assistance and tools as well. So I
Chris Rainey 37:34
need that one as soon as possible. How do I yeah, how does it show up? Is it within one sort of user interface and you switch between agents? Or do you not even know that agents are behind it? So you're asking your questions, and it obviously then sends that to the right agent, just out of curiosity, just to give it to people, to give a visual of like, what that looks like in a practical sense,
Jenny Griffiths 37:57
it's a super interesting question as well, Chris, because we had a lot of discussions about how we should present it to people, because you want things to be as seamless as possible, but equally, you want people to know they're using AI so they can opt or out of it. So it's quite an interesting one academically, and I can nerd out about that for ages as well. But essentially, you know, it depends on the product. So with Oracle grow, you'll see lots of these AI features embedded within the user experience anyway, okay, the way that you can spot, if it's we normally say AI assist, and we've got a kind of little wand icon with some stars, and essentially, so, for instance, if you're looking at writing a job description using Gen AI, you can, you could write a job description in line, or, equally, you can tap on the AI Assist button, and that will just automatically populate it, and then it's up for you to check it and press send. So yeah, we try and embed it where we can in the flow, so that you don't have to keep breaking your attention. But equally, we try and flag as well that this is AI that's being utilized in the background.
Chris Rainey 38:56
Yeah. And then for the benefits one, which is super interesting, because benefits is such a complex, yeah, it's, you know, like that. That area is right for innovation and to make that more personalized, because it can be overwhelming for people. I even though friends of mine that didn't even know they had certain benefits in their organization, right? So it's the ability to be able to ask a question, and it knows the context of Chris is in the UK. This is who he is, this is his role, this is his function, and this is what he has. Is that something now that's going to be employee facing, so employees can ask that Exactly,
Jenny Griffiths 39:29
and that's where a chat interface is really useful as well. So this kind of, I hate the word chat bots for you, conversation place is probably way nicer than saying the word chat bot. So yeah, it's, it's an interesting one. I always like to try and take a step back and work out, you know, where it fits in the user flow. And also, people have different styles of interacting with technologies. So you might love a chat bot, but I might love just clicking a button and getting it emailed to me or whatever. So yeah, trying to kind of thoughtfully design. Which products will fit which interface. But for benefits, it's a conversational interface.
Alejandro Modarelli 40:05
Yeah, I have seen, I have seen the benefits solution, you know, it really acts like an HR advisor, okay? It takes you to that limit, you know, things like, it would tell you things like, you know, last year you subscribed to these, I don't know, dental provider, but this year we changed, you know, be aware that the change brings these different benefits, nice so, and you also have a kid, you know, and you will need to consider this and this and it will guide you through the process, yeah. So
Chris Rainey 40:34
even being proactive as opposed to reactive as well, to say, Hey, I know you just had a kid, his new benefits. You might want to have a look
Alejandro Modarelli 40:43
at. Basically, when you go to the benefits application in Oracle, yeah, it will pop up. And we start to guide you through the process of selecting your benefits for the next year, and doing those comparisons and guiding you through, yeah, the changes, the changes in your lifestyle, what they think you know you should be doing, etc. Wow.
Is beyond what you've seen. Is it
Chris Rainey 41:08
so they still separate then interfaces? Is it or is are they within one interface of what we're describing? So do I still have to go to the benefit portable to activate the benefits agent? Do I have to then still go to a separate portal to activate the other agent, or is it in one interface? Does that accuracy?
Alejandro Modarelli 41:29
Okay? They win a the way I've seen the benefits one working is, you know, you go to your benefits enrollment, right? So you will go to the benefits enrollment and say, Okay, these are the benefits for this year. And on the right, you know, it will start to give you all this speech and say, Well, this is how you should select this year, and we'll give you the
Chris Rainey 41:47
recommendations. Yeah, I'll just ask you, because I've seen companies roll these out where the different agents are sitting in sort of a, I don't want to say word intranet, because I hate that word, but a portal, and where you click into different areas of the portal you have different agents to activate. I think one of the challenges with that is you're not it's not all in one simple user experience within one interface. If that makes sense, to then be able to it can then kind of send the request in the right direction as well. But it's incredible that that we even have the opportunity now to give our employees access to that. I'm sure the HR team is very happy. Probably reduced. I love to see the data on the reduced amount of requests and tickets that they're going to be they're going to be getting as well. So I know it's just launched, so be really excited to see how that evolves.
Jenny Griffiths 42:36
Yeah, we check back next year. I'm I'm interested as well. I'm not sure if we can ever get this data, but to see how many people take up more benefits based on this. So for instance, I was testing the software. Found out I could get free glasses. I was like, never see if people take up more benefits because of it.
Chris Rainey 42:54
Yeah, and that's the truth, right? Most people don't really know and check those things. And depending on what stage of your life you're at, those things change. Right? You could be whether it's getting a house, whether it's having a child like you know, maybe you're near retirement. Most people don't go and check back and see kind of what they have. So I love that. I feel like that's a great place to start, and an area really, that hasn't had much attention, in my opinion, over the years. What about the, I know you mentioned before, the Gen I innovator program. Could you talk a bit more about that?
Jenny Griffiths 43:30
Yeah, so we ran a program with a humanitarian organization, and they've got about 10,000 employees. So that was really interesting. And we basically said, like, these are all the innovations that are coming out. Some of these are available now. Some of these you're getting an early peak. Which ones would you like to test and deploy? So they prioritize features. They did goal creation for employees. They did feedback automation, summarization of performance reviews. So they were the kind of main three that they went for, oh, an about me, writing your about me page within the HR system as well. And we effectively measured with them productivity gains and how much people are interacting with that system. So that was a small kind of study that we did with one organization, but we're rolling that out as well. We've got a team now in Oracle who are looking to work with organizations and work out which features they want to turn on, because all of this is included within Oracle Fusion subscription. So to use generative AI, it doesn't cost anything more. There's not an additional fee, but we just want to raise awareness of that you can turn on these features and try and make it as easy as possible to get them switched on. Nice.
Chris Rainey 44:44
So if you're listening right now and you're using that turn it on, or you probably get some people listening, even though I had that
Jenny Griffiths 44:54
like me and my glasses, yeah,
Chris Rainey 44:56
exactly. And that's another part, most people aren't. So I'm glad that was. Are excited to do this episode and share more. Does anything Andrea, any other use cases that you can think of, that that you've come across, or stories that you could share, that that we've that we haven't spoke about.
Alejandro Modarelli 45:10
Well, I think, you know, another one that I've seen in within the Oracle space was the cascading of goals, I think is quite useful. You know, mainly if you really have, you know, a really, let's say, organizational cascade of roles, because it really can identify what of those, you know, company role, company goals are important, and what is the relevance for your role specifically, and what is the percentages that they will apply, etc. So you can imagine that, you know, it's again, unbiased, you know, very precise in terms of what the recommendation is about your goals and also time saving for you, writing your your your goals for that year. Yeah,
Chris Rainey 45:56
half of the battle is just getting writing it down and doing the work as well. So definitely will help, because I feel like we could do a whole series together on this. So we just sort of tip the tip of the iceberg. But before I let you both go, what would be your parting advice for HR leaders who are just sort of getting started on this journey, because, as we saw by the data, many are just getting started on the journey, and then we'll say goodbye. We'll start with you, Jenny, first.
Jenny Griffiths 46:23
So I'd say first up is speak to your HR software providers, see what they've got, see how much it is. For instance, you know, with Oracle Fusion, it is free, and you can try it. So I can really kind of sit down with people and work out what's being offered to you now, and how you can measure performance, and how you can kind of check that ROI. So the second one personally is just learn as much as you possibly can. So take AI into your personal life. Experiment with it. See what resonates for you. Think about where your gaps are, like my memory, and see where it would be the most useful. So just don't be afraid to play around with it. It's, you know, it's the time to experiment. And the more you know, the more useful it's going to be in the future. So, yeah, just get stuck in
Chris Rainey 47:09
amazing. Thanks, Jenny, 100
Alejandro Modarelli 47:11
I will add up, you know, to that. Yeah, it's a you know, the technology and what is available is important also, you know, look at your organization. Look at your workforce, find where you have benefits of implementing these tools and also help your your employees to learn and to change as you deploy these tools. I think that's the compliment to Jenny's
Chris Rainey 47:42
view, yeah. And I love, if anyone is using Oracle, definitely connect with Jenny and the team, because I love the fact that you're helping companies discover which use cases, because I think that's a challenge for many this great that you've launched that team to help organizations focus on which of those use cases that where they have opportunities, but appreciate you both coming on the show as well. It's what a time to be alive honestly, that we and we just getting started right on the journey. So thank you both for showing your insights, and I wish you both all the best until we next
Alejandro Modarelli 48:17
week. Thank you very much for having us. Thank you, Chris, yeah.
More from the HR Leaders Podcast
Jenny Griffiths, VP of AI Innovation at Oracle and Alejandro Modarelli, Partner at KPMG.