Making Workplace AI work for you.

The brain in front of the computer is you...

Join your hosts, Alex and Jason, and their guest Tomi Karafilov – expert in Workplace AI, as they discuss the value that AI can bring to your workplace.

Enjoy a lively discussion on the value and pitfalls of investing in technology that can either supercharge efficiency if implemented and managed correctly, or the opposite if not.

Understand how this technology can support and enable you – the brain in front of the computer…

Want to suggest a topic for discussion on a future episode?  Simply email Alex and Jason on o2i@softwareone.com. 

Transcript

Alex 0:02

Welcome to Optimize to Innovate, a show where we help organizations stop wasting money on things that don't add value to their businesses and understand the technologies that actually will. Join us as we share practical insights into the latest trends and innovations with industry experts across everything from software and FinOps to cloud, data and AI. We've got some great guests with us today, so let's jump into this week's episode.

Jason 0:26

Thanks, Alex, and in today's episode, we are going to be focusing on something called Workplace AI. So many of our listeners will be very familiar with generative AI or Gen AI, and they'll have used that through chatbots and other things like ChatGPT. What we're seeing, or what we're going to be talking about and unpacking today, is actually how organizations, technology vendors are bringing that Gen AI capability into the workplace, bringing it through the front door, not the back door, not people doing their own thing, bringing in their own subscriptions, but actually organized structured implementations with a very focused goal but very purpose-driven. So to lead us into an exploration of this, I'm really, really excited to have our first guest on the podcast and somebody who we know as having a lot of experience in this area, Tomislav Karafilov. Tommy, could you introduce yourselves and your background, please?

Tomi 1:23

Wonderful. Hello, thank you very much. Thanks for having me. Thanks to everyone who is listening. So my name, as you said, as you heard it already, Tomislav Karafilov. I'm working at Software One in the digital workplace advisory team. And yes, I like everything around technology. I'm interested in it. But um in the era of AI, I'm doing a lot of uh co-pilot uh projects, Microsoft 365 co-pilot projects, uh, co-pilot studio agents, so a lot of things branded with a specific word like co-pilot or like AI. But everything we are doing has to be in focus of people because people are using AI, and that is a good balance to have this always in mind when we are talking about workplace AI.

Jason 2:08

Tommy, thanks so much for that great intro. Just confirming to our listeners that you are the right person for us to have on this podcast talking about workplace AI. The first question I have is what do we mean by workplace AI? How is it different to other forms or types of AI that people talk about? Or is it actually just a category of hype? Because we know that there's there's a lot of hype around AI. There has been so much froth in this area for a year or more. What is workplace AI and why is it important for us to focus on this category?

Tomi 2:42

When you look at AI currently, you have maybe AI that is producing content for social media. Do I need this on my workplace when I'm a social media guy? Yes. But when I'm dealing in the banking sector or I'm uh working for something else, maybe it's not so important to create very fancy images. Or when I'm driving a car and I have AI in my car that says you have now to go left, right, I have a navigation, it's handling when someone is passing the street and so on. It's also interesting by sitting in my office, not so many people are passing the view to my monitor, so I don't need AI to restructure my setup. So looking at workplace AI means that you have to think about what is the workplace, who are the people, what is their job. So, what do they have to do in their daily business? What is their use case? What is the company doing, and then find for AI the right use case to implement this to make work for people at the digital workplace, yeah, what easier, faster, fancier, cost relative, better, or something else in front? Help people, that is the thing that I'm always saying.

Alex 3:56

We hear about there's a whole bunch of different partners that provide different workplace AI tools. So, common ones I've heard of, you know, we know of you mentioned Copilot already. There's things like Amazon Cube for business. Um, there's a whole lot of AI tools built into Google Workspace. Is that what it is? Is it really focused on that textual and PowerPoint type stuff, or is it actually wider than that?

Tomi 4:20

It's wider than that, and the access to AI is so easy. Let's remember 2023 ChatGPT becomes a little bit alive. We have a web interface, we can put in a prompt, we get fancy, really great results back, and now everyone is talking about AI, AI, AI. What is it? What is what can I use it? What do what kind of AIs do I have? And is it really important for the people to know that is AI? Currently, it is that because everyone is talking about this AI thing, but at the end, again, people want to do their job. And um, sometimes it's not so important that maybe the person who is using AI knows that AI is in it. But as a company, when we are talking, you need to know what is AI enabled, who has access to my data, what is processed. So different levels about who is responsible for I, what should I consider when I'm talking about AI and what is helping me as a company, and that is the most important thing. Use AI, and it doesn't matter all that you said, Alex, uh, before we have different kinds of AIs of implementations, we have things that are assistance that are helping us at the point where we are currently being writing a Word document, doing things. We have agents that are waiting on inputs from us, we have agents that do things automatically. So different kind of AI, but again, at the end, it should help me to be more productive, more satisfied, having maybe more time for things that I need multiple hours before. Now I can use a researcher or an analyst uh agent, can do things in minutes. That's something that is cool also for you when you use it. So AI, using AI, it's cool. And when you want, okay, you can copilot also ask to create an image that's also possible. I think currently everyone is focused on the point AI and asking and doing things. Looking in the next maybe two to five years, we will have so many AI, and the word AI would not be so present as we have it now.

Alex 6:21

As it becomes, it just becomes a natural extension of what we're doing.

Tomi 6:25

And the interesting part, for example, looking to agents, an agent is an app. That's it. And when I have two, three, four, five agents, I have maybe two, three, four, five apps. And when I'm looking at my digital workplace, I'm doing also different things. Writing an email, reading a document, looking at my appointments, preparing things, doing a little bit of research. Everything could be AI-enabled, AI-powered, and helping me doing my job.

Alex 6:51

Interesting, you mentioned agentic AI. I was gonna ask you about this later on. So we keep hearing about this whole agentic AI thing, right? Agentic AI. Last year it was generative AI, now it's agentic AI, right? So, how is this whole trend affecting workplace AI specifically? And it sounds to me like in the next 12 to 18 months, we're literally all gonna have like an executive assistant on our laptop. They'll do all our work, you know, for us, and then we can just chill out and and let the AI do the work for us.

Tomi 7:21

If that would be so easy, then I'm not for me personally, I'm not sure if I want to have this or if I'm looking for it. But everyone talking about AI um is also having, and that is something that is in our brains. We had looked at a few movies, we have seen some robots doing things, sometimes AI is scaring us, and so on. But looking at the future, I would say having different types of agents. So an agent is an app, an agent is something focusing on specific data, having a few instructions that is defining the behavior of the agent. And when the agent is not just waiting for a prompt, when it's doing things maybe because it's triggered or because things are happening, then we are in this agentic AI part. And think about when you go to your digital workplace, you open your laptop and you're logged in. And the first thing that you're opening now is maybe something like your emails, something like your chat tool in the Microsoft World Teams, for example. Don't open this. Think about you have maybe something like an avatar that is saying, Hey, good morning, Tommy. Happy to have you here. You have received a few uh messages, some of them via email, some of them via messages. Uh, some people in your organization have changed documents. Um, I have read through it, I have give you some suggestions, and half an hour you have an appointment. This and that was discussed before. Do you agree with this? Should I send an email? Should I prepare things? So I think it would become more an interactive thing with this AI helping tool, but the brain, the brain, it's me, the man in front of the laptop.

Jason 9:03

And that's I think a really good point for us to pick up the the ultimate value or the question around the ultimate value of what workplace AI is looking to deliver. So the way I see it, Workplace AI is about helping individuals to become more effective and more efficient. But when you're working with individuals, unless they are people who are incredibly scarce within an organization, you're only going to gain so much back in terms of productivity for any one particular person. So the bigger gains I think maybe will come with teams or departments. I'm curious, Tommy, how much of workplace AI you see people thinking about from a systems perspective, thinking about processes even just between two or three people within a team and how they can bring the AI to help that that function or that collaboration become more effective. Versus people working individually, doing what I would call organic things, so finding their own way, growing their capability by finding their own path, but not necessarily taking a structured approach. What what do you see?

Tomi 10:08

The great thing about AI is I can interact with the I in my natural language. I don't need to be an developer. I could be. There are AI implementations where you need developers for. But thinking to your example, Jason, think about projects supporting agent. Everyone has it, but everyone has, for example, also 10 projects, and I'm interested in project A for this and that, in project B for something else. Another colleague from my team, same project, is interested in a few different things. So we can have an agent maybe from the same kind, from the same structure, with but with our specific instructions, and we are saying, hey, don't send me every time someone is pasting uh or putting informations here, don't send me a notification. I'm just interested if the project is late, if the budget is something wrong with, and so on. So I'm putting in my agent with my natural language and saying, hey, this is for me interesting. In another project, this is interesting. So this will change a little bit the view, not being spammed with notifications, not having to read through everything. I have my assistant helping me to sort out what's going on. That is, I think, one of these examples that can help people at the digital workplace to structure a little bit more their daily work.

Jason 11:31

Okay, so that's really interesting. Are you talking more about focus then? It helps workers to become more focused, or is it actually that they can find information that would have been a struggle for them previously?

Tomi 11:45

I would say both. Structure a little bit, help me to find things, but again, this are maybe different also technical implementations. When I have an AI that is helping me to find things, it's not the same implementation technically in background as when I have someone looking maybe some just on notifications, uh, just on chat messages and so on. So we will have different kinds of things, and again, I think for the normal persons using this, it's not important if it's this or that or that kind. The challenge would be make the UIs and the interface for the people so easy that they can do things to help supporting their daily work without being a programmer in background. So not using, I don't know, C, JavaScript, and what we have also programming, Python things and so on. But also important is when you identify in your digital workplace that you're reaching some point where you can't go many maybe a step more. Have someone in your company, maybe someone also supported by an external company, to help to find out and to sort out a little bit where is the problem, how can we support it? To give you an example, I've built an agent looking at specific documents, and 80% of the documents were okay, 20% of the documents were not okay, they were not AI ready. I couldn't get the information from the documents. So we will have also to look at what do we have for informations, where do they come from? Maybe this is something where technology can support us to tie up a few things, but at the end, me as using digital workplace AI, maybe don't know how this technically works.

Jason 13:23

Yeah. Gartner coined a phrase called productivity leakage, which was specifically focused on workplace AI. And it was looking at the challenge that people may use AI within their tools, within their applications to save themselves time, to be more effective, to be more efficient. The question is where that time goes. Whether they do more work as a result of having got through their their previous work quicker, whether they spend more time being more human in the workplace, that's being creative, maybe thinking, strategizing, whatever it is, so bringing value through that way. Of course, there is the alternative that they just uh pour themselves a coffee and go and sit down and put their feet up for 15 minutes or take the dog out for a walk or do whatever it is that they wanted to do that actually gives them a better quality of life, a better experience of work, but doesn't necessarily raise the productivity of the organization. And I'm thinking about workplace AI where organizations implement it at scale and what the expectation they have versus what they're seeing back is. Now that's a bit of a philosophical question, maybe, you know, that should we be actually trying to force where people take that free time and use it when they've been given these tools. But Tommy, do you think it's an argument that the future of AI raising productivity productivity in organizations will be aimed more at processes rather than people? Both.

Tomi 14:52

I need process supporting AI and I need people supporting AI. Um from the differentiation, what I'm hearing, they are saying an assistant is a personal assisting AI and an agent is a process-based AI. But I would like to give you an example. Think about someone who is working in a bank. I have done um uh sessions in front of bank people, and uh after the sessions, um, we were at dinner and I asked them, so what do you think about this whole AI stuff and so on and so on? And I was talking about um the assistance, the digital workplace AI stuff, and um he said, What I love in my job is sitting in front of a bunch of monitors, seeing how courses are going, buying things, selling things, and the rest, I need AI to help me to sort out maybe my emails, my meetings, my other things. But they have also AI supporting their main business. So I have process supporting AI, looking at how courses are changing, what do we did maybe last week, what is the comparison, is there something? Should we buy? Should we sell? There is something you have the personal intuition, as people sitting in front saying, okay, maybe that the system is saying this, but I know we should do this and that. So that is the the thing the human is doing, and then you have the kind of AI integrated into the system, and then the kind of AI just helping you to make things that are that are there, making a little bit yeah faster. Or when you're doing a research and before, you have to read through different documents, you have to go into a different system, you have to go into your CRM, and so on. And when you now say, hmm, I received an email, customer wants to order something, and you have an AI that knows, oh, this is my customer. Uh, we sold this and that in in the last things. Hey, we have similar customers uh ordering the same and this and that. Hey, we have uh from sales a new uh package that we can also sell and integrate. Hey, prepare the whole information that took me before, maybe one hour or half a day. Now maybe it's in two or three minutes there. I can read through it and then I can decide, okay, now we can do this and that, click on two things, and then it's going on. And this would be a process. We are currently at the ramp up starting phase. Everything is implemented here a little bit, there a little bit. Let's talk again about digital workplace AI, maybe in two or three years, how things are changed between now and then.

Alex 17:13

Yeah, agreed. I I think one of the things or one of the underlying things I'm getting from what you're saying links to let's call it overwhelm or or digital overwhelm. I think every single, as every year goes past, the amount of information overload that people are experiencing, you know, you mentioned notifications, messages, emails, DMs, you know, LinkedIn, et cetera, et cetera, we are all coming under fire, if you will, with more and more information. So actually, one of the things I think that will come out of this is AI helping to filter some of that noise and actually allowing us to have that little bit more brain space, which actually is a really good thing. Even if we're talking about, you know, you talked about productivity leakage, Jason. Just having that little bit more of a breather as an individual, an information worker, etc., I think will actually improve the quality of what people are generating, well as you know, helping us on the productivity side. There is one thing though that kind of, you know, Jason and I talk regularly with uh with customers, and there's a there's a lot of consistent trends, I think, that we see. And one of those is concerns around things like digital sovereignty, uh, which is a which is a massively growing trend. You know, we had GDPR came in in 2018, but ever since then, even just in Europe alone, we've seen what like six-ish massive pieces of legislation, you know, AI Act, Dora, et cetera, et cetera. We are always asked around, you know, is this thing secure? So we know that when you mentioned earlier on generative AI came along a couple of years ago, everybody immediately started dropping company data onto Chat GPT, which I'm sure was not exactly ideal for organizations. But you know, what what do you think from a customer perspective? How do they protect that sensitive content that's being generated in these solutions?

Tomi 18:53

As I said, we are currently in the ramp up phase, and we are trying to use AI on things, on data we had for years, for 10 years, for 20 years. It is structured like it is. We didn't know it better before. This is a little bit negative. To say it positive, we found a way to support our daily business as we did it before, and now we have AI. In looking at AI, great AI, you have just with great data. And when I'm talking about AI and data, there are also two different uh perspectives. I can have an AI model that is trained with data, and I can have an AI model that is looking on data. And the difference between this is when I'm looking at ChatGPT, at Copilot, at generative AI-based models, they are pre-trained with data. Maybe I don't know where this data is from, who has trained it, how it was trained, what is the thing that is necessary. Okay, but it helps me to understand what I want to do. I give two instructions to it, I have data tied up to it, it helps me to find information. It's good for things like it is, but going the next step, why not having your data and training your own AI model? The difference is with training your own AI model, you will get responses much faster and much quicker than when you have to use a GPT model transforming things, going to an index, going into search, finding information, doing things. And um that is the difference. Great AI with great data that I can use just to attach it to AI or to train AI with. And from the security perspective, you are also right. We have to see what kind of data could be used for AI to train a model, what kind of data can be used for AI to search about things, and also keep in mind sometimes you have data that shouldn't be touched by AI or anything else, because it's your property, it's your knowledge in the company. Be careful. What I also have often is that people are saying we have a lot of files in our local data center, and I say, okay, and what's the problem? Yeah, that doesn't work with AI. I say, yes, we have local AI hosted in your data center, accessing your data, not anything is going out, it's all your. So don't miss up or or say AI is not possible because we have data here or this or that. We have to enable, and that is the thing where we are. Now, with having AI in mind, we have to see how we have to transfer and transform our data to be better, more AI accessible, and also to reduce the whole silos that we have in inform uh in all our companies and so on. Make something like a data lake, one data storage thing, two or three, but from the idea, having data centrally stored so that you can bring AI on top and get information out that sometimes you didn't wear aware because with prompting and with AI, you can ask questions. You couldn't imagine to ask them before. Four and now you get results in quick time. Everything needs to be secure, and security is access, permissions, and security tied up to your data. Sometimes, for example, it's good to have data for a few people that is done with permissions. And sometimes in your data, you have specific informations about your intellectual property, about credit card informations, people informations, um, the hirings, I don't know. That is something that is tied up to the data itself, so that when it's flecked, like, hey, we have here specific data, please don't use it with AI, that's also good. And that you have to keep in mind to structure the data in your company in regardless of having now AI into it.

Alex 22:45

I I completely agree with you. I think one of the challenges organizations have had in the past is you know, that it's that shadow AI piece. But if we can provide the security, the tools, and to your point, on the data front, the data foundation uh in a secure fashion, in in a in a way that you know you have your data sources, you have your applications and services, and you have your users, and you need that data platform to almost sit in the middle and provide the connections between those things in a secure fashion. That could be in the cloud, it could be on-premises, it could be both. But having a really strong data foundation, I think, is what's gonna set you up for success in this in this new AI world, whether it's doing generative AI fancy applications, whether it's doing workplace AI applications, etc. I think it's it doesn't matter which one you're going, starting with that really strong infrastructure and data foundation is what's gonna mean you have a flexible platform as well that in the future an organization can be agile because you mentioned already there a couple of different things around different model types, right? You've got large language models which do everything. You have more niche language models, which are specific to a particular industry or a or a particular vertical whatever. And these are going to change constantly. Like, as you said, two years from now, what's this all gonna look like? Well, all of those models are gonna be probably 10 times smarter than they are today. So what we can control now is what we're building is our data foundation.

Tomi 24:10

Let me ask you a question. When you go and do online shopping, sometimes you see, hey, you bought this and that. Other people's buying the same thing like you also bought this and that. Thinking at your digital workplace, assuming you are writing an email, do you want to see information? Say in your company, similar emails are written from this person to that customer in this and that way. Would this be great or would this scare you?

Alex 24:37

Yeah, so I think that that might be slightly inappropriate, especially given uh GDPR and so forth. Um, yeah, that's a that's a really interesting use case I hadn't even considered.

Jason 24:48

It is interesting if you think that maybe the data concerning part could be anonymized or obfuscated, and then it would show you effectively the style, the structure, the way it's broken up, or some of the key facts which are included. I think that's really interesting, and it it does it comes back to that you know, you were talking about data platforms, Alex, and one of the things about that I get excited about when I think about data cataloguing is if we can bring AI to data platforms and it can understand enough about the organization and the purposes of the data, then when people, ordinary people are looking to try and understand more about the processes they operate, the business processes that that they're absolutely central to carrying out, they can start experimenting with data, but that then they can get recommendations, they can get support, they can get explanation from AI, and it's going to enable people to do so much more with data in terms of generating insights and analytics. So it's taking this area of workplace AI, which ultimately is like Tommy, you've been saying, is how can I do more in the area that's specific to me with the tools that I use and the data that's relevant? But when you start looking then about forward-looking for the organization, how it can add value, how it can drive value, working with the data to generate forward-looking insights and being able to take action, you know, the right actions, data-led actions in response to those. That's a really, really exciting idea of where we could go with AI leading people.

Tomi 26:16

And we're not just having AI and assistance, we have also the classic automation that could be AI powered, but automation is still automation. So, what could be a difference? I can have an agent that is maybe monitoring my email mailbox and reacting on things when I'm saying, hey, when you receive something to this project, this is currently very interesting for me. Please give me directly a ping. It's not just a notification, make this and that. But automation is something like um you want to classify, for example, an email. I had a customer from the from uh a travel agency receiving a lot of emails for specific things. And uh would it be good to have an assistant when you're going into the email mailbox, looking at the email, and then asking the assistant, hey, what is in it? No, that's taking too long. But what we did was receiving these emails, taking the content, having a little bit of um uh classification. Uh we have some instructions and what was uh really important for us. And at the end, we get an AI extracted information overview where the people can look at this one and then to decide this email was spam, this email was something like hey, someone wants to book something, someone has an interest about uh getting information from the hotel, someone has this and that. So different things using automation, including AI, to help me in my process of answering emails and being there and responsive and everything that was AI powered. And as I said, that is automation um that you can use. And the crazy thing was before people that where I was talking with the customer said, hmm, I we don't think that would help a lot. And after two months of having this, the CIO gave me a direct call. It's not usually that CIOs are calling me, uh, and said, You wouldn't believe it, but I saved over 50,000 euros just with your automation, and it took us just a few days. So having in mind how AI can help you, it's maybe good again to have an agent, to have an assistant, and also maybe to have automation that you can use at your digital workplace.

Jason 28:29

I'm glad you mentioned automation because I feel that often it's the most unloved child in the family. So we've got automations, we've got um workplace AI, we've got agentic AI. And the thing about automation is, and I mean I um uh let me ask you this automation, do you see just that general field of automation becoming more intelligent anyway through the evolution of AI?

Tomi 28:54

Yes, I think that's that that's the case. And it could be that I'm automating an agent, that I'm just using a few things, as I explained before, into this, but automation, we have automation everywhere. We can use automation for data preparation. Hey, I'm receiving data from a specific uh vendor or something else, or it's it's um an invoice also. Yes, I can use AI to extract information from these documents. The other question would be why we are sending still documents when we can send just the data in JSON format, but that's another question. But there AI can help us um to extract informations to prepare things to be used then in a second step by a second AI, or we have multiple steps where we have to look into our um daily business, where can things help us? And um, using AI, interacting with AI in my natural language can do things sometimes a little bit better and faster than programming things to get an idea. And if you reach a limit, talk with the people that can develop and then make it really fancy.

Alex 29:56

I'm really curious. So let me put myself in the shoes of an organization, right? They're seeing all this change going on, they're seeing their competition or or other organizations are adopting this AI thing, but you know, for whatever reason, they've held back. Maybe it's security and governance, maybe it's the skills, maybe it's the technology. What are you seeing in terms of like those big blockers for organizations adopting workplace AI? And are there any things that you've seen where, you know, if if I'm an organization wanting to adopt, what are some of the first things I can do which will make it that little bit easier? Help me to get it, you know, just ingrained a little bit into my organization and then grow from there.

Tomi 30:35

Demystify AI. Everyone has a specific thinking about what could be AI, but AI is not magic. AI it's technology. And I need to understand what I have for processes in my company that every company knows. I need to understand where is my data stored, what kind of data I have, what do I have for people, what are the use cases of the people. And um, that is something that I have to have to keep in mind. And at the time we are now, everyone asks, is there AI? Is there not AI? But think in two, three, four years, it it's a functionality that is here. It's the same like we have written emails, now we are using chat-based tools. Before, oh, and now everyone hopefully is using more chat-based tools than writing emails. Um, it's a matter of time. And sometimes when I'm talking to customers, I say, hey, this whole AI stuff, it's really easy. We can do it in half an hour. We can uh set up a document and writing in any usage of AI and the whole company for everyone is forbidden. Sign it, CEO signs it, distribute it to everyone, and we are done. Then a moment of silence, and then uh that's not what we want to do. Okay, I say, okay, now we can start to talk about what is your intention, and then you always also see with whom you are talking. So is it IT, is it the department, is it the team, is it someone from C level? They're having different perspectives on it. Um, what is the cost? Yes, the cost is important. AI is costing us money. The end users are important, so we have different aspects to have in mind, but again, digital workplace AI, think about the people who should use AI and make it for them easy.

Alex 32:17

Yeah, um, absolutely makes sense. There's a phrase that Jason and I like to use uh about human beings, which is what's in it for me? Uh, you know, as an individual, if they don't understand what's in it for them, whether they're the C-suite, whether they're an individual contributor working on a particular, you know, line of business, we can make it real to that individual and see the benefit for them, then all of a sudden it's going to become more interesting. Uh interestingly, I think part of the EU AI Act is actually uh saying that organizations must by law um help their employees to understand this technology, what it does. There's that there are specific requirements being held against regulatory-wise against organizations where that enablement and that empowerment of individuals to understand this technology is then hopefully will then make everybody feel that little bit more comfortable about that next generation and that next step of taking it on.

Jason 33:07

So, Tommy, thinking ahead to the next 12 to 24 months, obviously the there's a real uh rate and pace of change around workplace AI, so it's already moving extremely fast, which might make this a bit of a difficult question for you to answer. But what are your predictions as to where you think workplace AI will be? Say 12 to 24 months from now, you know, are we gonna be seeing um individuals' jobs radically change? Are we gonna see people being released from the workforce because they're no longer needed? And what what do you think?

Alex 33:44

And just just so we're clear, if any of your predictions are incorrect, we will be bringing back on the show and slating you for your inaccurate predictions. So we we we don't take prisoners here, I'm afraid.

Tomi 33:55

Um, I'm also in Microsoft MVP. And as a Microsoft MVP, I have a little bit of more insights on what is going on in the product group, what is on their roadmap, what are they doing? A few of the informations are publicly visible, a few of them not. At the top, currently for this fiscal year, it's agents, agents, agents. So I think looking at the digital workplace, from things that I'm seeing from the tool providing companies, from the things that I'm reading, from the things people are dealing with, is having not just one agent waiting for things. So the hype where we are now going is having connected agents, and I think in the next 12 to 24 months at my digital workplace sitting here, I would be interacting with maybe 10 or 20 agents directly, and in background, maybe I have 100, 200, 500 agents also doing things that is supporting me in my work. So a lot more integration of this kind of tools in my digital workplace will come. That's the things that I am currently seeing. The technology is bringing the right interfaces, it's bringing the right APIs on it. Um, that is something that will come.

Alex 35:11

A follow-up to that, if these agents, because it's relatively easy to create an agent, and anybody can create an agent, right? You can be in a frontline role, you could be in an IT role, whatever. We're gonna see, I would guess, a mass proliferation of these agents. And then off the back of that, you assume with proliferation comes dependency, right? You're you've got big organizations with potentially thousands of agents, they become dependent on them. How are we gonna make sure that those agents are protected, that they're backed up, they exist? You know, what happens if one of them gets deleted tomorrow and it turns out that was a business critical process? Like, how how do we when there's such a proliferation of the number of people who can be creating them? How do we bring a little bit of control around that to allow us to make sure that we have redundancy in there and longevity in these things?

Tomi 36:01

At the moment, we are also at the beginning of how to monitor things, how to have dashboards, how to have logging, how to have an overview, and how to identify what is really the business critical AI app and so on. But is this really new? No. When I'm going to introduce a new CRM system, we have the whole things in B4. When I'm uh switching from system A to SAP, we had this B4. So I think AI, maybe it's something that is new from the technology, but um problems that we have in companies are the same that we have for years, and we have it for agents also. Um, do you have an overview how many Word documents you have uh written in the last month? Maybe yes, maybe not. Is it important to know how many agents do you have built in the last months? Maybe yes, maybe no. So it's a tool, it's another tool that helps me to fulfill my um my work. And if I'm identifying things, so if I'm raising up the pyramid to the top, and I'm identifying business critical things, please use the same thing that you have in your company already for business critical applications. Have a managed service, have reaction times, what is with updates, everything still stays valid as we had it also before.

Jason 37:20

Yeah, that makes a lot of sense. So the things like version control, you know, change management, all of these things, even the you know, CICD. Uh so yeah, absolutely, all of those professional enterprise grade capabilities.

Alex 37:34

For all the Battlestar Galactica fans out there, all of this has happened before, all of this will happen again.

Jason 37:41

Okay, moving on. Tommy, we'd like to get a recommendation for people who would like to learn more, find out more about Workplace AI. Where would you suggest they go and find great sources of information?

Tomi 37:55

So I was asked in the preparation also, what is my book recommendation? And I said, I'm not reading books. Uh, I'm well living in the digital world. I'm uh a community guy. I'm following people, for example, on LinkedIn to identify what's going on, what are the interesting things? I'm seeing references to articles, I'm following blog posts from different directions. So, what is interesting me to see what's going on. So if you are patient in things and if you're interested in things, try to find people. People are inspiring me that helps me to learn, to see what other people are doing, and to have a little bit of an idea where maybe the technology is going when I'm introducing also the yeah, the vendor blocks, the new features that are coming. That's what I'm doing.

Alex 38:47

Fantastic. Well, this this technology is moving so fast, I guess. Uh, any book that you write now is already going to be out of date by the time it's hit the publishing, isn't it?

Tomi 38:55

Yes. They asked me also to write the book. I started just with the with the content, and uh after a week I said, okay, this I can remove, there is something new, and so on. That's difficult. But um, going back to the people sitting in front of their digital workplace, not everyone must read the whole day, blogs, LinkedIn sessions, and so on. Again, people want to do their job. And if there is something that is important, hopefully you have in your company people that know your company that are preparing content, and maybe they have something like a newsletter once a month or something once a week, or you have a community built in it. So there are different adoption and change um things that you can implement to help your people understand what's going on and um do it.

Alex 39:43

Awesome.

Jason 39:44

One particular source of information I'd like to recommend, it's it's one view, it's one angle, is actually Microsoft's Work Trend Index Report. So this is produced annually, it's a survey across 31,000 people in 31 countries. And I've gained a lot from looking at this before to understand, you know, what's the sort of strategic direction of workplace AI, what are some of the key challenges that are being experienced. And there is also a work trend index podcast as well to support that. So so two sources of information with a Microsoft um angle, of course, which you could go to if you're interested in finding out more specifically.

Alex 40:24

That's that's obviously after everybody's liked and subscribed this podcast, isn't it, Jason? Just so we're clear. Absolutely. Yes. Well, look, I think we're uh we're just about out of time for our first ever episode. Um so before we wrap up, Tommy, I just want to thank you very much for your time and coming and joining us today. Super uh insights. How can if anybody wants to follow you online, if they found you inspiring, is there uh an easy way for them to do so?

Tomi 40:51

Tomislav Karafilov on LinkedIn or on any of your favorite web search engines, and I think you will find a lot of things.

Alex 40:59

Awesome. Thank you. Uh including, I guess, as we're talking about AI, then it would have to be perplexity. I'm sure we'll find you on there.

Tomi 41:05

Yes.

Alex 41:06

Fantastic. Thank you, Tommy. And so with that, I'm gonna wrap up the show. So thank you again, Tommy, for joining us. Uh Jason, thank you for co-hosting with me today. And as for our next episode, if you're interested, we did say we are the Optimize to Innovate podcast. And so in the next session, we're actually gonna be talking a little bit more about optimization. So if you're unfamiliar with the topic of FinOps or you want to learn a little more, uh, then we're gonna be diving into that topic with a couple of FinOps experts. Uh, so really looking forward to that. If you like this episode, don't forget to hit subscribe on your favorite podcatcher and leave us a review. It really does help more people find us. And if there's something you want us to cover in the future, don't hesitate to leave a comment or let us know via socials. We're at software one just about everywhere. Um, or we'd love to hear from you by email on o2i at software1.com. That's o the number twoi at software1.com. Until next time.