(EN) Azure AI - how the business is adopting to AI

Hello to MVP Voices, the cloud and developer podcast with your host, Ralph Richter. Hello MVP Voices community. It's a new episode of the MVP Voices. And today I have a special topic, which is Azure AI and how the business is adopting to AI and that brand new technology. And happily we have a special guest today, which is Max. Max, I'm handing over to you. So who are you, where are you from and what's your daily business? Thank you, Ralph. Yeah, I'm Max Mecher. I'm a cloud solution architect working for Microsoft in Germany. And I help large customers to adopt our technologies. And one of them is Azure Open AI Service. A little bit about me. I'm coming from a consultancy and gathered a lot of experience there to make customers happy. And then four and a half years ago, I switched to Microsoft and helping now from the other side to make customers happy. Yeah, and maybe people remember you as a fellow MVP as well, isn't it? Yes, only a short time. Once I once I took the blue badge and joined Microsoft, I had to give the MVP avoid back. Yeah, never one and a half year for SharePoint technologies. The fun fact there was that I was not like actively doing anything with SharePoint anymore and already was focusing on Azure technologies. Yeah, it's funny enough. And glad to have you and that you are still supporting the communities a lot, which I really appreciate. And not me. It's the whole community appreciating for that. So Max, AI that seems to be all over the world, leading all news topics and give us a short introduction about AI. Yeah, AI, especially like open AI is in everybody's mouth right now. And it's very trendy. Yeah, also from like adoption perspective, this new service that was like launched in January in like three months, they reached 100 million active users, which is insanely popular and growth. And we finally see adoption in the market also for customers to use them in their products, in their services, in internal topics or tools from chatbots, documentation platforms, whatnot. Because of this new way of integrating and talking to AI and making it so easy, the adoption is easy to understand. Like when you look at this from a historical standpoint, AI is not something new, right? They started with AI in like 1956 or something. This is like older than my father. And yeah, but coming from a theoretical standpoint, then this evolved into machine learning and NLP, natural language processing and all this fancy technologies where you need to be highly skilled PhD kind of style to like leverage technology. But yeah, specialized models, all the complexities, lots of training data in the efforts made it very expensive to leverage AI technologies until I came and started with these last language models that made it so much easier to consume the stuff and get crazy good results. Yeah, so there are already cognitive services out and you also mentioned that we have machine learning. What is the special difference when we talk about stuff like AI and nowadays and what is it compared to the past when we look at machine learning and when we have cognitive services there? I mean, cognitive services especially on Azure is not very different from what OpenAI offers. Most of the services come as API. You provide data, you send it to them, for example, a PDF and you use the OCR endpoint to get the text out of this. Or you submit speech and you want to transcribe it and get text out of it. But with this new technology in the collaboration with OpenAI, you could also use now the large language models from OpenAI directly in the Azure platform. And with that you get amazingly better results, especially without providing context, the zero result prompts that you can give. And OpenAI is really good at getting you results on the first prompt without providing context. And that is the eye opener. Yeah, got it, got it. That sounds pretty cool. So you say that I can handle my tasks by AI using my voice or texting to it. So there's a special name for this and I want to understand on how is it done in the back? I mean, what would you want to understand here? How the large language model is being exposed? Yes, that's a good idea. Okay, so Microsoft hosts it, you get an API, like an endpoint, HTTPS based, you get an access key and then you can send your data, your prompt to this large language model and get the answer back. Depending on the model that you choose, because they're like a couple, you get different results. Some is good for summarization, some is better for code, some is good for sentiment analysis. But we see that more and more converge into these chat GPT models because they are more general and cover holistically all the cases that we see in the enterprise. Yeah, I got it. Just shortly Microsoft published the preview or the first release of Copilot for their M265 products. I'm pretty keen. So Copilot was there already using Visual Studio Code, GitHub and Visual Studio. And so now that piece of AI evolves to Copilot, excuse it. Yeah, everything becomes a Copilot and every product will be extended by using a GPT model or Copilot in Microsoft language. From Excel, Word, PowerPoint, Outlook, also the other cloud products like the Dynamics CRM, the Office 365 platform, Azure with the Security Copilot, the Windows Search Bar also gets a Copilot. Lots and lots of different AI enhanced products and services will be there. That's so fancy. Okay, got it. So once in a time, everybody will get access to that stuff by having an M365 subscription. And what do we have a special, I mean license to get it? I think this is not clear yet. I haven't seen anything in this regard. So let's see what comes out of this. Yeah. So for the Office 365 Copilots M365 Copilots, they said in the next months, whatever this means, we can see it. And yeah, I could try a couple of those Copilots already internally. And yeah, cool stuff to come and to see and lots of productivity enhancers. Pretty awesome. So Max, while you were pretty close to larger businesses, how do they adopt AI nowadays into their daily business? Is there already a starting point where they're getting off from and what would you recommend is the best point of starting using AI within the business? So a clear yes for looking at the new technology and adopting it and also like coming up with ideas where this would fit into the enterprise. And we have seen crazy demand on these technology. And I think in my career, there wasn't any service that was that hyped than OpenAI on Azure. Like my data and AI colleagues are super busy in introducing OpenAI on Azure, helping the customer to onboard there, discussing the questions about data privacy, differences to OpenAI public service, or also like an ideation how OpenAI could be leveraged in the existing products, new products, what they can get out of it, what's the expectation, what is coming soon. Like GPT-4 or Image Generation and all the various topics that are ongoing here. And because it's such a broad topic and versatile technology, we see demands from all different business areas, from internal services, HR, marketing, tech support, customer support, new developments, chatbots, call centers. It's super, super useful to also integrate this into existing stuff. Yeah, that's true. When we listen to Marc Sinevich and he says, he's very, very keen to see what's coming up with AI. And on the other hand, he's also afraid about AI. I mean, we have to talk about responsible AI a lot. Don't you think so? Yes and no. I mean, it's super important to use the technology in an ethical way, also in compliance to regulatory requirements like GDPR, and this is super important. But once this is sorted out, you should focus on the benefits of the technology and how to get something out of it. Because I also said this on the Azure Global Bootcamp presentation in the very end. We are not getting replaced by AI. This is far away, in my opinion. But we are getting replaced by people who are using AI because they are getting so much faster, better results, higher quality, and also can solve problems that they couldn't solve before. Yeah, yeah, that's true. That's very, very true. But we have to take care by using AI, isn't it? So when we look into the data, and for instance, when we look into medical data, we have their strong bias due to the fact that the medical data is based upon mail and not more than 75 kilos and a height of 1.75 centimeters or meters. And due to the fact, answers may be biased by such data. What do you think will be the way to deal with such biases in AI? Yeah. I always hear this question, like, who is training the large language model and who decides what is right or wrong? Yeah, and the AI folks call this alignment to give the model the direction, to say what is racism, what is sexism, to teach it and also give it a direction to say, okay, this is something I will answer or this is something I'm not helping you to build bumps or whatever. Crazy example that we saw in the beginning of large language models. I think OpenAI, as a company, is doing a great job in looking at these ethical considerations and balancing the model to come up with good answers, but they might not be perfect. Yeah, for sure not. That means, as a conclusion, we have to have an eye on what's coming out of our request from AI or CHETGPT for being more precise for the moment. And we also have to think over the answer we're given by CHETGPT to prove it if it is a good one, based upon our own knowledge as well. Is it fair? Is it transparent where this is coming from? But this in the same applies. Let's look at the old legacy way of coding where we saw this forest with GitHub co-pilot relying codex models that OpenAI provides. What have developers done before? Before AI? They went to Stack Overflow and copied code. Was it the right code? Was it fair? Was it correct? The developer had to decide. It was faster. That's why Stack Overflow was so insanely popular. Doing this all by yourself and then figuring it out without documentation, without help, whatever. Nowadays, you write a comment and tell the AI what you want to have and you get good results out. Yeah. One new thing came out by having this technology around is the prompt engineering. There are already classes around prompt engineering and it's hyped as the new top notch job nowadays. What do you think about prompt engineering? I only see this as a short-term trend, bubbly kind of thing, and not as a new engineering discipline that you potentially also attended universities. Nowadays, of course, skill required to write good prompts and also if you embed the large language model into your application, getting good results back in the form that you needed to process later on in your application is challenging and requires skill. No doubt. But it's not something I would classify as a new engineering discipline. Let's see how this develops. Pretty clear what we see in the LinkedIn recruiter demand is that they want to have prompt engineers with 10 years of experience and whatnot. And all the new trends that must be in the age of 21. Yes. Unbelievable what's coming up on LinkedIn. That's a truly saying. Besides a short note to the audience, so whether you are going into AI using Azure AI, for instance, there is a responsible AI dashboard accelerator kids for healthcare out. So use that if you're in doubt and check it out. I will put the link into the show notes so that you can easily find it. That's a true thing. I'm also seeing the prompt engineering as a short discipline, but would you recommend companies to stick in that for short and getting knowledge about and gaining knowledge and teach their employees there? Yes. But just out of the perspective that the folks need to get comfortable with AI and how to use it. This is also what we see in the enterprise. If they come up with ideas and then try to solve it with large language models, interesting new solutions also come out of this. Hey, we can now solve a problem that we couldn't have solved one year ago. Yeah. So when we see the timeline and going a little bit back, we had the cloud topping up all over. Then the next topic was IoT going all over together with big data somehow enhanced. And now we're talking about AI. And I guess all these technologies combined together empowers businesses to deal better with their big data and data stuff at all. Isn't it done that way? Yes. It's like prompting your own data and getting results out, getting recommendation based on your data. That it will be a huge driver of new technologies and new services, new APIs. Yeah. Similar to what we see now, if you look at Twitter, there are hundreds, thousands of companies that go to market every week. And insane growth of new services of sometimes only thin wrappers around open AI, but they provide new services that were not possible before with this little effort. Yeah, that's pretty awesome. That's right. Yeah, I guess also that dealing with data within the companies, doing forecasts or seeing trends, analyzing whatsoever within the data of a company will be a big, big topic for AI because it can. It can learn by itself somehow, right? I wouldn't say it can learn by itself. It's really depending on how you integrate the technology into or the data into your technologies. But what I mostly see is that it democratizes the access to the data because you can query it in your own language. German, English, Chinese, whatever, doesn't really matter because it understands the context. And if you tell it to do SQL queries in the back or provide this, do other API calls, then it can help you based on the ginormous training data that is already fed into the large language model. Yeah. What I really seldomly see is that folks train or fine tune the existing models based on their own data. Yeah. This is seldom. True story. When we're talking about AI running in the cloud, having business data in it, how is it and sure that the data is not going to anybody else? Okay. That is also where Azure really shines in comparison to the public OpenAI model because if you like use chat GPT for your research and whatnot, prompt can be used for training further GPT models. Yeah. This is also like described in the terms that potentially nobody reads when you sign up for the service, but it's a free research preview. Yeah. There was a little gap in your connection. So was it? You can repeat this. Yeah. Okay. So this is where Azure really shines in comparison to the public OpenAI API because if you use chat GPT for your confidential research, you must understand that this could be used in training of further models. It's a free research preview. So this could definitely happen in like feeding it data and giving it feedback can help build better models, but potentially do not want to do this with your confidential private enterprise data. And you should not. So having Azure in place here, Azure OpenAI service ensures that your data still stays your data. Everything is in your environment and it will not be used for further training of new models. And then it becomes like a one way street where you send your data to the model, get the result back, but the underlying model does not change. Okay. Good. That sounds good. So and I can also do a tiny training or how do they say to name it when I, when I optimize my own Azure AI models. Yeah. So fine-tune. Yeah. Fine-tune it. Yeah. Right. Thanks for helping out. So fine-tuning is the new, new thing then. And what you're saying is when you fine-tune your model by using your data, it stays with your company. Nobody else is using anything of the training you've done. Yes. You have a contract with Microsoft in this case. Yeah. And the terms are specified how and what comes out of this business relationship, but there is a contract. Yeah. That's, that's pretty easy. And it also puts you in a good place if you have regulatory requirements because there is not like leakage of data, privacy concerns, whatnot, because it stays in your Microsoft boundary. No. So when you're using an edge browser with a current release and using Bing, Bing for search, it's already based up on AI powered, right? Yes. You can, you also use the chat version of Bing. Yeah. And this will give you the enhanced dialogue based search engine where you also can like compare search results, which I personally find really interesting. Like that's the couch fit into the trunk of my BFW where traditional searches, like you need to search for your car. You need to get the width and the height of your whatever trunk and then look at your couch and then figure this out by yourself. Yeah. That's true. So you mean the Bing chat based search knows the size of your couch and also knows the size of the trunk of your BMW and can tell you, hey, buddy, that's not fitting in any. Yes. So I did this recently and I said, hey, does it fit into my car and mentioned the brand? And the first reply was, yeah, how big is your couch? I did understood the context. Yeah. And then could ask intelligent questions to like help you find the correct answer. Oh, that's cool. And yeah, there's the really big improvement also in like what what customers or humans around the globe and see and how they can leverage this new services. Yeah, it's really big democratization of new offerings. Yeah. Well, what are your favorite prompts to try out? What I am the auto out of customer feedback also did this comparing stock prices. Should I buy Microsoft stock or Google stock? And then it comes up with what is the what is what is comparable what's not comparable. It will not give you a recommendation. It's also cool to see that it doesn't like advice you to buy something. But then shares, hey, growth of this share was like this growth of this share was like this and the average is like that. And and yeah, yeah, links to see for yourself. This is like the transparency that I also want to highlight again. Okay, I'll tell you how it came through this conclusion. Yeah, it's funny enough. I've never thought of asking the search engine about with the couch fit into my car. So, yeah, this gives a whole bunch of new questions I can ask artificial intelligence to be answered instead of going over to my couch, measure the couch going down to the car measuring taking the measuring of the trunk and then compare whether it will fit or not. I mean, that's that's one thing. It's not related to business, but it's related and and showing on how they I will change our lives. So this is very important, I guess. So what I really often use for chitchat for is like translating content or like improving emails that I write or yeah, making presentation better for non native speakers. So this this context adjustments to like send an email that the other party can easily digest it is also something that large language models can really shine or understanding big documents. Yeah, we did this example with the Azure Global Bootcamp agenda and the speakers and asking it what is the next sessions and all this. But I also did it with 250 pages of document with the annual statements of a company. And then ask it, hey, what was the highlight of this document and it like had everything ready was pre fat with the data and could like query this. And it gave me the highlights that the company in the last year had special programs during COVID pandemic. What what I transfer sustainability and the economical growth. You could ask it how the board members have changed for the strategy is and like prompting the data was not only fun, but it was also very insightful and way way easier for me to then reading 250 pages. Yeah. That's true for for real data and getting an idea about the document which is 250 pages long just asking some questions is is pretty neat. You then can interact with it and say, hey, please write me an email with the five highlights of this yearly revenue for that and this and that audience. And it comes up with a perfectly written, grammatically correct email with you like as a human being would just take one hour to write it. And then you have like the spoiler plate that you can adjust three or interact with the model again and say, hey, please mention this and that point or elaborate on that point or translated to German, English, Italian, whatever. It's just insanely cool seeing how this is being generated on the fly. It just crazy. Yeah, same happens with the developer guys here. So when you see what what happens to visit studio or visit studio code where the chat now is available to build code. And also in the in the past with co pilot and a from GitHub. I mean, there are numbers out which says that 40% of actual code is written by AI. And that's a I mean a serious number. Yeah, 20%. Yeah, that's that's what they say. And how as a business so we have spoken about as a business we would say get in touch with with AI for sure start over, get your people hands on by learning prompt engineering. And what would you recommend that next. So I have the technology and I have also an idea about how to use it. So what would you recommend for for a company to do next. Yeah, start with. And to say this in this call, start with MVPs. Yeah, that's very great. What's what's then happening when you start with an MVP so learn how it behaves and how you can improve it. See how they the end users interact with this and get the results back out sometimes. You really get through surprising results right because folks will use dialogue based interfaces differently than you are based interfaces. And this is also like a question that I have for myself will we have like complex you eyes in the near future or will this being replaced by like the dialogues. Where you say hey, do this and that and that and this and then underneath the AI will figure out how to use the program in and choose that you get the right outcome of this. Yeah, I think we will have both for a long time until people are getting into it. Yeah, but that may change for sure. So then we will have like three input ways right we will have a CLI. Yeah, we will have the UI and we also will have the dialogue based or maybe the dialogue AI based version will then generate the CLI and command lines and to make it work. Yeah. Funny enough on global Asia Munich, we had that request about which technology is the most interesting one for you. And it turned out that the metaverse was right by two points. So what do you think will happen to metaverse and the combination with AI? Do you think that will make an impact to both of them? I mean, I will have an impact on all industries or technologies. And yeah, I'm not seeing it so broadly like adopted as we might have seen it like two years ago where everyone was talking about metaverse getting things like second life. I think the push for this is gone. Yeah. So maybe there will be adoptions, but not as demanding as we have seen in the past like last year, metaverse was in all. What was the big hype? Now it's AI, GPT models and large language models. In the end, the businesses will decide if they get the outcome that they need. Yeah. Do they get their business objectives? Can they fulfill their use cases? Are they faster, better quality and happier user, happy customers? And does that fit to the organization's strategic efforts and priorities? Yeah, that sounds like a well-known story here. So yeah, for sure. So metaverse is one of the buzzwords coming up in the past and is more over going down. We don't see that much HoloLens anymore. On my opinion, once these technologies are getting more and more into each other, I guess that can be a starting point for a new immersive experience on how we deal with technology in the future. Having it all around us, I mean, not by wearing HoloLens due to the fact that it is more heavy than every smartphone is, but it can be wearing it all around with you. And this I'm thinking that companies bringing out AI models and all that stuff take over a large responsibility on how that technology is. We've spoken about that a little already by saying and stating that responsible AI is an important topic, but it looks like that it will be a few companies where you have to agree with their terms of responsibility and cultural understanding and you have to make sure that that fits to your own understanding of social responsibility as well. And again, we see lots of American companies defining this strategy, but also we see lots of push and movement in the open source community. There was this one model or the weights that have been leaked from Facebook in this kick off the ginormous motion to have open source models with billions of parameters that you can potentially also run offline on your Raspberry Pi at home, on your own services offline from the Internet and do not have the control from external providers and leverage it there. So see how this developers like having your AI in your pocket on your smartphone will potentially also be a trend. So let's come down to and let's name the now available services on Azure, which is which are at least six services based upon the Azure AI service. So we have their phone recognizer, the bot service. We also have a metrics advisor, the video indexer. And for accessibility, we have the immersive reader, which helps a blind or not reading or not persons not capable to read to understand written text. And then we have cognitive search. And on top of that, we now have these, these new models like Dali whisper. And so I'm not 100% sure whether they are no built into Azure AI. And we have chat GPT there. So these, these are the new technologies we have to look on. We also have Azure machine learning. Yeah. Machine learning. There are tons of services. It's really hard to also like, um, follow this. And I also like when I'm being asked how much, how many services we have, I always say 384 just to like drop a number and see how, how big Azure has become. And what, yeah, ginormous toolbox you have to come up with perfect solutions. Yeah, that's so true. And this is also like one, one thing that I always hear from customers in the future, we all only use open AI then. But this is like one part of the puzzle, but not the entire puzzle. Yes, you will have like Azure storage to have data in there. You need a pipeline to bring it over. You need APIs or containers or whatever technologies out there to come up with a solution that just fits into the business case that you want to solve the problem that you want to solve. And yeah, making it work in one platform. Yeah, I will, I'm really, I'm convinced that I will have the biggest impact on technology these days. What we've seen since microcomputers are, are developed or being engineered. And I'm so sure it will also enhance not only applications, but our lives. And I mean, it's very tangible, right? Like you can prompt it, you get good results back. Very different to the previous big hype that we've seen like three years ago with crypto. Where we saw like huge demand, but no benefits for the general population. It was not democratizing banks, replacing money or whatever, but for this AI services. Yeah, you, you prompt it, you get immediately good result back. You can work with this. You can tell it to like improve, make it different, correct it, and then get jobs done that would have taken hours in minutes. And this is what they like is always eye-opening. Yeah. I like recently presented an opening a talk in front of 150 managers in one company. And I did this example with the yearly revenue. And they were like very familiar with the results and the statements that were made in the yearly revenue. So I prompted it on stage and said, give me that, give me this, and then write an email. And I have never seen an audience that was so hyped during the talk. They stood up, went closer to the screen to see what is going on. And yeah, like in prompt to interrupt at my presentation and ask, hey, do this. Ask the AI for that. And to like also to challenge it and get a feeling of what is possible because they like, they of course knew what Chichi PT is. It wasn't the target show in German news. But yeah, like seeing it interacting with your own data is a totally different experience. Good. That's really awesome. Yeah, for sure. So what do you think will happen to jobs in regard to AI services and technology? And I think we will see different jobs and also like, also like even more challenging jobs, but also the demand for the jobs will, will rise and there will be more demand. Similar to, well, I always come with this example. What do you think happened to accountants when Microsoft released Excel? Did we see less accountants? Well, it's a little bit tough to combine or to compare the release of Excel towards the job due to the fact that it is more or less a tool which needs a human to be used. Automated once and then profit for the next years to come is the real what Excel is providing. Or let's think bigger. Internet. Do you think Internet reduced the jobs because access to knowledge is now like broader and things could like be solved easier. Knowledge is accessible, whatever. I don't think so. I think we will see bigger demand and better outcomes for customers and humans around the globe because this is so accessible and democratized now. Yeah. So the, so in my opinion, we will see another round of transforming jobs. A transformation will happen. Yes. It must happen. Yes. Hopefully happens in our education system. Yes. Oh, that's hope. But don't don't let's start that discussion. We'll take hours and ages to get it done. Yeah. Well, that's a true fact. Well, education system has to be renovated a lot. Kids that do book summaries without reading the book. Yeah. Or solving problems that they could not have solved before. But it also shows like the possibilities and like the if what they can get out of the technology, how they can leverage it in their daily lives. They have this in that question. Why is the sky blue? Answer it for a six year old. That was not possible before. You had to read the book. Then do the work yourself. And now you come up with the prom in the eye helps you to come up with a good answer. Of course, you need to like think about the results and is this reasonable? Correct. What are the sources and like also do it you affect check. But still for for education, this will be and has to have a very transforming effect. Yeah. For for true. That's a that's for sure. It has to. Yeah. I'm with you. So that's lots of challenges. Also for the security folks, there will be lots of challenges. Just think about the fishing fishing industry. Yeah. And not for fish that you can like fishing like am fishing spear fishing, whatever. Well, you now can write emails that are perfectly tailored to the recipient. Yes. Language correct. Grandma correct. Adding facts potentially sourced from somewhere. Facebook, Instagram, whatever that into the model and then perfectly tailored to the recipient and how folks could. Yeah. Understand if this is harmful or not. That'll be a big, big, big issue for the security. That's what's true. That's true. Yeah. Yeah. Yeah. I mean, on the other hand, it also enables the security because I will point on will put the finger into the hole where it is and say, hey, you have to fix this. This is the other way around. So also a great help for the businesses. So what do you think, Max, is our business ready for AI? Our business in general, when is our business in general ready for something? Was the business ready for container technology? Is the business already ready for containers? Yeah. It is always a transformation. And if you are in the company that invests in talent and skills, then yes, it's a very approachable technology that you can leverage to get better results. But we need to have like a culture of innovation and try things out, see how they scale, how they behave, how to get good outcomes to also like encourage experimentation. So you think it's more or less a maturity we can reach instead of having this short term? We should bingo say yes. Yes. We should start this journey early and get early results and become comfortable with it and then think of how you can integrate this into your core products. Similar to what Microsoft is doing. Yeah. Like integrating this in the core products, windows, office, the entire office suite, all the clouds, dynamics, Azure, M365. Yeah. This is an enormous effort. This is not something that they do overnight, but they said, hey, this is giving our users a better results, helps them to become more productive similar to what we have seen in co-pilot, GitHub co-pilot, where developers just become fan of the technology and honestly become very dependent on this because it is so cool to give it a prompt code, comment, whatever, and you get surprisingly good results out of it. That's true. Co-pilot is a huge benefit for coders as well. I'm using it as well. So that said, we have to say to all companies as a summarization, go get your heads into it, get your hands on, try it out, figure out how can it benefit to your business, keep people trained, knowledge, use talents, and focus on innovations. And then you'll get it done and you'll get maturity on those technologies as well. And a good starting point is always the Microsoft documentation, for instance. And when you never tried it out, just go to Bing, open the chat search engine, and then try it out and have fun with it a little bit. Yes, the sandbox environments for this, or the starter kits. The example that I showed as a global bootcamp is just mind-blowing, damn cool to play with. They come up with AI-generated documents, and then you can query them based on a fictional insurance company. And you ask it, hey, is this included in my insurance plan or whatnot? And then you can see the documents that sit beneath it. Just cool. True. Mike's one last question. When you listen to this podcast, whom would you recommend to listen to this podcast we've had recorded just now? Well, I think it's pretty generic, right? Not too technical, from business to IT. I think we covered a good fit for everyone. That's cool. Thanks, Max, for being here and having me. We hope for sure it was a pleasure. And community, this will go out to you. Next session and episode will be again in German. I'm having a special guest for you called Way2 broadcasted for you. And this will go out in about four weeks. Thanks for listening and keep following, commenting, liking, and talk to you next. Bye, Max.