00:00:01:

00:00:26: Welcome to The Optimist, the real-talk podcast with leaders who are shaping Europe's digital future.

00:00:32: My name is Moritz Schoenleber and I am the tech community lead.

00:00:35: We talk about AI a lot also in this podcast but maybe not so much about the hardest part Because while the technology is moving fast, most organizations are still stuck in a gap between pilots and real impact.

00:00:50: So today our moderator Der Rikert is joined by Matthias Emler – his partner at Hovert!

00:00:56: Together they talk about three themes Why most AI initiatives fail to scale What it takes move from pilots to real systems.

00:01:05: And finally how governance data & organizational alignment determines success.

00:01:12: and see you later.

00:01:16: Mattias Emler, partner at Horvath it's a pleasure to have you here.

00:01:20: It is great to be here.

00:01:21: thanks for having me.

00:01:23: I guess before we dive in i'd like to ask how's working for you in the consulting business?

00:01:27: In this era mean AIs everywhere people are building agents do even get the opportunity interact with humans on day-to-day basis of your work.

00:01:35: well actually more than formal times.

00:01:38: So if I look at my time schedule and traveling calendar, every day in a new town here because so many of our customers are wondering what direction to take with regards to AI.

00:01:50: And since these discussions are so intense they would like them in person right?

00:01:55: This is why i'm currently travelling even more than the past!

00:02:00: And I enjoy it

00:02:01: right?!

00:02:01: Because its just great be at so many major companies seeing how evolve and develop in these dynamic times.

00:02:09: And you know, it's funny hearing what you just said almost one is AI causing more questions than answering questions?

00:02:15: Is that a problem

00:02:17: at the moment?

00:02:18: I would say its maybe balance.

00:02:20: yeah.

00:02:20: so In some areas It's definitely Causing more questions.

00:02:25: in other areas That's definitely helping to answer question very fast

00:02:29: and i guess we can.

00:02:31: Before diving into the specifics of what you're doing in the workplaces, I've got people like you coming on a weekly basis to my Optimus podcast.

00:02:40: And I really feel that I'm working with elite athletes or elite performers in their own field and they are playing at your edge-of-your game business as an AI.

00:02:54: When did your fascination with technology start?

00:02:56: Do remember when first time said wow!

00:02:59: Well,

00:03:01: for me personally that was at a time where AI wasn't on the scene yet.

00:03:07: I started as consultant twenty years ago actually in CFO advisory which is already very related to data and i found more passion of data than digital yeah?

00:03:16: And combined these two topics then maybe ten year ago AI became much more prominent.

00:03:22: we as Horvud ramped up our first ATI team.

00:03:25: this really saw okay something exciting happening And since that time, together with my team I'm working to gather our customers and gaining value from data.

00:03:37: From AI in finance and beyond right?

00:03:40: In operations commercial procurement... ...and this is up-to now really the field where I say we're all more passionate than the excitement for me.

00:03:52: You talked about the early years and you know growing up in that world as a consultant around big data, particularly set into finance while this was already underway from two thousand ten onwards.

00:04:03: In a major way he had it.

00:04:05: twenty fourteen fifteen growth of the IOT movement connected systems and then industrial IOT these are all layers on that journey.

00:04:14: to get your point where can use systems Let's say AI, where there was machine learning in those early years through to now when people are building agents.

00:04:22: To help us make better decisions In this journey do you think why is the AIS wave that we're seeing?

00:04:30: Now Why Is it different to what We had with big data and IOT and IIoT And all Those other build-ups?

00:04:36: to This point Yeah

00:04:39: i mean The most likely answer Because It's not rule based anymore seeing how the AI acts with data and gives answers, it's just like you feel a bit like your working together.

00:04:51: They're human right?

00:04:52: And this has not been the case even if we had big data and mass data structured unstructured in Big Data Lakes.

00:05:00: If You Were Just Working With Descriptive Statistics or Machine Learning Even Yeah!

00:05:05: This is for sure also why this excitement was so big when they started.

00:05:09: I've never seen technological development around data since the internet that causes so much enthusiasm at customers and consumers seeing how these chatbots, now the agents act on behalf of us.

00:05:25: That is for me the main difference right?

00:05:27: Can i ask you personally if your working in a scene agentic or AI in commerce business What tools do you use on a daily basis?

00:05:37: And what are you building for yourself, and your teams.

00:05:40: I mean there many areas also when you look especially at our whole company.

00:05:46: For me personally There's very down-to-earth use case.

00:05:50: that is most important to me.

00:05:52: It's challenging by AI.

00:05:54: I think the two ways of doing it.

00:05:55: You can have white blank sheet and ask AI give me draft this side Do not too often.

00:06:00: so i still like start my own but then get challenged And this helps me to that.

00:06:05: they are still my personal ideas.

00:06:07: My experience is fully in, but it's kind of enhanced by the AI.

00:06:12: if I look at our teams and also more younger consultants and also data scientists we have...I think there even more like PowerPoints prepared market research has done minutes written by AI This for sure where consulting changed tremendously develop solutions for our customers, they use AI and this has caused a tremendous productivity gain there.

00:06:40: So that is even more obvious.

00:06:42: Yeah it's fascinating you know.

00:06:44: so I want to go back something we were talking about before the evolution of AI coming from machine learning.

00:06:48: And You Know i've had so many diverse conversations and I feel like I can't get a clear answer from anyone About what the rules are with AI?

00:06:56: What do need in place?

00:07:00: Give you an example, I speak to a lot of people who come from Middle-Stan family owned business and they'll tell you.

00:07:08: They don't know where it go with digitization.

00:07:10: They haven't even finished digitizing their processes in connecting the systems.

00:07:14: that only how does start with AI on top?

00:07:15: That's one problem

00:07:17: group.".

00:07:18: And then had a fascinating conversation as in Sweden couple months ago when i spoke to CTO of Nordea and he told me, if you haven't done your homework.

00:07:26: If you don't have good data governance or if you've got a lot of structures in place from the beginning then you won't be able to work effectively with

00:07:35: AI.".

00:07:36: Then I speak certain founders that spoke at sea level from loveable while I was in Sweden.

00:07:41: they said," That's bullshit!

00:07:43: You can just build and connect services over top but we do not need this structure is

00:07:46: in-place!".

00:07:48: There are so much chaos there no one who could really say What the rules are, how do you start out?

00:07:53: Are there rules and does it depend on the case?

00:07:55: what's your take of all this.

00:07:57: I think that is to a certain extent for sure true And its right combination with what he said.

00:08:04: so for sure You need framework.

00:08:07: i always recommend my customers start With a lean strategy but we have clear direction.

00:08:12: Where other potentials related To business model or specific processes The way you manage Your business And then digest what are the right technologies, What data do you need?

00:08:25: What other right models also for you.

00:08:27: How could further develop or Do You have to further develop your organization To gain all that value?

00:08:33: What governance structures do you Need not just for AI Also for Data For The Processes For IT Systems Underneath and last but not least Empowering Your People.

00:08:43: Prepare For A Change Process and change management that addresses the specifics of AI.

00:08:50: As I just said, AI causes enthusiasm but it also causes a lot of concerns.

00:08:55: this needs to be addressed... ...to get the whole AI movement going in your organization And then you need on top good AI value management.

00:09:03: In the end It's not about AI as itself for AI sake Its having AI to create value.

00:09:11: These are ingredients kind of have proper setup.

00:09:15: On the other hand, you say it's just switching on some services and its going.

00:09:18: I think there is also a lot of truth to that!

00:09:20: Have this simple rule when companies ask me how do start as they started with chatbot?

00:09:26: And then go to agents or automation?

00:09:31: These are three steps.

00:09:34: You start chatting using standard tool no matter from where if Google or Microsoft or Anthropica.

00:09:40: So yeah.

00:09:41: Then next step is combine it.

00:09:44: Look with your data you have in your company.

00:09:47: With your own IP.

00:09:48: This is where?

00:09:49: You make a difference to all the other companies combining large language models with your own Data, and then you have the basis to go into agents And go for much more higher levels of automation and Automate even full process chains and that it's actually quite simple here.

00:10:07: start with chatting Go-to-automation go two agents and therefore It's bit of both but you can Act based on such simple guiding principles.

00:10:17: You first need to establish a holistic framework and then you're good to go.

00:10:21: I really like that analogy around, beginning with what's simpler allows you be general.

00:10:27: That is obviously chat pot of some sort And i think if u look at when ive here when i speak Im engaging all the speakers for our program For tech coming up in just two months time.

00:10:39: So many people tell me they have.

00:10:42: They have probably, let's say forty percent of the projects that they know about in a company.

00:10:47: That are on the record.

00:10:49: someone has got oversight off and not many of them were actually paying back in terms of e-bit.

00:10:55: Let us put it this way And I think there is been lots research with this number.

00:11:00: between eighty five and ninety five percent of pilots don't payback in long term.

00:11:04: So i like that idea starting simple With with this aspect, what's the approach to pilots?

00:11:11: Because this is a danger I hear from.

00:11:12: A lot of people that are just using AI in use cases and in ways that maybe not be kind of safe for the organization or at least on the radar.

00:11:22: how do we deal with them particularly regulated industries?

00:11:27: That's absolutely very crucial topic.

00:11:30: And to me first step is also accepting that AI smaller cases that in combination and as a whole create a full benefit.

00:11:42: It's not like with classic IT where you have, as a DAX Forty maybe four D major initiatives they all big They all had for clear business case from the very beginning.

00:11:51: there were clear program management.

00:11:53: They run for three years And then they're done.

00:11:54: yeah With AI.

00:11:55: I just talked to a CIO and she said She has over three hundred nearly four hundred AI initiatives running That are already visible.

00:12:03: that also reasonable at the very top of the company all quite small.

00:12:08: And this is a very challenging aspect about AI.

00:12:11: and first step to gain value from here, it's to accept that you need to manage what I like to call a seek-and-find picture right?

00:12:18: Having these several cases out there where improvement is done What is crucial... You still have direction.

00:12:25: It's not like giving just some tools to the organization and then everybody creates its agents on their own behalf.

00:12:33: You still need to have direction, you still need digest.

00:12:36: what are major strategic use cases or process change that needs to be optimized?

00:12:44: And this direction should come top-down.

00:12:47: This is something which happens right now.

00:12:51: Then we need clear about value drivers behind those cases and not just monitor have invested this amount of money.

00:13:00: And I expect this return, you need to be clear.

00:13:02: what would drive the value saved working hours reduced throughput time or by new business models customer satisfaction?

00:13:12: if your AI cases are about better business management it might a bit more complicated to find the right drivers, maybe there's just some anecdotal evidence.

00:13:20: But even if you have just anecdotal evidences it is still evidence and we should use that to digest.

00:13:25: where do they really create impact?

00:13:27: And look another thing I'd like to talk about because as you mentioned, unlike what you talked about how you begin this strategy Do you have concrete example of an individual company?

00:13:37: but when came in with lack clarity people weren't sure There was cultural challenge for adoption fear of AI and how you unpack that in terms of an ongoing journey.

00:14:11: bring all to the table and have all their thoughts in ideas.

00:14:15: And at the same time not spend hours, or is it not spent years on talking down to people?

00:14:20: We found a very smart and also motivating facilitation mode too.

00:14:25: do workshops come up with clear idea but still having time listen them hear they experience this.

00:14:35: That was for me seeing Coming from this huge company all these construction site managers and so at the same time white collar workers coming together Digesting on an AI strategy, and it worked.

00:14:48: And It was really like they were All taking their thoughts bringing in ideas we actually done after some weeks only.

00:14:58: that Was for me an example how you can address This second fine picture In a quite straightforward manner.

00:15:06: I like it too because, obviously the people who have to deal with on a daily basis and often they'll get an outcome of those AI-driven processes as well.

00:15:15: It's top management that wants to push forward in terms integration.

00:15:19: Look another classic you see in new projects You've got company A new CTO.

00:15:25: let say New CTO comes In.

00:15:27: she is brilliant.

00:15:28: She has an AI pilot project.

00:15:30: He isn't successful As everyone had hoped or maybe there wasn't the buy-in from management, and this causes failure.

00:15:39: How do you integrate to the next AI pilot?

00:15:43: It doesn't just get killed in terms of having a culture of negativity within the

00:15:48: company?".

00:15:49: First it needs clear transparency about why it has not yet been adopted because the algorithms are just not capable in doing this task, then you maybe should stop that use case.

00:16:05: If it's... Because data is not yet at equality or at availability and needed to be successful than we need some measures to improve our data.

00:16:15: And there's a change aspect which I think most cases is the strongest challenge Here.

00:16:24: its really finding out why people for whatever reason do not adopt to that use case.

00:16:33: And having the right tone from the top and listening to people, finding ways together as a team to make this use-case still being used in processes by addressing what are the reasons why it's not yet adopted?

00:16:50: This is my experience – the most trickiest part!

00:16:53: Yeah, I think it's interesting thing because you know i've talked quite openly on the podcast in the past about some of the real fears that are around AI.

00:17:01: Because ten years ago an automated process and a factory can replace someone who puts screws in whereas now an automated processes can replace me.

00:17:13: so its'a real fear.

00:17:15: how do we balance this?

00:17:17: the need for humans, because we have a certain amount of wisdom and knowledge in our organizations to be able do business.

00:17:23: We need better look at each other and sit down and have

00:17:25: coffee.".

00:17:26: These are things that AI as you mentioned... You're constantly on the road because you need to be with humans but same time many processes know more than

00:17:39: me!

00:17:40: best ways to talk about it openly and best in person.

00:17:44: As he just mentioned, this is also maybe a reason why I travel that much.

00:17:48: In many areas the concerns can be addressed in way that's fine for everybody For sure.

00:17:56: you reach higher automation levels with AI And we reached higher automation level in areas where five years ago no one would have expected That We Would Talk About Automation Here.

00:18:06: When You Now Talk To Middle Management some of your tasks could be automated soon.

00:18:11: They have never been expecting that, right?

00:18:13: In many cases it is still clear there are other and rich task to do so they can also become a positive perspective about that

00:18:23: for you.

00:18:23: I mean as i said at the bleeding edge of implementation of AI in business right now.

00:18:30: where do you see it heading next?

00:18:32: five or ten years?

00:18:35: first further develop in the same pace as we see it last three years, which is already a challenge on its own.

00:18:45: I'm sure that this agenting AI will really allow for automation levels and areas where you have not expected them.

00:18:59: Diverse different customer types and they never managed to bring a high automation level in there with the standardized process within one system.

00:19:07: And already gave up some years ago, right?

00:19:09: Now they say well we can have that diverse different processes in order to cash process still automated And I think these are areas where we will see a lot of change.

00:19:22: Where also the classic ERP approaches, what would be changed?

00:19:26: We'll have to see combinations of classic ER pre-projects with agentic AI.

00:19:30: AI become kind of commodity and maybe just as normal as surfing on the internet... ...and that more people learn really how to use this agentic air in their daily work.

00:19:45: There's one thing we already talked about, which is for me like make it or break part of it all where I'm not yet fully sure how will end up.

00:19:54: But We need to find good solutions for data Not just for data availability and data quality also the whole Data privacy topics And these are aspects that to me you're not yet Fully solved.

00:20:10: There are good solutions out there.

00:20:11: For sure companies use AI, they must be solutions for sure.

00:20:15: but here I see also still a challenge where i'm not yet fully sure how all of it will be solved over the next five years.

00:20:23: and you in your work Matias.

00:20:25: we'll see an automated Matthias Emler In The Next Five to Ten Years And You Can Sit On Your Balcony And Drink A Beer.

00:20:30: Or Is It Far Off Into The Future?

00:20:33: Looking at my reality currently, and I mean AI is out now for already a couple of years.

00:21:03: But going into transformation, we're still all humans.

00:21:07: This is the moment where customers' consultants come together and drive that change.

00:21:15: here.

00:21:15: I'm pretty sure this won't be done by an avatar And this is also looking into my team, and to all the Horvat colleagues here where we are so much excited about.

00:21:27: To support different industries out there that different major players up there to develop in for that.

00:21:32: I think what's going on with the future of the next year will be about us

00:21:36: as well?

00:21:37: Perfect!

00:21:37: Well look it has been such a pleasure having you today.

00:21:40: That was Matthias Emler partnered at Horvath.

00:21:42: You can actually catch him at Tech from the thirty-first May till second June In Heilbronn tech website.

00:21:50: The program has just gone on so you can see all of the program sessions on our website, tech-europa.org.

00:21:57: Go and check it out!

00:21:58: And I look forward to seeing you at Tech end of May beginning of June.

00:22:01: Thanks very much for being with us Matthias.

00:22:03: Thanks

00:22:03: for having me, Dave See ya soon.

00:22:10: There was Dale talking to Matthias Emler.

00:22:13: As always we'd like to close this podcast With our TechLounge segment where bring in voices and questions from community questions we've gathered ourselves, from our audience and sometimes even within the tech team.

00:22:25: So nice to see you Dale again!

00:22:27: Thank-you so much for your time.

00:22:29: Let's see two words...

00:22:30: So let me ask you talking about AI and the pilots and the scalability of all this how do make sure at Tech in Heilbronn that were not only showing demos on stage but also bring scalable content to our viewers and to the audience?

00:22:46: Yeah I think it is an important question.

00:22:49: First, this really goes down to the key of our DNA at tech.

00:22:53: And we want make sure that... We don't just talk about perspectives from industry itself, perspective is just from research but actually combine it together.

00:23:01: so as part of the Tech Programming and we have leading researchers from professors like Professor Neil Lawrence from Oxford University world-renowned AI researcher, you know professor's from HEC Paris, Israel University professors from diverse areas where they can bring in their theoretical knowledge around AI and make sure that the approaches we're having actually makes sense, in terms of long-term development.

00:23:25: The second part is having industry involved.

00:23:27: so if course when you focus very heavily on people who have practical experience We are putting those people on stage like CDOs CIOs, CTO's for major companies Who give us there insights into what it means For their daily business.

00:23:40: And finally as we had from this podcast just now, insights from consultants who are dealing with it every day and then give us the best practice examples of how we approach these in our businesses.

00:23:52: All right thank you.

00:23:52: so AI is a big topic.

00:23:55: I mean there's also another.

00:23:59: already saw the agenda online.

00:24:01: see that we have a house of defense this year.

00:24:05: I mean, geopolitics is big in the news.

00:24:07: it's also big... In all business aspects.

00:24:11: so what exactly is the House of Defense serving at our tech convention?

00:24:16: What does the aim of this?

00:24:18: Yes i think its great.

00:24:19: you asked because your intention isn't just to put on yet another defence event Because of course defence has an issue growing importance in our economies and the defence events sector is growing fast as well.

00:24:33: But we want to create for tech, We know that we've already got leaders from middle-stunt family owned business SMEs who come to tech to get informed And many of these people right now.

00:24:44: I'm often at meetings where i meet owners founders of this businesses Who are saying I'm actually looking laterally right now.

00:24:52: The economies, you know if we look at the auto sector it's not really booming and we're seeing a real crisis for the auto-sector at the moment And they are looking to ways that can diversify.

00:25:00: An obvious candidate is defense.

00:25:03: So when talking about of course putting leaders from the Bundeswehr on stage so that contractors from the big defence prime companies can sell products and services We talk about putting in mixture of defence people, defence companies those smaller family-owned businesses, they can actually get involved and enter the defense economy.

00:25:23: How do you get onto the procurement list of these companies?

00:25:27: And how do make sure that your business is ready?

00:25:29: How to transform current operations into something which could feed in to the defence economy

00:25:34: itself?".

00:25:35: Yeah!

00:25:35: So we are honouring this approach off the economy... They want to diversify them and also have a diverse interesting program we can offer to everybody who comes to Heilbronn, looking forward to this and I see you there.

00:25:51: And thank you so much for your time today Dale!

00:25:53: Thanks very much Merz.

00:25:55: That's it for Today.

00:25:56: Thank You Very Much For Joining Us On The Optimist.

00:25:59: If you enjoyed This Episode We'd Love Your Feedback on Spotify Or Whichever Platform You're Listening From.

00:26:15: Thanks for listening to The

00:26:17: Optimist.

00:26:18: For more insights

00:26:19: and updates

00:26:19: about

00:26:20: tech, visit us at tech-Europe.org.