AI Won't Fix Your Business. Your Data Will.

00:00:07: Welcome to

00:00:07: Hobus, not a strategy and Sasha.

00:00:11: To your podcast.

00:00:12: we're doing the dual live stream podcasts.

00:00:14: I would say it's The Finance data and

00:00:18: people

00:00:18: podcast also from from.

00:00:19: but yeah so my name is Christian underwood.

00:00:22: i'm founder and CEO of strategy frame AI and host off hobus not a Strategy.

00:00:27: And I'm welcoming Yeah two other AI founders or AI consulting founders, I should say.

00:00:36: Dennis Schütz from Strategy Bridge AI and Dr.

00:00:39: Sascha Hagemüller.

00:00:40: welcome!

00:00:41: Thank you very much Christian.

00:00:43: we're very excited for this.

00:00:44: yes absolutely so.

00:00:46: it was your idea that we do that and you invited me.

00:00:49: So thank you very Much That i can now ask the questions but just as brief introduction Can can both of you give me just an elevator pitch off what your company is doing?

00:01:01: Just in a brief sense and Then then I will explain how I fit into that.

00:01:07: So and Dennis, please Please you could start

00:01:10: sure.

00:01:11: so You could say we are a global database, but with an analyst on top.

00:01:17: So what we deliver is data any form of data on any given company in the world.

00:01:23: you can see.

00:01:24: so starting from financials ownership date contact data whatever And on top of that, we are allowing users to do real research on markets.

00:01:36: So generating long list right market studies but also company studies.

00:01:41: I think that's what standing out the most at on a given company.

00:01:45: you can generate benchmarking analysis multiple valuations whatever your needs and in audit grade.

00:01:52: i think That's The thing.

00:01:54: so we're targeting high quality with What We deliver?

00:01:59: not only data, that's what strategy bridge is about.

00:02:02: Yes?

00:02:03: Thank you Sasha!

00:02:05: Yeah thank you very much Dennis and Christian.

00:02:08: I always say we are tech-enabled finance transactions in value creation services boutique.

00:02:13: so yes we do consulting in the PE environment along with a deal cycle of private equity investor on any finance related topic.

00:02:23: but our USP is to bring together CPA backgrounds transaction expertise and a data engineering in the AI side of things.

00:02:34: And that's pretty unique on the market.

00:02:37: to have both, yeah, expertise is sitting at one table and then really implementing projects and solutions for our clients instead of just painting beautiful slides.

00:02:49: so that's little bit the approach that radial follows.

00:02:51: Yeah,

00:02:52: great.

00:02:53: And I'm in between both of you.

00:02:55: so Dennis is helping to acquire the companies?

00:02:59: You're helping sell... And we're doing the strategy in between.

00:03:03: No one really cares, but in the end should care.

00:03:08: so I

00:03:09: know a couple of clients from you Christian and they do absolutely care.

00:03:12: i can tell

00:03:13: yeah Absolutely Yes.

00:03:16: We have a topic today and it's a little bit provocative.

00:03:19: um...I'm not sure if The Livestream on LinkedIn is already working But anyway..we will put that On our podcast streams So You Will Find It Anyway.

00:03:31: So we have the title of Why We Still Hire in The Age Of AI or should say why you are, both of you still hiring.

00:03:39: I'm not hiring at the moment.

00:03:41: and how AI is rewriting the M&A deal cycle?

00:03:47: In every investor deck that has an AI slide almost none has any result.

00:03:53: Why?

00:03:53: that's the question.

00:03:54: and yesterday?

00:03:56: The AI company deeper from cologne announced a cut of I think one-fourth off all employees.

00:04:05: So they're around thousands, and there cutting two hundred fifty to become an AI first company also in their operations.

00:04:12: So the first questions goes to you Dennis.

00:04:16: so You still growing still growing in your team and shipping AI for M&A.

00:04:21: What does that tell us?

00:04:22: That you're still growing in team.

00:04:26: So, I mean speaking of Deepel and all the other news on the news floor where they now cut people because of AI... The question is do we really have to do it with AI or did they over hire in the past as a good chance for them to get rid off people?

00:04:47: Because I think the reason why we still hire is very simple.

00:04:50: Two engineers with AI are still faster than one engineer with AI, right?

00:04:54: That's the...the very simple math that we have and what we're doing it not just punching Claude or whatever in face of our clients.

00:05:04: for that i would definitely need not an engineering team But what we do is different, right?

00:05:10: So AI is enabling us for sure.

00:05:12: It's making us a lot faster without a doubt but the hard work still.

00:05:16: for instance data engineering.

00:05:18: Right it very difficult in this age to get high quality data Without intersection of humans.

00:05:25: so our Data Engineering team obviously growing.

00:05:28: We need to fulfill highest quality requirements there And then also, maybe that's more a general topic of AI.

00:05:36: But the question we have to ask us is not only do we make our clients faster?

00:05:41: That's probably the simpler take but more important can we make them even better and Also there?

00:05:47: so when we innovate When we deploy new features There are still tons or things To be done outside The typical AI bubble Generating new data points.

00:06:00: Again, what I mentioned in my first words on business analysis and audit grade right?

00:06:06: This is not something that would simply come out of AI.

00:06:09: so we need to still work a lot with financial mathematics etc.

00:06:13: Pp end.

00:06:15: as the growing company And most probably in the audience would agree We are still an A people's business rights.

00:06:20: also your sales team still needs to grow work with outbound sales to not have first person contact with your clients.

00:06:33: So I think these are the heavy hitters and obvious reasons why you would still hire as a growing company.

00:06:39: Yeah, on LinkedIn i experienced that for Outbound You can do everything now with AI but all of us in really trust environment, so trust first and there I think you need real people just to get that done.

00:06:57: But Dasha how does it look like with your team on the let's say at end?

00:07:02: On the other side of the Deacycle?

00:07:04: Yeah It is actually similar.

00:07:06: if we would be a consulting firm That would operate globally already And have research teams With two thousand employees for example then Of course they will have an issue too.

00:07:16: but i think Lucky enough that we founded this business four years ago, and you're roughly thirty five people now.

00:07:23: And still have a very low overhead for the size of our team in order to question is rather okay.

00:07:30: how can be increased output per team member?

00:07:32: it's the same what Dennis basically mentioned But the output requirements are growing way faster than I could hire people, and they still need to hire people.

00:07:43: And actually it's a similar issue if you go into a portfolio company of a PE for example after closing off their business... ...and its all about getting this AI strategy working in finance or around controlling reporting stuff like that?

00:08:00: The first question comes up is where does data come from!

00:08:04: And the issue is that many of these companies have legacy systems, they've old systems.

00:08:13: The trend was also in the past.

00:08:16: for every problem that existed in a company new software was implemented from very siloed perspective.

00:08:24: and then when you go towards value creation or toward an exit all this data pools don't talk to each other.

00:08:31: Yeah, so you somewhere have a pricing table and at the same time.

00:08:34: You have procurement prices but again do want to contribution margin automatically?

00:08:39: But these two data sources are not talking to each other And they cannot be accessed automatically.

00:08:46: So um A lot of stuff is happening around The Data Engineering as well As Dennis mentioned.

00:08:52: So we need people that actually Are able To bring the data from legacy systems into a central data warehouse.

00:09:01: And nowadays it's still difficult to tell Claude, hey please create a VPN connector into that system without talking to the IT department or potentially external suppliers for these things.

00:09:15: So eighty ninety percent of actually getting an AI strategy going for our portfolio firm is still The Data Foundation and this will also be the case regular medium-sized entity in Germany that you might be aware of don't have a controlling or don't Have sufficient data.

00:09:36: So either the data needs to be generated order data is somewhere and it needs to Be accessed, and only if that Is yeah taken care off then AI can reach its full potential because It actually meets The context of the data.

00:09:50: otherwise You always have to feed it again And at the end do the same mistake that you from a siloed perspective And create your own AI tools and then ten years later you have the same issue nowadays with all these heterogeneous software solutions.

00:10:04: So companies are pretty much on the verge right now to make the same mistake again, To be honest!

00:10:12: I think it would very good if they wouldn't this time... ...and think through a little bit more in depth.

00:10:20: Yeah absolutely It always comes down to processes & data when you want to enable the AI systems or any other digital system.

00:10:31: So, and what we are experiencing?

00:10:33: that because we have a hybrid solution for that so they're bringing in their data where not... Because were in strategy process.

00:10:43: after doing this on one side of the data from Dennis and his team ready hand for our corporate clients They can use it.

00:10:51: We also crawl market data.

00:10:53: That's really helpful.

00:10:55: But answering really simple questions, where do you earn your money?

00:11:00: And what are your customer segments.

00:11:02: So a really simple question's they're not prepared for that normally.

00:11:08: or What are your top five customers?

00:11:10: so therefore You're implementing then the system on more sustainable way.

00:11:15: you have more time and I think therefore the funnel coming from Dennis going through this strategy process, then end up in your system also is really helpful for clients.

00:11:28: And get them more clarity!

00:11:30: That's pretty important...

00:11:34: Yeah, yeah funny note on that is I Think if you look at LinkedIn Sasha as the face why we should actually do this.

00:11:43: I think every month your posting link Sasha Our new join us and like five new faces Showing up right so?

00:11:53: Uh-huh i think.

00:11:56: just a side note Very typical for what we're talking about now.

00:12:00: okay

00:12:01: crazy

00:12:04: Yeah, I think we're struggling a little bit with the live stream.

00:12:07: So it's our first livestream.

00:12:09: so i have some different data information.

00:12:12: from one side they are writing its not life and other part is writing that it IS LIFE!

00:12:19: Take me a little bit into sourcing.

00:12:29: So what actually really changes when I get serious and not, you know the hype cycle?

00:12:36: And so... When i come to corporate clients they're telling me now oh we have now co-pilot!

00:12:41: We do that for the CEO office or analysis part etc.

00:12:45: with copilot.

00:12:50: That's better than nothing, right?

00:12:53: So for sure that very nice AI can help in lots of ways.

00:13:00: Instead of doing a Google research, want to know the ten most important companies in a segment.

00:13:05: The quality you would get from any AI system will be quite nice.

00:13:10: I can tell that especially because we also started in the pre-AI era right?

00:13:15: In the beginning of twenty twenty and how We came up with this at that time was majorly centered around keyword search Which was okay, and you've got something right.

00:13:29: but probably the so-called false positive rate.

00:13:31: So companies showing up that shouldn't show up.

00:13:33: Was sometimes over fifty percent?

00:13:36: So you had to look at three companies to get more or less certainly a good hit.

00:13:41: And what changed since then?

00:13:44: I think art in technology wise.

00:13:46: two things.

00:13:47: first of all We now have more power in structuring text data which is powerful.

00:13:55: do is when we get a company in our database, what's the product service of that business?

00:14:01: What are end clients?

00:14:05: What is the where in a value chain?

00:14:06: as to business?

00:14:07: This is a manufacturer.

00:14:08: Is it a distributor or both and also at your business model.

00:14:13: so very important data which we can structure, And then the second advantage of course semantic search.

00:14:18: So that way instead of looking at simple words We have the ability to really yeah understand even though they don't mention the same words They're still doing the same essentially and That's where the power gets them.

00:14:32: The problem why i would especially like users that do resourcing is very important.

00:14:38: so getting a list of companies you want to buy or, Like if you use an AI system, You're still a strong slave to zero for instance.

00:14:55: Right?

00:14:55: So you will find the companies that want to be found and That's a big problem.

00:15:00: so how we solve it is simple.

00:15:02: We are doing that same thing in our database of fifty million businesses.

00:15:06: So we have a completely Democratized data set.

00:15:10: they can show that really every company that needs To Show up Really shows Up.

00:15:14: another Problem Is these AI systems are not built to generate long list.

00:15:20: right good luck of generating a list?

00:15:21: Of couple hundred clients.

00:15:23: that's about their calculation capacity works with and that is That will be also at least the midterm problem With ease.

00:15:33: so years ahead until you can get high quality lists of hundreds of companies again in a local environment in a dedicated setup where we can find you in calculation costs such that we can generate it with sensitive cost, let's say sensible cost rather.

00:15:51: I think that is where the power kicks-in of still niche solutions like ours.

00:15:56: Yeah great and i just hand over to Sascha because always get a question Is AI ready for some finance calculations?

00:16:07: I had talk yesterday.

00:16:08: someone said AI cannot do this.

00:16:11: So and there's so many myth out there that they do not believe it can simply do one plus one thing.

00:16:20: so and you're doing in finance Danny is doing really great calculation.

00:16:24: so Sasha your statement on dad please.

00:16:27: It definitely can.

00:16:28: I mean i still stick to what if mentioned before the database of a data context needs to be provided.

00:16:37: so this is definitely the truth where I say whether real power lies.

00:16:42: but, If the data is provided in a proper way, or if you for example need an integrated financial model.

00:16:48: Or business plan based on drivers or assumptions and nowadays in twenty-twenty six start to set up a manual Excel file honestly your like looking backwards.

00:17:02: so no financial modeling project needs to be done completely manually nowadays.

00:17:10: If I look at Claude, it can create integrated financial models and i did that like a couple of months ago already And then I was already laughing as a finance guy because in the first run The balance sheet didn't tie up so.

00:17:29: But then it actually corrected itself and said, oh the balance sheet is not tying up.

00:17:35: And has the same amounts on each side so we need to correct this?

00:17:39: Then they go through with stuff like that especially within financial modeling.

00:17:44: also due diligence.

00:17:46: We already do a lot of things AI based.

00:17:50: That doesn't mean you have tremendously.

00:17:54: Let's work around it, but you can actually go deeper and actually create the insights that are more interesting than just paraphrasing.

00:18:01: The annual reports and trying to use or used a budget for manual data work, How is that business creating money?

00:18:16: What are the verticals that are worthwhile to consider.

00:18:21: Can the business be cut into specific verticals, so that a company doesn't look at yet and stuff like that And how it's actually money flowing through their businesses?

00:18:33: The financial diligence can be created much more commercially also that you really understand the business.

00:18:41: Look at verticals, look how money flows through a business and what value actually is created by this business that we want to buy.

00:18:49: so it creates room for us even deeper than before.

00:18:59: I would definitely say that AI can help in this regard.

00:19:04: Of course you need to look into it from a human perspective, from an expert's perspective but not using it as like... ...not an option anymore for my

00:19:16: perspective.".

00:19:19: Not using it, but also like it was in the past.

00:19:23: I would say not every axle sheet Was perfectly developed and without any mistakes.

00:19:30: So there have been people doing yeah failures And wrong calculations and wrong code i would say In the past in their axles.

00:19:40: so that's a way things are happening.

00:19:45: Maybe is

00:19:46: Exactly what addition on that?

00:19:49: Of course, completely agree like the modeling and everything that there was.

00:19:53: never did the hard part though.

00:19:55: Right?

00:19:55: So what is getting commoditized now?

00:19:58: as I think again something at a speeding you up crazy but um Like building the model populating it with the right data That was never the heart part The hard part.

00:20:11: that is getting assumptions, right?

00:20:13: So very well thought through assumptions on how you calculate a beta let's say.

00:20:19: How do the forecasting of the business which it not only function history but also your best take at moment in future market and I think this really where as Sasha explained when human still kicks really get trust on your assumptions, because that is a hard function.

00:20:40: And just getting the result by AI not knowing how exactly it came up with and you will never know... That's the problem!

00:20:49: Maybe adding in that regard especially when we think about financial models like this I see very high danger our generation has an advantage right now Because We built these models from scratch.

00:21:03: Yeah, late at night a hundred times.

00:21:07: So we know exactly how the result should look like and how an investor looks at it.

00:21:12: if I'm very young employee now when they just use AI blindly so to say not questioning what i'm actually creating that's big mistake in them very much.

00:21:22: you are concerned about future.

00:21:24: education will look like in consulting firms because people also need what they actually want the AI to create at the end, because otherwise you don't have a clue.

00:21:34: We'll see how that will show up.

00:21:37: but as you say then it's like really knowing what the goal should be and the result is driving your prompting style.

00:21:49: so I think still important to really think about that.

00:21:54: Also, when we think of our data warehouses and connect these with AI solutions probably half the time goes into the prompt explaining what do you find?

00:22:08: in which column is this huge database?

00:22:11: What does it describe?

00:22:13: where are issues or gaps like that?

00:22:18: The technological part isn't an issue but prompting correctly and feeding the details into this solution will actually enable a better result at the end.

00:22:28: This is maybe also where people struggle currently sometimes that the prompts are just not sufficient to enable AI do their job.

00:22:38: Totally different perspective from my side, because we're shipping solutions which goes directly educated and customers.

00:22:47: I would say, don't get me wrong.

00:22:50: They're really well-educated business people on our client side but maybe they haven't done strategy in their whole business life.

00:22:59: so therefore why we have real frameworks?

00:23:02: And it's pretty clear what happens with the data and our agents with a different tools are processing that Assessing the prompts that are coming and they're always not on the level you need them.

00:23:20: So we rewrite the prompting, direct user to different tasks help them also showing visually what end result should look like give examples and agents etc.

00:23:37: is directly presented in this format.

00:23:40: so even do it directly now And you can then export that direct because we do not believe.

00:23:48: That people will be how to say it become the best prompt us in the world.

00:23:55: so and, So doing strategy with voice, but it totally changed the game for people because they knew their sales.

00:24:08: They know that supply chain things.

00:24:10: so there are experts in really great fields and just have to talk when they talked enough context And then our tool should do the rest.

00:24:21: This is a little bit more of my perspective Because we also have consultants in certified partners.

00:24:29: But in the end, uh...in our perspective it should be done by the-by the end customers.

00:24:34: but for sure how to have a consultant in future years?

00:24:38: I think there will be total different skill set that they are focusing on and it would be different people also not as i'd say..i was never right finance guy.

00:24:50: so i studied political science at first study nothing to do with numbers then went into strategy which is a real finding thing.

00:24:59: It's just let say, twenty out of the eighty percent.

00:25:04: but I would say that makes it...I think there will be different people in consulting doing different things tomorrow.

00:25:13: so and therefore I see that is changing little bit.

00:25:17: But

00:25:18: your clients are running around now all saying make no mistakes instead

00:25:25: Absolutely nobody.

00:25:27: but when it comes to we have for example, We are also doing for nonprofits and its really funny.

00:25:55: Where do they come from and isn't that a real progress?

00:25:59: Yeah, it's good point.

00:26:01: I mean you do strategy as well.

00:26:03: so strategies always are broad term.

00:26:07: we have an AI officer but actually our AI officers someone who can code understand LLMs create workflows in agents is hands on person.

00:26:19: i think what?

00:26:20: currently?

00:26:21: the issue sometimes that there is a gap between creating, how should I say an AI strategy or an AI roadmap on the slide in Ivory Tower and Frankfurt versus then actually implementing it somewhere in your medium-sized or small to medium sized entity.

00:26:43: On country side where you have people fifty plus sitting at the finance department Yeah, coming around the corner and saying hey we are doing AI now.

00:26:52: And then following the path to say okay now everyone gets a chat GPT or Claude license?

00:27:01: Then you're suddenly an AI company which is complete BS.

00:27:05: yeah...to be honest We actually currently conducting study called Winning The Value Creation Game where we talk with approximately more than one hundred private equity investors about the whole value creation process.

00:27:22: And it's actually very interesting that because we have one question, that basically asks how high do you assess their AI potential in different corporate functions along like basically the port of value chain where your finance and then IT and HR and production procurement et cetera?

00:27:43: The potential is assessed very high in every area.

00:27:47: HR is a little bit lower, quite interestingly because they say okay there's still lot of human interaction and you need to create certain vibe but in every other department the potentials seem very high.

00:28:01: And then we do question after this saying ok interesting now let us know what concrete use cases are that your working on?

00:28:10: Then it was quiet at room.

00:28:13: yeah I think discrepancy Currently lies because again, the data context is not existing.

00:28:22: So AI cannot be really deployed and real use cases can't be really deploy if AI's not intertwined with every piece of data in their business yeah?

00:28:34: And just installing a few AI licenses here and there in their silos is not changing anything.

00:28:41: Of course, maybe a presentation is created a little bit quicker or an email is corrected or whatever but this has nothing to do with AI readiness or an eye enablement.

00:28:52: um In the business.

00:28:53: And therefore I think that uh That the AI thesis is not written in the investor's office The value around AIs really created on site In the countryside where the business is and whether changes with a change needs to happen.

00:29:08: And stuff like that, I mean i also had a conversation actually few weeks ago With an investor who said we wanna do or create an AI first company with one portfolio firm.

00:29:25: so The First Step We Actually Need Is A New ERP System Because The Current ERP Systems Not AI Ready Yeah.

00:29:33: So the step would have been to buy a new ERP system for probably five hundred thousand euros, which will need two years to be implemented.

00:29:45: and then you would have the same bad data in the same bed processes In a new environment.

00:29:51: And then this ERP provider would say now You can use our AI or whatsoever?

00:29:56: It's really

00:29:56: just not working.

00:29:57: no we're still not working.

00:29:58: so it's.

00:30:00: and these like false concepts around AI and which are also promoted by some AI specialists, so to say.

00:30:09: They're wasting money.

00:30:11: they are destroying value actually that you could really easily build with a central data warehouse.

00:30:17: connect your AI into the warehouse it doesn't matter what systems there are.

00:30:21: try to gradually improve the quality of your data connected to it then Build your agent and build you workflows instead of like relying on systems.

00:30:31: And stuff like that.

00:30:33: so they are really, really wrong decisions made right now from my perspective that destroy value rather than creating it

00:30:43: absolutely when it comes to let's say one function the sea level itself.

00:30:49: So what kind of solution do they really need?

00:30:52: And coming also again to, let's say M&A consulting boutique firms.

00:30:57: Dennis you're doing a great job and your team on that with out of the box.

00:31:11: So you do not need The Data Warehouse to build up because You're at the front end.

00:31:15: that makes it a little bit easier, but can you please tell us so how your building That kind Of data warehouse?

00:31:22: What Do you Need and what are the main challenges for the client accessing that data with the AI?

00:31:32: I

00:31:33: think thats the big benefit Focus right and focus on a niche as you said the M&A space.

00:31:40: so We know perfectly from A to Z. The whole process.

00:31:46: These experts need, they need data And what usually data means is of course it mean state on the single business or demand date, let's say.

00:31:57: So we need to make it as easy as possible that they get their data into our platform with all these security necessities right.

00:32:05: so nobody else will see your data.

00:32:07: but essentially that is the only time They would actually.

00:32:11: yeah I need to put something outside in and throw a platform and the rest this Yeah, our job.

00:32:17: I would say so when you give in that company name as a set the When we started with the sourcing topic and they sell sites case We would find Through our database all relevant buyers if we should miss one because usually wouldn't know That they're doing something in that space decline could easily add them.

00:32:38: but Usually what they will say is fantastic?

00:32:40: All Relevant bias have been found by the platform.

00:32:46: And then yeah, they need the contact data needs to know how to contact them.

00:32:51: and Then of course generating the outputs.

00:32:54: The marketing material let's say so from teaser two info memo.

00:33:00: Historically, what we have done is generated.

00:33:03: the heavy hitter analysis parts of that.

00:33:06: What they use as gut was in their own template and design.

00:33:10: The PowerPoint outputs could just drag-and drop it into their slide.

00:33:15: now with next update will also take over generate full fledged PowerPoint report.

00:33:22: so I think big benefit for focusing on this right?

00:33:29: Never possible.

00:33:30: if no, I don't know.

00:33:31: another use case comes in whoever needs a PowerPoint presentation.

00:33:36: A sales guy from the random sales business make a pitch with us.

00:33:39: that will be difficult but as long as our client focus are part of our ICPs M&A by side audit they would get exactly results.

00:33:48: they need more or less at the click off a few buttons.

00:33:52: yeah absolutely great.

00:33:53: um Yeah.

00:33:55: and then Please Sasha, we come to your playbook.

00:33:58: So please show us a little bit.

00:34:00: so how does it actually look like?

00:34:02: Because I hear now okay cleaning up the data building Data Warehouse and then in a portfolio company It's like oh my god!

00:34:12: We've never done that.

00:34:14: What is magic source That you can make really happen?

00:34:19: Yeah i would say To Make It Simple Behind every kind of system.

00:34:25: There's a database in the databases usually just to table right or a few tables multi-dimensions whatsoever.

00:34:32: so The simplest form is probably like a CRM system for example Like pipe driver, what's ever where you basically have an Excel file?

00:34:40: In a web context up to SAP for HANA something way.

00:34:46: we have huge amounts different data sources and stuff like that.

00:34:51: So at the end, every ERP is creating data sometimes manually, sometimes automatically inserted into these systems And every kind of insertion or bookkeeping booking whatever it's one line in this database In order to create a warehouse.

00:35:18: You basically just need to access this database somehow that the system creates and make sure you pull the data automatically from that database.

00:35:30: Usually, you mirror it into a separate database overnight so your main system is not struggling with the request And so you feed a central data warehouse with that data.

00:35:41: Yeah, if it's an on-premise system You have like a VPN connection or something?

00:35:44: If is the cloud system at sometimes easier some systems seven direct API and stuff Like dad.

00:35:49: So we usually set up that Central Data Warehouse by basically pulling The data from these various Systems.

00:35:56: okay then like V lookup trying to somehow intertwine them.

00:36:01: yeah buy I don't know, a project number where you basically connect for example then time recordings with project calculations or stuff like that.

00:36:10: Or via product numbers that you connect the pricing table with procurement costs from the warehousing system.

00:36:21: so create close data warehouse which allows it at end to to connect data that hasn't been connected before, and this is actually the most complex or task with the most effort required.

00:36:38: And then you visualize that data can happen via Excel which happens in Power BI whatsoever.

00:36:44: we are very flexible on that regard.

00:36:46: Then usually see the quality issues because it gets transparent as soon as they get transparent For example, product cohorts not maintained and so on.

00:37:03: And based on the transparency you can then walk backwards in that process.

00:37:07: understand where is data governance leaking?

00:37:11: Do I have a CRM system?

00:37:13: but sales never takes care of certain fields because they are not obligatory or there are no drop downs, but free text fields for example.

00:37:21: And then we can think about the future data governance concept where you say okay these might be mandatory fields in the future this right rather via dropdown instead off a free text field and stuff like that.

00:37:33: So basically already have central brain of the company with all data in their which is consolidated?

00:37:41: Which gets cleaner over time which is updated once per night.

00:37:46: So you also look on more or less real-time data, and then you visualize that with Power BI or Excel for example And than have a view of your P&L life Your balance sheet, your cash flow statement.

00:37:59: You're working capital, net debt whatsoever.

00:38:01: But in the other side Also on operational KPIs where can dive into contribution margins Per product, per client, per sales channel, per region whatsoever and really make in-depth analysis.

00:38:15: And what we then do?

00:38:16: when this central data warehouse is created, getting cleaner and cleaner... We connect an AI solution where currently working with Langdok which is a GDPR compliant that gets connected to the Data Warehouse.

00:38:32: This actually our main solution.

00:38:35: A controlling or reporting guy can basically chat in natural language without using SQL statements or stuff like that.

00:38:46: So the solution basically translates natural requests into SQL language, gives back... The database gets back SQL language which then is again translated into Natural Language and you can ask them okay with Which clients are we creating the highest losses?

00:39:02: in which regions?

00:39:03: Are we selling the wrong products that we purchase goods and services, where prices got increased but we didn't forward them to their customer for example.

00:39:13: And stuff like that.

00:39:15: then you really have an AI based reporting and controlling it also value creation engine That can support You on On The Way To Create Value again All dependent on the database connecting, connecting.

00:39:29: The data warehouse with that AI solution and make it work is a matter of like two to four days.

00:39:35: to really make at work by creating the data warehouse in really making sure that the databases correct Is a matter off?

00:39:42: A couple of weeks rather sometimes month In order To Make Sure That This Is Working Really Running?

00:39:48: Yeah Okay And I found It Really Interesting.

00:39:51: So That Would Have Been My Questions.

00:39:52: How Long Does It Take You?

00:39:54: so Dennis Heaven out of the box solution with their own database that comes directly to clients in the M&A field.

00:40:02: So we are doing the hybrid mode Crawling data, but also using data from the customers and they're going to their systems.

00:40:09: And we say for strategy We just need that kind of data from your controlling and we connected with The outside data and you can chat with everything over there.

00:40:17: Also the project and the OKRs.

00:40:20: You are going deeper in let's say the Data Foundation building the lake.

00:40:24: how a data lake house?

00:40:27: AI on top of that to make it work in a longer run, I would say.

00:40:32: So also three really different use cases of AI but also usage of AI on Top AI baked-in as pure and here i think we see how consulting is changing in the field To the problem and to each problem.

00:40:54: There are different solutions, And we're trying to find The best ways that Are really possible?

00:41:00: So so yeah looking to look into the time I would like to ask now let's say Just one last question for both of you.

00:41:12: And then just one overall question.

00:41:15: We had for those not listening on LinkedIn now because they can't and there may be a listening in our podcast, so we're sorry that it the stream didn't start.

00:41:25: in the event?

00:41:32: with AI when you're working with software, there can always a bug.

00:41:37: It's not like a PowerPoint slide where we can control hundred percent of the typo and that everything is right.

00:41:45: so You have to be little bit more careful.

00:41:51: but also some things happen.

00:41:53: We are sorry for this But will ship it over our podcast streams anyway.

00:41:59: So Dennis overall Being in their journey coming from the banking world where you originally started Where do you think is it going?

00:42:11: For you, In your special field on the front and to generate?

00:42:15: so You're optimizing.

00:42:17: The work for the people there.

00:42:19: They are getting faster and faster data quality's increasing.

00:42:23: but how far can you go?

00:42:26: So

00:42:28: What does your vision?

00:42:28: I would say

00:42:30: our vision Amongst many is we are of course preparing for a world where more or less most of our users will be agents.

00:42:40: That's very relevant scenario which we prepare, so that we build our APIs and MCP coverage in the way.

00:42:51: whenever somebody doesn't want to do hand work get it perfectly visualized whatever they need but instead let an agent do what ever he wants And that we still feed them with the right not only data, but what you're creating on top of that is information.

00:43:08: Right?

00:43:08: So a valuation, for instance.

00:43:12: Nobody cares about the just all number you get out of it right?

00:43:16: You want to understand really how did they get to that number?

00:43:19: what are transactions or similar businesses that happened?

00:43:22: whatever got here?

00:43:23: What is like a size adjustment A performance by Adjustment off their business so To deliver information Really deaths.

00:43:32: I think world we're focusing on and there Of course these solutions survives.

00:43:38: That not only gives the easiest access today.

00:43:40: That is, of course important.

00:43:42: But also as you said the highest quality in data and this just dirty work sometimes that needs to be done in creating all these data sets, this pieces of information such that you can get the best possible output off it.

00:44:00: But there's also another scenario.

00:44:02: so I think... Also and will sustain for a couple years i'm sure is still have to use us tell their clients look we are humans doing that full expertise no database capturing my forty years of experience.

00:44:21: Also still for them, I think we are demanded to build the best possible user experience still for humans and ensure that they know each step what is happening.

00:44:32: They can intervene also That it's a very important part Still in the future right?

00:44:37: What if you're unsatisfied with data You get?

00:44:39: Is there A simple way To intervene?

00:44:43: big Important UX topic to also Deliver dead so that somebody With decades Of experience or Years experiences, usually also something worth mentioning.

00:44:53: They can still get exactly the outputs they need just a lot faster Just with access to information that wouldn't have otherwise.

00:45:01: and so what we're seeing is going forward.

00:45:04: you incredibly speed up in what percentage of your process can be completely automatized with a generic tool that is heavily increasing.

00:45:21: But then you still have the parts where details matter, everything belongs on the foundation as Sascha mentioned minutes ago.

00:45:31: And that needs to be on point and this is what we prepare for.

00:45:38: This essentially stands for us a business to live our highest possible quality.

00:45:44: Yeah,

00:45:45: great.

00:45:45: And Sasha from your side are you looking for let's say a fixed agent set when you're entering tomorrow to accompany and they were collecting the data in building the data warehouse?

00:45:58: Automatically.

00:45:59: is that kind of vision or something else

00:46:02: I mean?

00:46:03: i mean You can think about every scenario.

00:46:05: To be honest we are on high alert.

00:46:09: Yeah, be on the forefront of all the developments out there.

00:46:15: I mean i remember me saying half an hour ago currently you still need data engineers in order to actually create a data or to connect systems into the data warehouse?

00:46:28: If that is still the case for one year maybe you prompt an agent and say hey please contact me through the client's interface.

00:46:42: These are the access, this is the password and they use a name for it.

00:46:48: Yeah I'm sure that's might be... This might be the case.

00:46:52: And i still think there's lot of groundwork to done in industry out here.

00:46:59: There will also be manual work because usually businesses aren't as far away from us.

00:47:06: But but that is definitely a challenge we think about.

00:47:08: and also on the other hand like if you, If you think about financial diligence for example where you have a data room.

00:47:15: For example That is filled with with data by either target business And all of the data room Suppliers out there I'm sure they build their solutions for this because This might be new way Of doing business to them.

00:47:32: To be honest We're actually creating environment to do due diligence in the data room and all the due diligence might also be shifted into an earlier stage where I could imagine that a lot more due diligence is already done before the LOI instead of now after the LOE and all these things.

00:47:52: On the other hand, I have to say i'm an optimist by nature as an entrepreneur.

00:47:59: for consulting firm there's like us, maybe not the large ones but for a boutique firm there's an option right now to actually increase margins on our businesses because every product at the end has value.

00:48:18: So financial diligence that is required by the bank will always have price.

00:48:24: in one or two years this might be half of the current price.

00:48:31: But then the only question I have to ask myself as an entrepreneur is how many of these do I do more?

00:48:37: Yeah.

00:48:38: And if, uh, If i'm able to provide um...I don't know if the diligence costs like a third over due diligence nowadays but I do six times as many in the future and then increase my margin.

00:48:52: so for us that's also huge lever or huge opportunity to actually become more profitable in the Yeah, so we are pushing towards everything.

00:49:03: We try to change as quickly as possible and even challenging for us.

00:49:07: sometimes it's also challenging for the team.

00:49:09: I still have discussions where people come in say hey we cannot write that many proposals at the same time.

00:49:14: i said this is the wrong way of thinking.

00:49:15: The right way Of Thinking Is How Do We Create An Agent Or A Workflow or Process That makes us write proposals quicker?

00:49:24: or ideally that an AI automatically generates a ninety percent version of a proposal in the future.

00:49:30: That's the way of thinking, and it is to change what needs to happen.

00:49:34: And I think the biggest mistake right now for every company out there is wait For the solution that will emerge because It might be smart actually buy the tool In year which works best.

00:49:49: but The main issue that remains is, the change in a team doesn't happen.

00:49:54: And then the changing way of thinking doesn't happened.

00:49:57: and this changes only happening when you try out stuff yeah?

00:50:01: This sometimes against nature or for German human being but um... You have to try it!

00:50:08: When you tried test it also go into certain risks with it Then actually win at end because your teams ready to adopt the right tool in the moment it comes out.

00:50:21: So yeah, trying out being on the forefront of it paying someone in our team who is an actual AI engineer, actual data engineer maybe hiring more of them and trying to push that forward as quickly as possible.

00:50:35: That's the way to go.

00:50:36: but In reality I'm not in PowerPoint Absolutely.

00:50:41: And

00:50:42: then you also have the aspect of accountability, right?

00:50:45: I think that is something where everybody questions how will that evolve?

00:50:48: so we take full accountability for what we do.

00:50:51: a consultant especially with the bank case you made.

00:50:54: they are very interested to know when something doesn't work who's the one to be blamed.

00:51:01: yeah i think it was still persist and still needs this or saves our ass

00:51:11: Absolutely, so to bring it to the point.

00:51:14: It's just do-it with AI.

00:51:16: So I think that's really important and let me please share Just kind of our vision that we're doing And for me is having both of you also here in the podcast but Also a kind of that solutions In our solution?

00:51:29: We are talking about a strategic decision intelligence because That is in most companies not there The Strategic Decision Intelligence.

00:51:39: one person with a gut feeling, with biases etc.

00:51:43: And when we want to make that better... ...with better data and more clarity from outside in-data the data in the warehouse so on all those stages bring it together in simple form.

00:51:56: We will launch this on the first of June On your mobile.

00:52:00: So that's basically the idea to ask all of that kind question, having dashboards for the burning KPIs of your business.

00:52:09: But most of them today don't know what are their most burning KPis?

00:52:15: This is a way to go also for our clients and we're on journey now!

00:52:21: We do not where it will end but were sure its better start now or yesterday than tomorrow.

00:52:30: One last sentence to maybe, yeah let's say PE partners watching here this live remote podcast.

00:52:39: What is the move on Monday?

00:52:41: They should do Dennis.

00:52:45: Yeah reach out of course

00:52:46: if they...

00:52:48: We put everything in show notes for sure!

00:52:53: No but seriously I think Sasha put it in the right phrase right waiting is the worst thing you can do and usually if like time has money.

00:53:06: You have two options.

00:53:07: basically you can now try to start out what many do and I think for good reasons.

00:53:11: is the make decision right.

00:53:14: So, if they don't make the issue and we've seen that no more of a dozens off times it's when your start making.

00:53:23: he will know the actual result even though the impact of what are doing earliest months later right after weeks you never have something where your feel trustworthy and saying okay.

00:53:33: This isn't heading.

00:53:34: were we wanted to this has major flaws.

00:53:36: it will still continue to invest time right nobody.

00:53:39: psychologically we don't see some cost as humans write.

00:53:42: we continued to optimize and there is the, Advantage out there, right?

00:53:49: There are many tools.

00:53:50: They're many tool providers that focus on a single niche their handling which did it already successfully for hundreds of other clients where you immediately the second You buy the next day will have access.

00:54:00: we ever working solution.

00:54:01: That would work and so what we actually tell to our clients is okay fine Do your make please very happy with not standing in your way.

00:54:10: take our solution as the benchmark.

00:54:13: Take it.

00:54:13: That's right when the market delivers right now and try to make us obsolete with your Make Efforts.

00:54:22: And I think that is something, it would challenge everybody.

00:54:25: do not waste time.

00:54:27: start working solution right now.

00:54:29: we are very happy.

00:54:30: get into competition With a Make Effort.

00:54:33: i think thats the fairest you can Get It!

00:54:37: Absolutely great Sascha.

00:54:38: what Would You Say?

00:54:41: Audit Your Portfolio Companies Data Foundation.

00:54:44: i Think That's The Biggest Step.

00:54:47: Don't think about AI and an AI roadmap.

00:54:50: Think of how you get the data together in order to enable AI at the end, please do not The idea that an ERP system might solve your problems.

00:55:03: Yeah, it's really about the data.

00:55:06: and If you can work with a within IBM AS for hundred solution from nineteen eighty nine if the bookkeeping is clean You can still build on SQL database around it and digitalize the whole process to make it AI ready.

00:55:19: So so don't believe the usual pitfalls in PE.

00:55:25: Really think about the Data Foundation And then take Do it step by step and start implementing on Monday.

00:55:31: Yeah, that's uh...

00:55:33: That's great!

00:55:34: And just my last cent to it.

00:55:37: so hope is not a strategy?

00:55:40: You need a strategy but for strategy you need everything.

00:55:43: You need the data ,you need the processes .You need AI But you need people in your portfolio companies but also in your PE team.

00:55:52: So therefore I would say bring them in harmony and just start tomorrow.

00:55:58: So therefore I would like to thank both of you, Dennis and Sascha for this first live session we did.

00:56:05: And we just did it so... We made some mistake but will be better the next time that we do it!

00:56:12: Thankyou very much for listening and put everything in the show notes.

00:56:17: You can reach out all of us if you have any questions on LinkedIn.

00:56:22: we are always available looking forward.

00:56:25: Thank you

00:56:27: too, sorry to the audience.

00:56:29: Bye-bye!

00:56:29: The

00:56:30: strategy

00:56:37: magazine, hope is not a strategy.

00:56:41: What do you expect in our new magazine that once appears in the quarter?

00:56:55: Each one special focus topic where we really go deep into it will be strategic decision intelligence for first time then two to three case studies showing how companies with the Strategy Frame AI make real difference and also expert contributions interviews and here behind the scenes of Strategy Frame AI.

00:57:20: So if you want to subscribe to our magazine, feel free to contact us at www.hoffnungistkeinestrategie.de.