Taking Small Steps: Implementing AI Tools Gradually The Future of AI as a Base Technology with Daniel Rohregger

In this episode of Decode AI, Michael and Ralf interview Daniel Rohregger, a Senior Solution Architect and Microsoft MVP for Microsoft Co-Pilot. Daniel shares his experience with Co-Pilot and discusses its potential for boosting productivity in various use cases. He emphasizes the importance of caution and research when using Co-Pilot, as it is still in its early stages. Daniel also highlights the need for companies to prepare for AI and explore its potential in their business processes. The conversation explores the implementation of AI in companies and the potential impact on job roles and performance reviews. It discusses the idea of using AI, specifically Copilot, for performance reviews and the benefits and limitations of this approach. The conversation also touches on the future of AI and its potential as a base technology in every company. The importance of considering ethical implications and taking a thoughtful approach to AI implementation is emphasized.

Takeaways

Microsoft Co-Pilot has the potential to boost productivity in various use cases, such as generating new product ideas and brainstorming game show concepts.
Caution and research are necessary when using Co-Pilot, as the default model may not always provide accurate or high-quality results.
Companies should prepare for AI and explore its potential in their business processes, considering technical readiness, tool readiness, and people readiness.
AI technologies, including Co-Pilot, differ in terms of data access and control, with Microsoft Co-Pilot having full access to the company's environment.
Companies should take small steps and implement AI tools like Co-Pilot gradually to ensure data protection and avoid potential issues. Implementing AI in companies requires change management to address employee concerns about job security.
AI can assist with performance reviews by providing objective insights, but it should be used as a supplement to personal experience and feedback.
AI has the potential to be a base technology in every company, providing valuable insights and streamlining processes.
Consideration of ethical implications and the need for guardrails is crucial when implementing AI.
Taking a thoughtful and agile approach to AI implementation is recommended, focusing on short-term plans and gradual steps.

Keywords

Decode AI, podcast, Daniel Rohregger, Senior Solution Architect, Microsoft MVP, Microsoft Co-Pilot, productivity, AI, use cases, caution, research, company readiness, AI implementation, job roles, performance reviews, Copilot, benefits, limitations, future of AI, base technology, ethical implications

Links

• https://www.rohregger-it.de/blog
• https://www.youtube.com/@rohreggerit/videos
• https://tr.ee/TL0XoVzhIX

AI, Microsoft Build, OpenAI, language models, AI development tools, hardware advancements, Google Gemini, technology development

Transcript

Michael (00:01.332)
Hello and welcome to the latest episode of Decode AI. Today I'm not alone with Ralf in this podcast. We have a guest with us. But before we start with the guest, welcome Ralf.

Ralf (00:17.646)
Welcome Michael. It's a pleasure to be here. And it's really cool to have a guest in here in our podcast this time talking about AI again. What a wonder. And I'm pretty keen to get to know about our guest and I'll drop his name now and maybe he can introduce himself a little bit. Daniel, this is your stage.

Michael (00:28.532)
You

Daniel Rohregger (00:40.73)
Hello everyone, nice to have you here. Great to be on stage together with you here. My name is Daniel, Daniel Roegger, Daniel Roegger as some of one namey. My name, sorry, my role, I'm a Senior Solution Architect. I'm also Microsoft MVP for Microsoft Co -Pilot since a few months now. And yeah, my daily business is like doing a lot of Co -Pilot, a lot of Microsoft stuff, a lot of working with customers on these topics.

And yeah, I can't wait for your questions. I'm very excited and I'm happy to be on this episode with you today.

Michael (01:19.732)
We are happy to have you here. You mentioned you are an MVP for Co -Pilot or Microsoft MVP for Co -Pilot, not the minimum valuable product. That's another thing. What does it mean? What have you done to be an MVP for Co -Pilot?

Daniel Rohregger (01:32.922)
Yeah.

Daniel Rohregger (01:45.466)
So I was starting really early with the copilot topik and my company was a so -called jumpstart partner for Microsoft Copilot. And I was jumping right in even before the official release on October last year. So the release I think was in November and we started already in October. So I got the possibility to have a firsthand touch on the very earliest product Copilot.

And since then, I was really excited about that. I was posting a lot. I was finding the first cool things out and I was talking about this in our monthly podcasts. I was talking about the customer events. I was talking on public events and yeah, with all of that. And I've not talked about cool things before. I was already integrated in the community quite a bit. We're talking about Microsoft Teams. We're talking about Planner, all the modern work and

hybrid work toppings. And so things add up and someone said, Hey Daniel, I think you could be one, someone eligible for being in Microsoft MVP. And I tried it and I had luck and that's that, that, that's how this was going on. Yeah.

Ralf (02:56.078)
Pretty cool.

Ralf (03:00.11)
Pretty cool. So you've shared a little bit about your daily business already. So what's your daily doing when you're dealing with a customer then?

Daniel Rohregger (03:05.626)
Yes.

Yeah, so I have two main areas of focus. So one is the whole digital work part as a cloud architect. I'm designing the digital workplace for customers, meaning everything on the collaboration and on a productivity side. So for Microsoft Teams, for OneDrive, for SharePoint, everything that is included with all of that is part of my daily work. I'm also having a big...

hard for adoption change management, which also comes in very handy when we're talking about co -pilot, because this is the second part of my daily work. So we are currently doing a lot of workshops with customers. I see a lot of customer requirements. I talk a lot with different areas from the customers and we are once doing like the, yeah, the looking, the first touch points. Yeah. We're doing the first touch with the customers.

And we are also doing like the first implementations with Co -Pilot already. So quite a, quite a long journey already.

Ralf (04:12.206)
Okay, cool. So when I'm listening to you, you spelled out Senior Cloud Architect. Does that mean that you have also development experience?

Daniel Rohregger (04:20.954)
No, not really. I know what it's doing and I have some customers who have these requirements, but we are currently not focusing on that. We're really focusing on the productivity side. That's where also where I came from and where I also came across the M365 co -pilot.

Michael (04:44.852)
When you talk about AI and Microsoft 365 Copilot, you mentioned productivity as well. Do you think it's a real productivity boost?

Daniel Rohregger (05:01.818)
Not for everyone, but yes.

Ralf (05:04.366)
Hahaha!

Michael (05:05.14)
I think that's the best and honest answer. Otherwise I would just challenge you with some theories.

Daniel Rohregger (05:07.578)
Ha ha ha.

Yeah. Yeah. So what I've done already, I have had a workshop with, with, some project managers. We had an eight hour workshop, a full day workshop, and we talked only about how can Co -Pilot help us in our daily life, in our productivity as a project manager, because this is also my past. I have done project management, with, IT projects for like five years or so.

So I quite know what I'm talking about and what kind of at least challenges we have in our consulting life, in our, you know, IT life. And I see a lot of potential here. Example, for example, especially for project managers, but also there are quite a few things where, or quite a few areas where Copa really can provide already now a big, big positive impact on our daily productivity.

Ralf (06:06.222)
That's cool. Thanks for sharing that. And listening, when you were talking about AI and that you started early as an early adapter with Copilot, was that your first touch to AI or did you have any experience before? What was your way going to AI?

Daniel Rohregger (06:24.954)
Well, my journey to AI was not as long as some of you maybe already have experience with that. So I was really coming from a productivity point. So of course I was somehow impressed with the open AI capabilities, with the picture generating, with Dolly, with of course, chat CPT, but maybe only one year, maybe one year and a half. So I just...

get help from it by doing some of my social media, lead inspiration, not a big fan of really copying or let it do on AI. I always get like some inspiration on that. And I already have done that before Co -pilot was released. So this was quite a short journey. And I've not taken part on the whole machine learning and deep learning AI stuff. I really came in across right from the side from the productivity area there.

Ralf (07:23.118)
That's cool. Thank you. That's awesome. So it's really that you've just got in touch with AI just, let's say, two years ago. Okay, cool. And is it possible that you can share some of the use cases with your work? I mean, you have had shared that one out of your business or the free company, but are there others as well that you can share with us?

Daniel Rohregger (07:33.594)
Yeah.

Daniel Rohregger (07:46.746)
Yeah.

Yeah. So there is also one, one part I want to say. So I think because going back to your initial question, does Co -Pilot really provide a productivity boost to everyone? And I said, not for everyone, but I think most of us can find a case. I think that not all of us will find enough cases to get a return on invest for a whole Co -Pilot license, but I'm sure

many of us, if not all, will find some kind of use cases. So, and this is also what I do. When we talk with customers, we talk with different departments. Sometimes we talk with the department leads, and sometimes we talk with just member of the departments who are working there. And that's where I get a few use cases from. I had, for example, one customer

Michael (08:46.26)
Thanks for watching.

Daniel Rohregger (08:48.762)
that was releasing products. And they were also doing a lot of new products in the year. So they have a wide area and one of them was cosmetics. So they were thinking about, hey, what kind of different new cool product ideas can we get? And we have designed them a co -pilot use case that was focusing on new products.

One of them was, for example, just, hey, we are brainstorming, looking into the market, looking into existing knowledge. What kind of products do we have? And combine that with current trends. What could be a easy to use product that we can make easily? And maybe Copilot generates 20 ideas. 19 of them are crap. But the one idea that we find, maybe it's valuable, or maybe it's, it's, it's

faster and more efficient to get these ideas by AI. And then we give the AI the task to design the product and give us more insight. And we do this quite a few times. So we have like maybe four or five new products described by AI, maybe also we generate the pictures by AI. And then we go into our development offices. Hey, hey, look, I have five new products ideas. Maybe even from these five, maybe also two or three are shit again, but...

Michael (09:57.908)
Good.

Daniel Rohregger (10:17.242)
we can then go on and maybe develop a new idea from these existing ideas or even take one and go further with it. And again, then AI again comes handy. We ask co -pilot again, hey, can you give us more details? Can you give us a description? Can you give us a picture? Can you give us an idea how that would look like? And so on. So we go into the different next stages and we always get use of AI when we need to.

generate, hexes, generate, whatever. And this is one very common use case. We also have another that is quite similar. We have a customer who is doing TV productions. And this is very similar because they're always in need of new game show ideas. I'm sure when you think about game shows, for example, you know, either on the public television or on private ones, they're always like mini games and so on.

Michael (10:58.036)
you

Daniel Rohregger (11:15.354)
And it's quite a hard time to get generate new ideas. And that's a really good discipline of AI currently. Copilot and also many other AI are really good in generating information, generating ideas. And even if the idea is not perfect, then a human with human knowledge and best practices can work it out and does not start on a blank page. That's really currently a great, great feature.

what we are doing with Copilot in the really traditional natural way. Of course, we have some other use cases. If you want me to go on a little bit, we have, for example, when focusing on chatbots. So Copilot is a chatbot that is integrated very well into the Microsoft ecosystem. But it's not only that. We already see the first customers integrating, for example, service desks and information system.

knowledge basis where we have confluence or Jira or something like that. And we are seeing more and more integrating customers. These additional third party products were also knowledge from a company is integrating that into co -pilot. So if someone wants to know information about something, he does not need to know where the information is based. It's completely irrelevant. If this is on SharePoint, if it's on Teams, maybe it's on confluence or in the ticketing system.

doesn't matter because you just ask Copilot and Copilot already today has the capability to expand its knowledge and its index to additional attached third party systems and giving like one interface where I can ask with natural language to find information. And I can really say for myself, the finding information part is a really cool one. There is still room to improve, especially when you know,

Michael (13:01.656)
you

Daniel Rohregger (13:13.178)
when you know how to use a search, when you have some kind of IT knowledge, you know your parameters, you know, you can say, okay, I would like timeframe from today and 24 hours. And we have like, excludes, includes, and we have, you know, all the different types of filters you would apply and Copilot is not there yet. Yeah. Especially when you would, for example, search for, hey, give me all my last emails.

I have sent to this and this domain. And you would like maybe mark the first part of the domain with a star. And because you don't know who was the recipient, for example, and this is not capable yet, but these are some basic search parameters that for sure will come in a future iteration of the product. And with the possibility for a normal user to just ask someone, Hey, I know Daniel has a presentation about this and that where is it?

Michael (13:58.004)
See you in a minute.

Daniel Rohregger (14:12.25)
And if Co -Pilot knows, hey, it's on a public place, then he will guide him to that place. And that is existing already yet, even for third parties. And I think this is a really, really great use case that we see a lot of customers.

Michael (14:29.748)
That sounds promising. A little bit scary actually if you can search everything just by asking something quite on a high level and you get some detailed information. But talking about the information, you said you bring some information into a system, you use information and someone can use those information from different systems and

Daniel Rohregger (14:31.674)
you

Michael (15:00.34)
How do you manage to control who got access to the new information, to use all this maybe internal knowledge, maybe intellectual property which is internally and you want to protect as good as possible?

Ralf (15:17.742)
Yeah, not every information is needed by everyone. So I mean, it can be digits about the company or about projects you don't have access to usually. So interesting question, Michael.

Daniel Rohregger (15:32.922)
Well, yes. And, but the discussion that we are talking here or that we're having here is not a new one. You know, access and role -based access management and, yeah, these, these kinds of strategies, behind that everyone just needs to see the information that he should see or that he, that, you know, we have some limitation. it's not really new. We had this when we talked about Delph, we had this when we talked about SharePoint search initial, initially, but

Yeah, so of course, first of all, we need to take care a little bit about our sharing infrastructure, meaning maybe we limit who or we limit the defaults in first place. That's where we start with the customer. So we look when we are sharing information, do we share that default with the whole company or just with the specific users or groups that should have access to that? And yes, that we have also

the opportunity to use like sensitivity labels, which gives a second area of security. And this is one of the great parts where Copilot comes in because Copilot currently, so only I tool that can use that sensitivity label, meaning if you got a sensitive document from another one, maybe you weren't aware of that, that it's very sensitive. And you ask Copilot, hey, generate me of this document in our PowerPoint.

And Copilot says, yes, of course I do that. And then you have the PowerPoint. Copilot automatically detects the origin label and applies it to the new document. Meaning if you are generating a new document with AI, it will have the same protected status and the same sensitivity label as the original document had. So you're not making a failure on a mistake. And there is some kind of protection built in. Yeah. But of course,

This is only for companies and we need to be honest about that. It already have rolled out information protection already, maybe got the users to use it. We see there is not a big, big usage on that. Many talking about that, but we see that the readiness on the field is not quite there where we want to have it and where it should be for rolling out an AI for a whole company. That's also why some holding still back. And we have some other controls.

Michael (17:31.46)
you

Daniel Rohregger (17:56.282)
that we need to look on first. That's besides the whole sharing topic. Also, do we have a lot of public teams? Do we have a lot of private teams? Do we have, based on our reports, a problem with overshare documents where we also know, hey, yeah, we had this kind of program for our students where we let them look into all of our different departments.

Michael (18:01.168)
you

Daniel Rohregger (18:21.786)
and they get access to them and nobody was really taking care that the access was revoked after that. So we have some like students maybe that are an hour in the last year and they have access to all the relevant company documents from starting at marketing, starting with AI, with HR, with starting in IT. Yeah, you know, and where they're running all through the different departments. We have those cases.

Michael (18:32.284)
you

Daniel Rohregger (18:50.298)
and we need to take care about that. And basically we have two answers currently for that. So first of all, and we want to be very restrictive and we want, or, and we are very unsure about our current SharePoint infrastructure when looking from a Microsoft point of view. Then we have some,

Michael (19:01.692)
you

Daniel Rohregger (19:13.722)
something that is pretty new and currently in public preview, depending when these podcasts episode will come out. It's called SharePoint restricted access. And these restricted configuration is like, you have no SharePoint sites in the co -pilot index, only the ones you specify. So meaning if you have like a

Michael (19:30.916)
you

Daniel Rohregger (19:41.114)
public SharePoint from your company where all the information that our public knowledge is stored on, then you just include that. It's like a white listing. So per default is none. And you only say, hey, I want this and that SharePoint site to be in the index for every user. And there is also the other way around. I call it blacklisting when I say, hey, we have the sensitive SharePoints for like HR or maybe you from

Michael (20:01.207)
you

Daniel Rohregger (20:10.842)
Yeah, the leadership team, for example, then you can also exclude SharePoint site from the index for every user. So these are like two mechanics when we say about, okay, we want to roll out co -pilot and we want to make sure that the data the user is seeing is okay. Then we have these two controls to make sure of that.

Ralf (20:34.03)
So summarizing what you're saying is there is no great control out of for this problem yet, but there are some concepts you can follow with to start with AI within your company. I mean, you didn't share a very big surprise that companies are behind looking to AI. I mean, companies are still behind looking to cloud. I mean, this is surely not a big surprise for us.

Michael (20:34.139)
you

Daniel Rohregger (20:55.13)
Yeah.

Daniel Rohregger (20:59.802)
Yeah.

Ralf (21:03.63)
But okay, thank you for the insight here on your thoughts and on how you see that information can be protected somehow with that tool set which is available for the moment. That's pretty cool. So that brings me to the next question. So how is yourself thinking about AI?

Daniel Rohregger (21:28.347)
Well, I'm so in the customer world that I'm always looking from a customer perspective, to be honest. Yeah. So I'm really thinking into every customer that I'm talking with. So when I'm thinking of my past, when I was a project manager, I would be really excited to be honest, to be a project manager these days and to use that and to get use of that and to

Michael (21:50.464)
.

Daniel Rohregger (21:56.57)
Yeah, get faster project progress with the help of AI. And of course, I need to admit, yeah, I'm already using that when I'm working in projects. So when we're working with customers and we need some kind of document, hey, why not give it to co -pilot and give a first draft? What I'm taking very seriously and I hope also everyone does is that currently

Michael (22:19.488)
.

Daniel Rohregger (22:24.474)
with the default model of Co -pilot that we have. It's not really easy to get what you want in terms of correctness and in terms of research quality. So it's very important to have a look on that, to do your research, to double check and maybe sometimes triple check with an expert. You know, when I'm doing cloud architect stuff like

writing roadmaps for customers. I easily could just like to co -pilot write a few sentences about every topic, about every technology, but in a customer scenario, I always need to recheck. And that's what I'm really doing. So I really try to give it more and more place in my daily work. And I'm still also experimenting because I want to see how it feels and how we can get use of that.

Michael (23:02.396)
you

Daniel Rohregger (23:24.506)
And also tell my colleagues to inspire them and to say, Hey, look, there is something new and it's cool and it's beneficial for every one of us. And let's explore, but with caution. That's what I'm also saying to the customers and what I also yeah, doing right now. A lot of exploring, a lot of getting my hands on new stuff and also trying to, to be honest about if and when the co -pilot is really ready to trust it more and more.

Michael (23:55.38)
Do you think there's a difference between the Microsoft Co -Pilot and other AI technologies? Does it make a difference from the perspective? And just thinking about the first step was the announcement of OpenAI. And I think that was something also from the public interest, from press perspective, you got just...

So many news about this AI stuff that it's now something that feels like it's important to have. Do you think that makes a difference between the different AI products?

Daniel Rohregger (24:38.746)
I mean, if you ask the companies, yes, it is of course, because Microsoft is per default having access to your environment. This is why some are still a little bit afraid of it because they say, hey look, if I have a chat GPT business, for example, and I only give the data to chat GPT that I want to give and a Microsoft is the other way around.

So it comes like with the big feature that is like the big selling point. It has the full capability and it has a full access to your documents if you want. So yes, I would differentiate a little bit. So the co -pilot is a great, great opportunity, but Microsoft really needs to watch its steps because if they're a little bit too fast, and that's why I am personally pretty happy that it's

They're slow rolling currently. I'm not so happy about the lot of lot of news announcement in every conference, in every place, every time we see new things coming up. And I would say, hey, slow down a little bit. Let's implement the thing step by step. Because if you remember how it worked with teams in the first steps, God, it was not easy all the time. It was sometimes really buggy, not stable, bad quality.

and co -part of this are really new, new product and they're doing a lot of it. And I really want that it's going well. And I really would say, hey, let's do it step by step. Let's do it slowly, but with a good progress. And then that could work out because the difference is again, it has access to the data per default. And that's why Microsoft is giving...

a little bit more potential for the users of the product and for itself to maybe have some bad impact or some bad press by, you know, oversharing information by giving people access to data they don't have because they didn't know it, but that could happen. And I really, I always tell this also to my customers before we do a small step forward by implementing co -pilot and then having like a problem.

Daniel Rohregger (26:58.842)
having like a data oversharing or a data protection issue, because then we go three steps back and we don't go one forward, we go three back. So let's go small steps, let's do it good. And in a controlled environment, that's why we are, to be honest with you, are currently only running pilots. We are running some bigger pilots, but we do not run the first big implementation. I know a few customers, they've rolled it out to a

bigger part of the company and also not to everyone. So let's do small steps to make sure everything is running and currently focused on the basics and make sure that these things don't happen.

Ralf (27:44.046)
Yeah, pretty, pretty interesting. Sorry, Michael.

Michael (27:44.308)
Do you think?

Yeah. because we were talking about co -pilot and other AIs in companies. Do you think that's something you have to use today? AI in general in companies, or is it something you should just discover? Be curious.

Daniel Rohregger (28:13.562)
So if you're not using that right now in a company, I would say, don't be afraid. But if you're not talking about it right now and not preparing yourself, then you're doing something wrong. So you're clearly not missing out by not using already maybe some chat bot or something like that. That's not a problem. But if you're not even thinking about that and not even preparing yourself, your stuff,

your guidelines, your IT, your technology right now. Yes, you're missing out. And I can tell you, we're already talking with some, with some machine powers, machine builders from Bavaria, you know, when you see them and what they're doing, you would not even think that they're really taking care of AI, but they do. And it's great. And they have discovered Copilot and they said clearly, yes, we now we know what it's doing. And now we know we are not missing out.

we have, we see Copilot and that's a perfect spot for a tool for personal productivity. And the guys and ladies, they are, they are building things, they are building machines. So their job is not sitting in front of a computer all the time. Of course, they have some workers, they do. That's what I said. Copilot is not for everyone, but for some. But they say, hey, these AI discussions led us to think even bigger. And that's the point. So

Try to see where in your business processes do you already have the potential to use AI and even just small bits. And that's where we see the customers are really thinking about now. And that's why that's where the most are a co -pilot workshops also lead. It's not doing a lot of lot of lot of M365 co -pilot implementations. That's not to be honest, the truth right now, because they're saying, yes, we see what it's doing and we see this for the personal productivity.

And we know it's not for everyone. And we wait a little bit until it gets better. And then we for sure will roll it out for some, for the right groups. But we are now talking about AI in a bigger, you know, not from a personal point, from a department point, or even from a process point. That's where it gets interesting. But you also need the readiness here. You need the technical readiness, the IT readiness, the tool readiness, and also the people readiness there.

Daniel Rohregger (30:36.25)
You need to do some change management if you're implementing AI based processes, because people are getting afraid of that. Am I going to lose my job? No, hopefully not. Your job maybe will change and maybe we can kill some new job offers right now. So that's where also the bigger companies are currently thinking, where we say, okay, it is going to kill some jobs. No.

but maybe we don't need to hire so much new stuff because our current employees can do better, can do faster, have maybe a little bit for a room for improvement because AI is taking care of the easy parts or of the repetitive boring parts of their job.

Ralf (31:24.174)
Okay, got it. That's cool that you have a statement there and that you share this statement with us. I like it a lot. So Daniel, there was a post on your LinkedIn stream and that post was like shaking me all through. It doesn't brought the discussion you hoped from it, I guess. And I'm a little bit curious to see if you're guessing what posting I'm referring to.

It has to do with people. It has to do with managers. It has to do with companies. And I'm referring to your performance review with copilot posting. And I was really thinking of, will I start a conversation within LinkedIn or on how can I handle this? And I decided to invite you to our podcast. So that was the trigger point really.

Daniel Rohregger (32:03.642)
Yeah.

Ralf (32:21.998)
You brought up some interesting questions in there. First of all, let's discuss the idea of doing a performance review with co -pilot. From your experience as an employee, do you really want your manager to prepare himself with co -pilot to have a performance review with you? It is an interesting point of view.

Share a little bit about your thoughts here, please.

Daniel Rohregger (32:54.682)
So also this was not really my idea. It was also a customer's idea. And of course it was meant a little bit as a joke, but to be honest with you, as a customer, set this one and brought this idea up, I get a little bit of goosebumps. I never had this in a workshop. And this why it came to me, I said, yeah, that will totally work. So from a technical point, you know, asking AI,

Michael (33:01.264)
you

Daniel Rohregger (33:21.594)
to give insights on someone else's performance or give insights on someone else's, what they're doing on their job, that will work because as a colleague, as a manager, you will mainly have access to the documents besides the mails that he's sending out directly to the customer without you knowing. But if he's posting things on Teams, if he is storing documents in Teams, you will have all the access to it. So there is a big potential.

answering your questions, would I have this on a meeting with my manager? What I like about the idea is that it is objective.

Ralf (34:03.47)
do you think it is objective? So I'm, well, all large language models have kind of bias. It is based upon the data they are having to train. So there is a bias in, we cannot, we can, so maybe we can agree that there is a bias and this bias will have an influence on the so -called objectivity here.

Daniel Rohregger (34:20.794)
Yeah.

Michael (34:27.424)
.

Daniel Rohregger (34:29.466)
Yeah, it is. Definitely.

Ralf (34:30.158)
So I'm not clearly and not 100 % sure if we can talk about objective view on that point.

Daniel Rohregger (34:37.338)
Let's frame it, it's possible more objective than my manager is having a bad day, having stress with her family, maybe came late to the office to do some traffic. I think it's even more objective than that. But yeah.

Ralf (34:42.67)
Haha.

Ralf (34:51.982)
Maybe, but still he will have his bad day and then even co -pilot cannot make that rid of him. But I found it very interesting. So I've also have seen the picture you posted there where there was that prompt given to co -pilot and the answer of it was like interesting because that means that

Daniel Rohregger (34:56.73)
Exactly.

Daniel Rohregger (35:11.29)
Yeah, I've tried that.

Ralf (35:20.622)
All the given data, like all customer meetings you had using Teams was analyzed by Copilot to give that answer.

Daniel Rohregger (35:29.658)
Yeah, at least those one that were recorded or transcribed. Yeah. So of course it is just an idea, of course. Yeah. But you know, from the other side, as an employee, you can also influence what Co -Pilot can see. So of course, you know, not transcribing your escalation meetings, but only transcribing the meetings where you get good feedback is also somehow influencing that.

And yeah, so the data behind is very interesting or is very important. And I think when we talking about companies that are doing really loud stuff in Teams and SharePoint and OneDrive and working with those things, then why should the databases would not be valid at least for a additional point of view from a little bit more objective point of view from an AI.

Michael (36:01.796)
you

Ralf (36:28.206)
Now I have the goosebumps. That feels a little bit more than being watched all the time by Brick Brother. It's just a feeling, gut feeling here, right? So it's not being representative at this moment. I don't know how Michael feels about that, but he speaks out of an employee as well as on a CEO position. So he maybe has a differentiated opinion here.

Daniel Rohregger (36:34.97)
Hmm.

Michael (36:56.42)
Yeah, definitely. The first idea is it sounds brilliant, right? You just say, hey, Jarvis, just give me a performance review of person X, Y or Z. And it sounds good because when you think about the technology, it sounds like it's touching everything from the user. So you think from the...

From the perspective of an AI, it all is everything. That's not true in the first place. It's just the things you share with each other, especially in Co -Pilot. The next thing is it does not help because it's not doing the very important human part, like discovering some notes, some tones.

maybe have some background, some background information. Also, some visual reflections or reactions can reflect something, who someone acts. If someone is really relieved, for example, because of a huge pressure. And if you have just some words and try to interpret them, it's make it different.

Michael (38:26.484)
You have a different view when you work with people and talk to people and you know how they act because you met them on a more personal level. I wouldn't say private, but on a personal level, you met each other on the bar and you just discussed something. You just got some insights into the family business or...

Daniel Rohregger (38:48.506)
Mm. Yeah.

Michael (38:54.132)
something you share about hobbies or something like that. And you just have another perspective. You have another culture to exchange some topics. And then it may be, for example, sounds rude in a transcript, but it was a joke because you know each other and you have a different, that's all something, some influences and points I would say.

Daniel Rohregger (39:15.77)
Yeah.

Michael (39:24.436)
It sounds good in the first place because it's congested right away and everything is easy and perfect because AI is so brilliant. But in reality, it's not there. It's not also not good from my perspective to get this on a very limited data, so -called data insights of a human being. So that's...

Daniel Rohregger (39:33.658)
Ha ha.

Michael (39:54.452)
definitely the wrong way from my perspective.

Daniel Rohregger (39:57.562)
I agree. I agree with you. And, and at least for the part, I would also not like to do co -pilot my full performance review. But when we switched sides here from a manager perspective as co -pilots, being a personal assistant, helping you with everyday tasks, running a team with 20, 25 people and having like January, February, and you need to do these performance reviews, you have quite a lot of work to do. And to be honest,

If you have not been in touch with those guys or ladies in your team in the recent, what do you do? You go on documents, you're going on SharePoint and you do exactly that. And I would top or I would again use the copilot AI generated performance review input as a first 20, 30 % and then top it off with my personal experience with them, with my

Ralf (40:34.35)
Yep. Yeah.

Daniel Rohregger (40:54.746)
subjective meaning with what I had from customers from other colleagues and use it like as a foundation. And in this place, I would say it is definitely more valid or more valuable as are again, not need to start off on a white page.

Ralf (41:11.47)
Yeah, yeah, well.

Michael (41:11.796)
I think that, I think the important part here, sorry, Ralph, is you use it as a flavor in the discussion and not as a full feedback and performance review. And that's definitely something you always have this AI. You get something and you have to double check, does it fit? Does it not fit? And is it maybe absolutely wrong? So that's also something you may have under specific.

Daniel Rohregger (41:31.226)
Yeah.

Michael (41:41.204)
circumstances.

tonight.

Ralf (41:44.718)
I would say so. That's really good, Michael, what you brought in here. Three little things. A team with 22 members is too big for a single manager, in my opinion. You cannot have a touch to all of them, especially if you're not into their business and you do not have daily touches to them. Get rid of that scenario. Split the team up and make it to smaller teams because that would be my recommendation because you can then...

better handle the team and be a better manager for them as well and see them growing. And appreciation is as well the thing what would be lost in that situation. It can be maximum on my opinion preparation for performance review. And I would even not handle that as the first 25 or 30 percentage to this performance of that.

particular employee because as Michael said, you have to double check everything. You have to drill into it to understand was the tone hit correctly? Was it a joke or not? And even the personal circumstances of that employee of that year maybe have an influence of his performance, which needs to be respected, which AI wouldn't know in that case, in my opinion as well. So you have maybe

to deal with personal circumstances where illness or family circumstances brings the performance down of that employee. But the years before he performed over and was a very good employee. So it can be that it is a lower curve that year, but not necessarily something you need to address to because you know the circumstances of that employee. So it has a double -sided sword, let's say here, and it needs to be reflected

Daniel Rohregger (43:37.562)
Of course, yeah.

Ralf (43:40.11)
pretty well. I liked it a lot that you brought up that topic so that we can discuss it. I found it tough to get done by LinkedIn because there are trolls in LinkedIn as well and that wouldn't be fun. I was trying to prepare some popcorn to have the stream on to see what happens there. Nothing happened so far. I double checked it today. So even today there are just two answers and that brought me not further.

Daniel Rohregger (43:50.458)
Yeah. Yeah.

Daniel Rohregger (43:58.714)
Ha ha.

Daniel Rohregger (44:03.002)
No, no, no.

Ralf (44:09.39)
So, yeah.

Daniel Rohregger (44:10.426)
I'm, but one, one point to add here, what I see my part in. Yeah. So I get these a lot when I'm posting something on LinkedIn, especially when we talk, for example, about mobile work or home office, or, you know, will we ever see again, an office from inside or not? I know the answer, you know, and, and it's very similar to hear the question and the answer to that is obviously no, Copilot will not do your net.

your next performance interview. But I think it's like, you know, when you go into discussions and you already know that the answer will be somewhere in the middle and the outcome will somewhere a compromise or a agreement between two parties. You need to have an opponent and we all know we have a lot of these guys on LinkedIn that said, yeah, it's good already and

Michael (45:02.092)
you

Daniel Rohregger (45:09.114)
AI is bad and all these things. So I try to be a very positive and sometimes a little bit more extreme counterpart to that, that I that we can maybe move or shift a little bit, or that we get agreement like, yeah, why not try it at least. So

Ralf (45:25.198)
It's fair enough. All good. Let's close this topic down. Do you post things via AI? Do you prepare your postings by AI or do you post humanly thought stuff?

Daniel Rohregger (45:40.346)
To be honest, sometimes my captions are that short that I think, okay, let's just write it on my own, especially when I'm talking on events, for example, when I want to recap events. But some of my discussions are influenced by AI. So when I say, okay, I know, for example, I already had three or four posts about home office and going back to the office and the stuff. So...

Michael (46:03.236)
you

Daniel Rohregger (46:06.906)
I'm asking AI to give me some ideas or some conversation starters. I use that. But to be honest, I experimented a little bit, for example, with hashtags made by AI and hashtags by my own. And it does make that sentence. I also have the feeling that the LinkedIn algorithm is not really caring about that much anymore. So, yeah, no, well,

Ralf (46:28.718)
Yeah.

Ralf (46:32.11)
about hashtags or AI -routin' content.

Daniel Rohregger (46:37.69)
I don't know that yet, but I think that there will be some like made by human batch in the future. Not only for postings, only for blog articles, for books and so on. Also for music, that's crazy too. But yeah, I use it sometimes, but not as hard as maybe someone would think, to be honest.

Ralf (46:53.454)
Okay.

Ralf (47:00.59)
I just was curious about that. Okay, Michael.

Daniel Rohregger (47:03.706)
Yeah.

Michael (47:04.692)
If you talk about the different AI flavors in companies, and you already mentioned if you are not doing it today, in working and be interested in AI, what do you think about the future of AI itself? Is this something like a

based technology in every company or would that be, sorry, Ralph, thinking about Apple again, say on every device and every situation, everywhere in life, what do you think about this perspective of air?

Ralf (47:40.27)
You

Daniel Rohregger (47:51.697)
Well, to be honest, I am currently thinking in like a one or two year horizon. We had a talk recently that we need a five year plan where we want to go. And I said, no, we don't need it. And it's completely nonsense to do that. We can do that in other businesses, but not in our ones. You know, when we're opening a restaurant right now, yeah, we can do like a three year plan or a four year plan.

but not in our technology part. So what I can say or what I think, we all need to be a little bit more agile. We all need to be a little bit more easier to implement new things. But I see AI, if you give me those two options, definitely more as a base technology. So AI will be everywhere. I also see a big potential for local AIs.

on every phone, on every desktop, not probably in the way that is Microsoft currently introducing it right now with the Windows plus PC, but I see a lot of potential there. And I mean, what AI is currently doing pretty good is understanding things, is generating things, is generating content. And you know, in every part, when you're visiting the doctor,

you have like this big of a paper map and there are all your things inside it. If this is digitized, you just go to the doctor, he presses one button and AI gives you like a five sentence most important things about this patient and the doctor knows very important things you don't need to tell them. Yeah. So, and these are small examples in everyday life. So that will become a part of our life and of every company.

on different devices in different ways as a base technology. Co -pilot for sure also will be implemented more and more and we will find more ways. But with Co -pilot, we always have only the individual and we see the AI perspective, we see processes, we see departments, we see, you know, bigger business things with more value, with more, yeah, easier ways.

Daniel Rohregger (50:15.258)
to make money, to be more profitable, to save time. And that's why we have in a broader way more. And yeah, so as a best based technology, I will stay definitely, but on multiple ways.

Ralf (50:32.59)
Cool. That's a really broad view on the future. I hope that we don't get dumb asses in the future because they just can ask AI and get an answer. So they don't need to learn by themselves. I hope it stays as an assistant with us being helpful. And this is the curious curiosity of like things I want to see in the future on how that will develop the people.

and on how they stay informed and stuff and so on. That will be very, very interesting because we nowadays have already to deal with those different situations where we have like fake news and stuff and all this kind of rumors in our news. And I'm really keen to get to know on how people or kids will be educated in the future.

Daniel Rohregger (51:17.978)
Yeah.

Daniel Rohregger (51:29.978)
Yeah.

Ralf (51:30.094)
and so on. So this is really an interesting thing. I can't oversee it yet, but it's, I mean, very, very, very interesting. So for us, this is it. Michael, do you have any further questions?

Daniel Rohregger (51:35.194)
Definitely.

Michael (51:47.06)
No, actually, it's not a question. But maybe you get the room to have a direct approach, something... How can I put this? Something brilliant you want to share with the audience, something you want to address to the audience, something you want to...

give the audience to think about something you just want to use as a closer for this session today.

Daniel Rohregger (52:30.298)
I think I can only emphasize and summarize a little bit. So when you hear, okay, we will definitely need to take a look on that. It will stay as a technology. And also, as I've said, there are still a lot of things unclear or there is the potential also for the bad guys using it and they will use it. So they will definitely use it. All the bad guys will definitely use it. So

You need to take care about the things that are going on right now. Make a plan for the future, not for the next 10 years. Make a plan for the next one or two or three years. Get your things together. Process these technology and also your people. Take them with you on this journey and see where it in your company leads you and where you see at first the benefit. But of course,

as Germany, as Europe, as everyone we need to talk about guardrails that will come more and more, but that will come automatically. Focus on your business, what you can do in a positive way and don't be too fast, be considerable and take your first step slowly right now, but start with that and please don't try to ignore it because

that will not go out. So you definitely will need to take care about it. And yeah, start right now with the first steps.

Ralf (54:08.142)
Thank you, Daniel, for being our guest here. That was it. And it was a cool session with you, the discussion. I liked it a lot. Thank you for sharing all those insights into your daily business, your thoughts, use cases as well, and stuff. Really appreciate it a lot. Thank you for being here.

Daniel Rohregger (54:12.282)
Thanks for the invite.

Daniel Rohregger (54:27.226)
Thanks for having me.

Michael (54:28.5)
Thank you also from my side. And it was really interesting to have this conversation with you. And also the ethical part, it's something really, it's really interesting. And I think we will have more of those conversations in the near future. But that's it for today. And I think we have a kind of close out already. Let me try this. Stay tuned, stay listen.

There was one.

Ralf (54:58.574)
You're close to it.

Michael (55:04.212)
Right, how is it correctly?

Ralf (55:05.742)
It's okay. Bye bye all. Thanks for being here, Daniel. Daniel, Michael, it was a pleasure to be with you here as well. Stay tuned, stay interested, sign up, listen up. Here we go. Bye bye. Take care all. Thanks for being here and listening.

Michael (55:22.26)
Thank you, bye.