EP110: Turning Insights into Action with GenBl w/ Dr. Andreas Koeberl @ milo.ai

Hello and welcome to the Process and Automation Podcast with the Automation Guys. Yeah, Arno and I are back after a bit of a sabbatical, I would say. And the last, last few months have been really, really crazy around the topic of AI. AI is advancing extremely fast and, um, yeah, I think every, every week there are like hundreds of announcements, new language models, new tools. This is better. This is better. So yeah, plenty to keep up for everyone. So the hype around AI continues. That's, that's clear. What we see, adoption seems to be still very low. I mean, it's what we hear as well from our clients and what you read as well in the news as well. Plenty of AI projects seems to fail. So there was a big MIT sort of report recently, 95% of the projects actually do fail. Probably not the best report. It was just 52 companies they interviewed, but overall, I think the sentiment is there that AI, yeah, the adoption is very low and people are questioning where is the value of AI. So Arno and myself, yeah, we have always been very sort of non-hype actually. I think, yeah, being in industry for more than 20 years, so we always try to and still try to uncover where the latest technology brings that real value. Of tangible results for our customers. So this is why we like doing this, this job, because we implement these things with real projects, real customers, and doing that for such a long time. So yeah, where's the real value for businesses, employees, and customers? Yeah. After that short sabbatical, Arno and myself, we continue covering all these things here on the podcast, all low-code, no-code workflow. So everything which is really there. So AI agents, of course, these days as well. So, yeah, we will cover all these things to help everyone really out there to make these technologies work for them. So, yes, of course, today, again, won't be really different. We have a format on the show which is very popular. It's our interview episodes. So with that thought, yeah, with that, we thought we'd do one interview episode again today and bring a great guest along, a guest who is I would say from, from all the previous conversations I had with him, so very passionate about AI and automation and the topic, as passionate as we are. So without further ado, with great pleasure to welcome our next guest. He's the co-founder of the AI company called Autonomous Minds and building a great product called Milo. Welcome Andreas. Welcome to the podcast.

Thanks. Thanks, Sascha and Arno. Thanks for having me.

Yeah, Andreas, we are very excited to hear your sort of insights today and learn a bit more about Autonomous Minds and the platform you're currently building, the platform Milo, you mentioned you're building at the moment. To kick off, can you maybe share with our listeners and with us a bit about your background, what you do, and a bit more about Milo?

Yeah, happy to do so. I mean, thanks for the introduction. So yes, I'm indeed one of the co-founders of Autonomous Minds. I'm an engineer by training, but I think the last line of code I wrote like 15 years ago, so I'm way older than I look. But yeah, I mean, I started my career in aerospace, actually, aerospace and public safety, 19 years ago as a developer and then system engineer, spent a couple of years at Microsoft and 9 years ago, I finally started building my own companies, built quite a successful sportsbook operation 9 years ago, which exited in 2020 and Since 2020, I was leading and turning around a game studio. So I moved into gaming, which is kind of very different to what I did before. And yeah, a year ago, together with Philipp, my co-founder, we started Autonomous Minds. And Autonomous Minds has the big ambition to— we need to use diplomatic words and not our marketing slogan here, but we want to unscrew Analytics a bit, not actually a bit in a big way. So our big ambition is to build the decision platform of the future because we strongly, strongly believe that this is one of the use cases for AI. We're not alone on this. Funnily enough, two weeks ago, Gartner reached out to us. They're looking for startups in that space because the large incumbents like Microsoft, etc., are not really satisfying that sort of emerging demand because decision technology exists for years, right? There is nothing that is really, let's say, accessible and easy to use, two ingredients that you need for the diffusion of any new technology. And what Autonomous Minds does, to cut to the chase, is we basically built an AI data layer that understands what's going on in your organization in real time, tells you why things moved, and then based on that why, triggers the right action automatically, operational action automatically. So you can think about it like a bit of an AI COO, right? So while we are in the analytics space and in this new emerging field of called generative BI, it's more, let's say, merging into operations and into actually decision tooling or decision platforms because when you think about what actually is business intelligence, a lot of people mistake this or misunderstand it as building nice dashboards, right? And it's like you raise a ticket and then 10 days after you raise the ticket, you get like a nice dashboard and a chart back that tells you something you already knew. But what BI actually is, it should, its intention always was and is help you make decisions, right? So what we do is we don't want a tool that helps you make decisions, but we wanna build a tool that takes the decisions and actions on your behalf autonomously. Wow.

Yeah, that's, that's very innovative indeed. Because all the tools we, we, we, we know, you just mentioned, um, it is really about displaying all the, all the data currently available. And of course there needs to be some brain then otherwise behind it to analyze it. And, uh, you mentioned there's a good point with the, with the age of the data. And then sometimes they are actually, well, we work with some clients, they are taking the data from the data warehouse. Originally from this data warehouse and this data warehouse, and then they're producing a massive spreadsheet and then they're loading it back into the system. So you hear a lot of crazy stories out there to come to results. And yeah, I think one thing we talk here a lot on the podcast is then the actual use cases. So when you talk about this generative BI, where do you think is sort of the biggest use case for that? So you touched on this briefly, but maybe you have some some impactful sort of stories already or some, some vision, impactful vision for the product?

I mean, the, the big use case is always what we call, sounds a bit abstract, but it's really important, is this why, right? Because that's the, the huge differentiator. That's the huge value creation or where the value creation happens. Because traditional BI and analytics usually tell you the what, as we said, you get the dashboard, you see that. I don't know, your Q4 pipeline is trending down, but that's not what you want to know. What you want to know is why does it trend down? Right? Like, think about the user in any business, and I'm not talking about what we call the technical user, like the data analyst or the data scientist, like the person who requests certain data or an analysis or report, right? What they want is an answer to a particular question, but what traditional BI tools do and the BI departments do is often not really addressing the questions, but basically tell them, yeah, you were right. Your pipeline is really trending down or your user journey is really trending up. So why is that? Right. And then you start doing meetings and you go into dashboarding and you drill down and you do more meetings. And then days and days after you ask a very simple question, actually, you might get an answer. And by that time, very often the moment is already gone, right? Because Mm-hmm. Okay. So it's, it's, it's actually quite a funny story. I don't want to derail from your question too much, but we, we talked to 19 gaming companies over the last, over the last 3 weeks and 16 of them, so 84%, admitted that their analytics is a complete mess and the majority of them, like 11 out of these 19, said that the moment they get the answer to their question, nothing happens because it's too late anyways, right? So it gives you a bit of a feeling of how big the opportunity out there is, right? In terms of just revenues or, as I said, the example of user churn, etc., you leave on the table. And looping back to your initial question about use cases, so we help our customers to understand the why, to simply answer their questions. And then additionally, basically trigger an operational action straight away. That's where the agents come in, right? So a good example is churn, right? So what we do, for example, in the revenue operations space is that we help our customers understand where they have churn risks because a traditional BI tool, Power BI, Tableau, think about, let's say we have one customer that sells a digital product. So what they do is Let's upload dashboards, show them, okay, my users, my customers are using my digital product, which is an ad tech and advertising tech product every day. So seems to be fine. What they don't know is that, that what their BI tool does not tell them is that that particular customer has raised 15 support tickets in the last 3 months and they have sent 3 complainer emails to various people in the organization, right? This is where generative BI kicks in. Don't only look at your particular data that is stored in a data warehouse. Generative BI understands what's going on in the entire organization. So it merges your data warehouse with your customer support tool data, with your conversational data, like support channels via Slack, emails, right? It plugs in Salesforce, your CRM, et cetera, et cetera. And make sense out of this, right? And only then the analytics happens on top of that, right? So think about, that's why I said in the beginning, think about it like an AI COO more than an analyst, because that's more or less what a COO does, right? Understanding what's going on operationally in the organization and then pull the strings to make sure everything runs smoothly and we don't screw up things. And that's, for example, a use case in churn analysis or in risk analysis. We have a customer that does risk analysis with our product because it's so powerful. What usually happens is you start with a small use case and then the magical thing happens where our users understand, moment, this thing understands really what's going on in the entire organization, not only in my department. The use cases naturally expand at the speed of light and the sort of questions that our users can ask now and get answers instantaneously, not with a dashboard, right? You can have a dashboard. We also do that, but that's a feature, not the value proposition, right? So that expands like crazy and is an extremely, extremely interesting field. And we believe that this will be one of the fastest growing sub disciplines in applied AI that we will see in the next couple of years in organizations. Yeah. Yeah.

Oh, wow. Yeah. Very valuable. Yeah.

Yeah. Sounds fascinating. I guess a lot of people is always talking about data intelligence or getting insights from data. But I think having the ability to actually act upon those kind of, I suppose, almost real time using agents is really valuable. I mean, we, we see a lot of the kind of traditional report consumption decision-making processes still, which, yeah, which in effect is very cumbersome. And a lot of people are actually sort of starting to understand that, you know, if we do see patterns with inside data, can we deploy autonomous agent to actually catch an exception somewhere and deal with it? So I could see I could see that the value your product, um, you know, can, can bring to organizations. And there's definitely a drive, you know, to extract insights, but also in a way that it's contextualized well, like you said. So not just our sales numbers are down, but you know, it's, it's looking at a wider sort of scope or wider lens. Um, and I guess from an agent development perspective, deploying agents to do something about it, right? So that, that sounds very, very fascinating. A question from my side, and I guess you guys, you know, being in this space in the GenAI space, specifically around sort of data, I guess your product have the concept of an agent as well that can do specific things. Things are developing and moving quite fast. So, and I guess that is something that you guys, Autonomous Minds, look at as well. So, so where do you see? Things going and heading, you know, in the AI space and where exactly does this, do you think Milet fits in?

I mean, nobody knows where AI will be in a year or two. I mean, I saw, I think a lot of people saw all these memes that are going around and in the last few days where they, you see Sam Altman claiming, I think it was last year that we will soon like this year or next year cure the first variants of cancer, but what they did now is releasing erotica, right? Yeah. Instead, it's very hard to predict anything. But I mean, jokes aside, this thing is here to stay. It creates value. I think everybody feels that you cannot call this a fad anymore. I think these voices will become— I think they will vanish soon. The question is always, of course, this debate about AGI or ASI. I think this is still probably probably 5 to 10 years away, also based on what we see, what we do, the issues we experience and the problems we try to solve. I mean, you would be surprised how much of our AI-native solution is actually good old functional code. Like, it's massive. I give you an example. The biggest challenge that we have to solve is hallucination because we're dealing with a lot of our data, of the data we deal with. So we deal with quantitative data and with structured and unstructured data. Yeah. When you deal with a lot of unstructured and Qualitative data hallucination is a problem, but it's not affecting you so, so much. To give you an example, if you use ChatGPT to help you to craft a 3-year marketing strategy, you don't really feel if 95% of it is accurate and 5% is hallucinated, right? That it still feels overall okay and you will still be happy and it will still be useful because it is a fuzzy concept. Whereas we, when we talk about the analytics space and we, when we talk about a lot of quantitative data as well, you cannot put the wrong number in front of a CFO. Like, it's not okay to show them 95% correct numbers, right? This doesn't exist in that world. So this is a huge problem, and we solve this by, for example, do all computations with code. So we don't use LLMs or any sort of AI to do maths because LLMs traditionally are quite bad with maths. So from that perspective, I said, I still think that this whole AGI thing is still a few years away. It probably needs a few revisions of the current transformer-based architecture. It's probably, and you see a lot of very, very powerful and influential scientists like Yann LeCun and even Andrew Kaparthy coming out last week saying it's 10 years away because First of all, we need to ensure that these things are good with science and maths before we can actually expect them to invent new things, right? Which is intuitive, but I think these voices were a bit over— or got a bit interfered by all the buzz we saw on social media. So it's heading into the right direction. Right direction, I mean, it will become very powerful and it will become more and more autonomous going forward, but I think the breakthroughs where AI and these agents will be able to replace humans for more than, let's say, for work that requires active thinking. I'm not talking about purely repetitive stuff, right? Like something gets triggered and you always trigger the same action. You don't need AI for that. I mean, you guys are in automation for 20 years. You don't need AI for that. It's also something that we tell a lot of our customers. Like one of the sentences I use a lot in almost every prospecting call is, yeah, but the cron job does it as well and way cheaper. right? It's like, can AI monitor my whatever Grafana, my logs and trigger something, right? As a, but why would you use an LLM? It costs 50 times more and it's less reliable. Just use a cron job or something else, right? So that are the issues that make it so hard to be predictable if next year we will have agents that can replace an accountant fully. Right? I don't think so. I think this will, will still take years, if not a decade. Where Milo sits in, in that. So I think nobody really talks about generative BI a lot. I think this is something quite new. It emerged last year. And Gartner thinks this can be a $100 billion market in the US and the UK alone by 2028, right? $100 billion. I mean, it's gigantic. So we going, we are going after that space because we believe that, that the value, because you were talking about low adoption a lot, and I think to create a lot of value in organizations, you need to address pains that a lot of people feel and having data accessible and getting simple answers to very simple questions sometimes, right? How we are doing versus budget. I mean, on Friday I had a call with a customer that said, it takes me 4 days to understand how we are doing versus budget. I mean, you wouldn't believe that, right? But you would be surprised how many organizations, big and small, have exactly that problem. And I think that's where we, the technology is strong enough and our technology, because we built everything in-house, we are not using any frameworks. Was a lot of work that has gone in, a lot of engineering work that has gone in where we can create short-term value already and then develop the whole thing into a decision platform. So this is our big ambition is Now we are trying to help customers fix their analytics and help them get answers to simple decisions, to make simple decisions, to get answers to simple questions. But probably 18 months down the line, we want to take the decisions, the operational decisions, not the big strategic ones, but the operational micro decisions that every person in the organization has to take every day, that a portion of that should be fully autonomously triggered By Milo, for example.

Very, very good. Very good.

Yeah.

Yeah.

You touched earlier again on the adoption side. I just remembered in this MIT report, they said 95% of these projects are failing because yeah, ultimately there's not enough value, but also over 90% actually using AI every day. So that, that was all part of that report. And then you think, how can 95% of these projects fail if everyone is using it? But that's the thing, what you mentioned earlier, everyone is using it for some not value-added stuff, right? For the company. Sure, everyone is doing a bit of ChatGPT here and there. And Arno, you mentioned that phrase a couple of years back and it really stuck with me. It was the, you could trim your garden with scissors, right? So you could do this very efficiently, right? You could have the best scissors, you and all that kind of stuff, right? But is it effective? No. Is it actually adding enough value in the right time? No, it's not. But, and this feels like everyone is doing doing the AI approach a little bit with scissors at the moment, right? So little scissors and you could do a nice proper thought-through kind of solution. And that's the big lawnmower everyone is looking after. Yeah.

It's because we're all trying to find a problem for the solution, right?

Yeah. And indeed.

It's like top management says, we need to do AI, we need to do AI. And then you, you do something just because it sounds cool or because you saw a nice demo. Yeah. We, we, the three of us here in this call know it. There is a huge gap between spinning up a demo and the prototype. Versus deploying something in production scale, right? Especially when it comes to AI.

Yeah. So yeah, that brings me to the next question. So very innovative technologies there every day. There's like the next fancy thing available to us. So it's like this next shiny object kind of syndrome we all have. It's not, there's this next tool or that creates these videos 10 times better. And do we really need that 100th kind of video to advance humanity? Maybe. I think, I think of course OpenAI and all the other big firms, they sort of self-feeding kind of thing. So the, the masses, massive masses get really excited about it and they train the models a bit more and more. But, um, yeah, not really that valuable, I think, for, for companies. And, um, what would be your advice for companies, mid-market enterprises to get started, especially around the area you are now focusing on with generative BI? Or was data analytics in combination with AI?

I mean, apart from the cheesy understand your pain, right? And know what you want to address. So I'll not, I'll not use this here. I think this, this should be clear that you should understand one clear pain point before you get started, right?

Because otherwise, it's not that clear for everyone.

I'll not measure if the solution actually brought you the cure or made it worse. Right. But I think for, for the analytics space, for the data space, for generative BI, the key for any solution, no matter if it's Milo or something else, is getting your data AI ready, which means it needs to be accessible and it needs to be labeled and it needs to be organized. The labeling part is the tedious work that you have to do and you have to do it for every AI. I read this MIT report as well and I think we even wrote a blog post about it because we have the strong opinion that especially in enterprise, as I said in the beginning, if you want to be successful, you need to touch the data layer. You cannot just start with the application layer and hope that something fetches data from an API via an API and throw it at an LLM. And then we all pray that what comes out is good. And if it's good, we hope that if we repeat it, it is still good, right? It doesn't work. That works in the consumer space, right? As I said, when you do a video, you actually love inconsistency. I mean, as long as the characters are consistent, but you love that. We call it creativity, right? Or temperature as it's called technically, but in the enterprise space, in the corporate space, you need stability, you need accuracy, you need reliability. And the only way to achieve that is quite simple by getting your data AI ready. The key to it is, just to repeat myself, is labeling. So what you do with labeling is you basically attach, literally attach a label to your data. It can be on a table level. It can be on a column level. If one structured data, you can attach metadata to your data objects that tells the AI what it represents, right? What does this piece of data, what does this data object within my organization represent? And if you don't do this, any AI product will fail. right? Because this is the essence and the AI is not magic. If you don't do this, AI will start doing guesswork, right? So it will come up with something, but it will just guess, guess. It's not guessing, but it's just all based on probabilities, but it will mess things around. Sometimes that might be accurate because if you have structured data and your column name is revenue, then, and it's the only column that is called revenue. Well, it will kind of work, right? But the moment you have a lot more complex organization and you have maybe multiple revenues or you have multiple tools that hold revenues that might not necessarily be the same, right? Like we had a customer that has a revenue in one data warehouse where the revenues are calculated in euros based on the daily foreign exchange rates, right? And then there is another set of revenues that are based on invoices that takes the the FX rate based on the last day of the month. So you can assume that there is always some discrepancy, right? That was a brilliant case to illustrate the exact problem I just mentioned. How can the AI know which revenue, for example, which revenue field to take, right? And if you don't tell it, then you end up exactly with what I just said. It will spit out something based on best endeavor and based on probabilities that happen under the hood, and you will fail immediately. So long story short, get your data in order, get your data labeled. Labelled, make it accessible, then you increase your chances by multiples to be successful in an organization.

What is the best approach for companies to approach that labelling?

I mean, some of it— I mean, I can tell you how we do it. Some of it can be done automatically. So some of it is done by AI because what you usually do is you give the AI instructions. So before it starts to label things, You give it the background of the table. So you could say, okay, this table holds Salesforce data and it, what you will find is X, Y, and Z, right? Or you give it, for example, a sales process as a piece of document, right? Like you give it context. So it starts to learn and understand what it can possibly find in your data. And that helps a lot with labeling, right? And then still a human needs to review that because it won't get everything right. So as I said, you have 3 different variants of revenues. The AI will not smell, it is not magic, cannot smell which revenue represents what. So the human still needs to do just a validation verification loop, right? So the human in the loop comes in here again and just says, okay, this is correctly labeled. If not, change it manually. This is work you have to do. If companies promise you something else, they will probably lie to you. I wouldn't trust them. It's just how things work, right? So it's a mix of automated AI-driven exercise that is done by a model or it's done by an agent, agent using some models, but there is always a human interlude that needs to jump in and needs to take a look if this is correct now. And I can tell you when we look at our support tickets and we look at I don't know the numbers by heart, but I can tell you whenever somebody complains about wrong results, the vast, vast majority, and we are talking about 90% plus, is usually bad labeling, right? Or the customer plugging in data that isn't labeled at all. And then obviously the smartest model and the smartest AI will get confused because there is something new that I have no clue what this is about, right? And sometimes it decides to fetch data from the new source without knowing what it represents, and then it shows up somewhere and it looks like a hallucination, but it usually is not a hallucination. It's just a wrong piece of data that exists somewhere, but is out of context.

Yeah. Right.

Okay. Interesting. Yeah. Thank you very much. So I think we come to this part of the podcast where we, where we have that tradition where we ask our listeners, no, we ask you to tell our listeners a little bit about you. So sort of more from the sort of the, yeah, personal side, not, not too personal, but less from the technical and business side usually. Yeah. Would that be okay to fire, fire away with a couple of questions?

Of course. Let's try it.

How would you like to take the first one?

Uh, yeah, sure. Um, who is your idol and why?

That's easy. I'm Austrian, so my idol is Arnold, of course. Arnold Schwarzenegger. A lot of people laugh at it, but, uh, I think he's just one of the most inspiring people in the world. I mean, you can, you have to imagine growing up from nothing, tough childhood, not speaking the language, emigrating to the US, becoming a— what people don't really know, before he became a movie star, he was a very successful real estate entrepreneur already, making millions, launching his own sports brand, becoming Mr. Olympia, right? Becoming the highest paid movie actor in the world without speaking the language, right?

Yeah.

Uh, becoming the governor of the fourth largest economy in the world, California, right? As a foreigner. So if that is not impressive, then I don't know what is. Yeah. So it's definitely very impressive. I don't know. Yeah.

I've read all of his books. Um, yeah. So he has also, um, an interesting business presentation. He gives, gives, uh, sort of to a lot of people around the world. That's, uh, yeah, it's very good stuff. It's all about putting in the hard work. So nothing will just happen like that.

Right.

So same what you have to do with AI, you will not have, um, quick values, quick wins straight away, but you must just playing around with AI. You still have to put into like proper, proper project work.

People need to understand.

Yeah. All right. So the next one is, um, yeah, what's the best advice you have ever received?

Best advice I've ever received. I think it's, uh, shock them with your honesty. Something that one of my early mentors told me is when I started my first management position, he told me that you're dealing with smart people, you're dealing with engineers, they anyway can smell the truth from a mile away. So shock them with your honesty, be open and transparent. Uh, that got me quite far in my career. So. For sure the best advice.

Yeah. Yeah. So no, no, create not too creative then. I guess with the AI and hallucinations, we sometimes see a lot of, I suppose you can't call it dishonesty, but you ask a question, you just think that, that, that's a, that's a very creative response because it wants to respond. It's never no, this, this can't be done. It's more, yes, it can be done, blah, blah, blah. And you think, I think that's a no.

So, uh, Did you, did you see the Deloitte, um, uh, report? Yeah, the news about that. So, so yeah, a lot of bullshit facts in a £400,000 or dollars kind of report.

That's ridiculous.

So just looking at sports, I don't know if you're a sports person who plays sports, but, um, I think everybody watches the Olympics. I, I, I love the Olympics. Um, if you were an Olympic athlete, what sport would you choose?

Quite a good skier, but I would choose tennis probably because I love it. I'm very, very bad at it. It's, it's embarrassing. I'm really bad at tennis, but I love it. It's just, I enjoy it.

Yeah.

It's a, well, paddle is the new tennis now, isn't it? I think it's the fastest growing sort of right now. Multiple minutes.

It is. Yeah.

So we, uh, yeah, we, we are fond. I love tennis as well, but, uh, yeah, we, uh, I'm a fond paddle, paddle player. Um, Sascha, I don't know if we've got one more.

Yeah, I think, I think I skipped one question earlier, but, um, it brings us back to the knowledge and then to AI in general. So imagine you could get all the contents and knowledge of a book instantly. Which book would you choose?

I need to think of a non-technical book because I read a lot of technical stuff. It's just a relic from my times when I did my PhD. But I think it would be, yeah, this is a good one. Africonomics, I think it's from Levitt and Dubner.

Yeah, I've heard about this.

Two professors at the University of Chicago, I think, or Chicago Booth. That's brilliant. I read it 10 years ago. I need to do a refresher. A really charming version of combining economics or basically explaining the principles of economics and data analytics with weird questions, such as, what was one? I think, uh, why the average drug dealer lives with their mom and makes less than minimum wage, for example. It's all about incentives and how humans run the world. Based on incentives and hidden incentives. So, uh, that would be definitely, uh, something that I can recommend and that I would have, uh, uh, that I would choose here because, uh, yeah, it's, uh, I like, I like any discipline that teach you how the world runs. I think that's the future, especially in, uh, a time of AI where let's say a lot of skills might be displaced, but I think the moment you understand how the universe and how the world works and runs, be it economics, be it maths, be it medicine, be it biology, I think that is still something really, really valuable.

Yeah, very, very much so. Yeah. Similar to that, um, I've seen the economics of AI at the moment, basically they're like these 3, 4 players, but everything sort of revolves around NVIDIA, OpenAI. So then they all feeding itself. There's a bit of Oracle, but Oracle buys NVIDIA stuff and then they give a contract to, it is, so it's a bit incestuous, isn't it? In a way. Yeah. But it is, it is, it is keeping the bubble up at this size at the moment, or it's growing, but it's just these 5 companies sort of, yeah, sort of, um, pushing it.

I agree.

Interesting. Okay. So yeah, all the questions, um, done for today. Great pleasure, Andreas, uh, for having you on the podcast today. So, Thanks for sharing all your insights and thanks for sharing a new topic, which is sort of emerging, narrative BI. I think we will hear more and more about this going forward. It's a fascinating thing. And then I think helping businesses putting sort of insights into action, I think that really helps from small business to enterprise. I think that, that, that will make stronger businesses, stronger economy, and can probably solve is that a lot of issues we have out there. And additionally, I think it's more fun to run a business if you have a tool who actually tells you how things are and advises what to do next. So I think that's quite exciting. And I'm sure some of our listeners would like to get in touch with you after that episode. So what would be the best way for them to reach you?

I think there is two simple ways. One is via email and one is simply via LinkedIn.

Okay. So yeah, I will put your LinkedIn And your email as well into the show notes or just the LinkedIn?

Ah, you can, yeah, just let's start with the LinkedIn.

Yeah. Yeah. Okay, perfect. And you mentioned, okay, you're launching sort of the product more and more to more sort of people who are interested. Is it okay to put sort of the signup link as well onto the—

Yeah, 100%. Perfect. A very interesting sandbox demo environment. So you can sign up free of charge, of course, takes a minute and then you can play around with some datasets and some scenarios to get and see how Mylo works.

Perfect. Yeah, that's exactly sort of our listeners. They are very keen on sort of practical advice and like to get their hands dirty straight away. I think that's perfect. So we will put that in the show notes and then they can reach out straight away and get started. Amazing. Yeah. Fantastic. Yeah. With that, we are coming to the end of the episode. Thanks again, Andreas, for joining us. And yeah, and also a big thanks and thank you to all our listeners. Yeah, and of course, as always, so we had a bit of a sabbatical, but now we will be back with another episode very soon. And until then, let's automate it.