AI capabilities of UiPath, Celonis, Metaverse, etc.

Nico
Hello Andy.

Andreas
Hello Nico, wie geht's dir? Oh, sorry, we need to switch to English. How are you, Nico?

Nico
I'm fine. The week was quite turbulent, but now the weekend is coming and I try to shut down a little bit.

Andreas
Yeah, I think that's important. Very hectic in the weeks and lots of things flowing forward. Maybe we hear that later in our learnings of the week.

Nico
Yeah, maybe. And how is it? You are at home right now, correct?

Andreas
Yeah, I'm now home. I traveled a bit back and forth also this week and I had to exchange my company car, get a new one. Then the audio system was not working so I had to again exchange it. So you have always that site pain points that are not really paying into, that are not really value adding, I would say they're not paying your bills, but you still need to care for them. So there was a lot of these things this week, I would say.

Nico
Okay, we have a lot of things to discuss today which is in the first part, like always, the news from the market and we have some great news and then we got inspired from the news of the market about artificial intelligence and want to set a focus on this topic today. It excites me a lot because I hear a lot of people talking about AI topics, also how AI topics integrate in process automation and there we will shed light on.

Andreas
Yeah, and I hope that we make it below the marketing level that we're not talking about AI on slides, but really about how we use technology and how maybe today's technology differs from the technology 15, 20 years ago.

Nico
It's a wide topic. Let's make sure that we keep under the 60 Minutes.

Nico
Okay, but let's dive into our Automation Mag first. If you haven't heard about it, you can just subscribe on LinkedIn and then you get the freshest news from the automation market directly into your LinkedIn inbox. And there we have the biggest news. And what inspired us for this episode is that Microsoft Power Automate has built an AI that reads user's wishes from the lips in terms of automation. What do you think about this, Andy? It's amazing, right?

Andreas
Yeah, I really like it. Sometimes I would wish maybe even that it's reading from the mind. Sometimes you have people that expect something and you don't know and then you need to fulfill the wish without you knowing. But I think it's great to see that basically there is some kind of sensing about what the human likes, what he wants to read the wishes. I think it's a nice use case to go for. I'm interested in how this will be embedded in the future, maybe into some commercial applications, how Microsoft will use it. It's definitely an interesting product to look at.

Nico
Yeah, in the first step on what we can expect, right? So it's already the first step towards automating the automation. That you don't have to build even in low code, the automations yourself, but you just tell a computer what you want to solve and it will magically build it for you.

Andreas
Yeah, so I think there are still lots of steps to go because there's always expertise needed. And it's not only that you save your sentences and then magically everything happens. But I also think that's the direction. So for example, it's not the main topic for today. Basically the Metaverse. What does it mean? I think with all the buzzwords aside, it means that technology and the digital world will get closer to the real world. And now you see basically that the stores in the digital world get more seamlessly closer to the stores that we have in the real world.

So you have basically the same environment. You can walk like into a digital store and see the products like you would go outside and you also see that. And I think that's the major thing. There will be lots of devices in future, so there will be maybe in ten years there will not be your iPhone, but you will have an ilense or some glasses and you will see through that. And so basically everything will be more seamless, it will be more close to the human and more close to the human nature.

Andreas
And I think also this step of Microsoft in listening and translating that information into the right things, this is also one step in this direction. So there's definitely a connection between these topics and the Metaverse.

Nico
Yeah, that's definitely exciting. What's coming up? There was another interesting thing we put in the Automation Mag, which is a posting from Philip Glöckler who is an investor, and he said that no code is eating the world, which is a reference to descendants software is eating the world 20 years back then, or maybe even 30 years back then. And he says that the no code industry, he explicitly calls it no code, not low code, is eating the world and is the biggest revolution since the iPhone in the industry.

Andreas
The thing is, you need to make technology applicable to as much people as you want to and in the end that's the trend. And there is more sophisticated technology that is more on the low code side and there is, I would say, more simpler technology on the no code side. And I would support that. I would say the no code development just has started and it will be everywhere in future and there will be basically more and more technologies being available for each and everyone out there.

Nico
Yeah, and I would say talking about this, there was a crazy thing going on at Tesla. It's already two weeks ago about the Tesla bot and you have the news andy. You just sent me a few hours before this recording and I haven't seen it. But you have seen it. There's a few YouTube videos on it. What was it about?

Andreas
Yeah, so actually I saw that when I was in the US. So there was this kind of nice show by Tesla and they presented basically their robot optimus. No, actually they presented two robots, so they have basically an older version that's already a bit more mature. It doesn't look very nice. You see all the cables and everything, but it's really a humanoid robot, so it can walk.

So basically the doors open and the robot was walking in on stage, it was greeting, it was forming a heart with his hands and greeting the people. So it was really cool. And they had a second version of it that couldn't walk yet. So that was a bit of a crazy thing because basically five people needed to carry the robot on stage because he couldn't walk. But it showed a bit how in which direction it should go because they had already some kind of a shell outside. So it looked a bit more beautiful. It was not open cables and things like that. And yeah, in the end I'm asking myself, where is this going and for what do we use that? So will that be your household helper or whatever?

Andreas
And similar to the question that I'm asking myself always again and again in, I would say, the software robotics RPA world, do I really need a humanoid robot that's mimicking human behavior for everything, that would not replace my SAP or Salesforce system by UiPath or Automation Anywhere and similar. I think you would not get rid of your kitchen aid or in the other kitchen device that you have by that robot. And maybe you would also not get rid of your, I would say in the end cleaning robots that you have or whatever, and they are not humanoid. They are special systems for that special purpose. But nevertheless, there might be things in the future where you have this basically I would say human, half human, half a nonhuman guy that's completely out of technology, that's supporting you. So I'm always very interested in which direction this will go. But Elon Musk sees also a business case there. You know how he is, if it's Tesla or SpaceX. It's not only that he's building that stuff. If you look into it, he's also making it a commercial success. And that's for sure some commercial thinking behind that.

Andreas
But that leads us to yeah, we are always getting more and more of that kind of artificial intelligence, more of that kind of technology can do things humans can do.

Nico
The thing what I was wondering a little bit, you explained it with this kitchen aid, they show a video on where the robot is watering some plants and carrying a box and yeah, I'm wondering does it really need to look like a human? I think it's a little bit also horror scenario. If you watch the Scifi movies and the cinemas back then with WIll Smith for example, how was it called again? I don't remember. [I, Robot, 2004]

Andreas
Yeah, there were multiple of them and they all didn't end well.

Nico
What is Elon Musk doing to us? On the other hand, maybe it's good to build the robots in such a shape of a human, because we are also doing human things. And why not building the machine in the first place? Like a human is built to do the things a human can do. I was just wondering why do you build it like this?

Andreas
And we learn a lot from nature. So, for example, we have airplanes with skin that close to sharks. We have lots of things where we learn from nature because through evolution, actually lots of forms and things that develop, they actually make sense. And is the solution for everything in our world a human shaped robot? No. Might there be some interesting things for that? Yes.

Nico
And I have also never seen a bird or a hamster watering plants. So probably it makes sense to make it a human shape then.

Andreas
Yeah, but on the other side, when you want to do that at scale to look into such a farm hall or farmhouse where you have thousands of plants, where you are growing paprika or whatever, you would not have a human robot going through and building the can and watering everything. I think you need a proper water system. So I think it always little bit depends on what you want to achieve. And there are different kinds of ways how you can use technology to do it.

Nico
Yeah, there was also in the podcast of markus Lanz and Richard David Precht, there was a nice talk about AI. I think it was not the last one, but the one before they talked about the future of AI. We will also have that in that episode where you give an AI the task of producing something, for example, producing Christmas trees, and then the AI is producing that Christmas trees. And it doesn't matter what the AI needs to do for it. The AI would also move houses away to have bigger plants to produce that Christmas trees.

Andreas
Yeah, because if your telesystem optimize based on this one KPI, it will optimize based on that one KPI. And if basically it realizes it will not take care for humans, it will not have feelings. So it will not say, okay, I shouldn't do that, it's not a good idea to do it. It will act based on the task you gave the AI.

Nico
That's interesting. Okay, then let's ask maybe also the basic question what is AI? What is it?

Andreas
Yeah, so actually, I think we already covered one sentence and I also looked a bit in the internet and actually even the sentence that's on Wikipedia was pretty nice. "Artificial intelligence is intelligence demonstrated by machines." So basically that means wherever machines or software or technology is playing intelligent abilities that you normally would assume within a person, within a human that's what's the broader definition of artificial intelligence? I would say it's always related somehow to technology. It's maybe technology 5.0 or 6.0 because it's still in the tradition of the technology that we know, whether it's the production technology and I would say the manufacturing robots that we have in automotive still basically the humanoid that you see is a progression of what we see there. And also when you look into the IT technologies that we have and what's coming with AI, it's a progression of technology and it's still in that space and it's basically going beyond what we had before. We had systems like SAP or Salesforce or any kind of other systems that doing basically, I would say, transactions, processing of transactions. And there are systems that are, I would say rather like in the end digital workplace solutions or Office solutions like Excel or PowerPoint or Word where you have a shell where you can do something with it.

Andreas
But now artificial intelligence is a bit different. It's going more into sensing, into listening. And I think that's what we had also with the example with Microsoft or your Alexa is listening or your Siri is listening. It's about reading. OCR, so you can classify documents and extract data out of it. Speaking, writing and combining this kind of skill sets. I think it's a lot about more close to attributes of humans and I would say it's a lot about decision making, Nico. It's about automating decisions based on big data. Where we would go then into the ML, machine learning, space.

Nico
I had a super interesting project back at PwC where we wanted to create an app for a supermarket. And the app should be supporting the employees of the supermarket to get more knowledge about the fruits they sell. Because we got the challenge from them that if you ask someone in the supermarket: "Hey, what is the difference between this apple from South Africa and this apple from Spain?", and they normally cannot answer you the question. So we were thinking about very high tech thing where you can scan the fruit as a picture and it will show you then the most relevant information about the vitamins, about the country and everything of the fruit. And then we had discussion with AI experts and if this is at some point possible to just from the picture recognize if it's a banana or a carrot for example, this would be probably easier. But there are fruits that look quite similar and there's also it's more difficult if you want to compare two kinds of apples. And then he said in future he believes that it's much more likely that there will be sensors that recognize the biochemistry profile of the fruit.

Nico
So you know you scan and it will basically smell and can, you know likw how good dogs for example can smell in comparison to humans. And why shouldn't an AI be able to do it in future as well? He said more likely to recognize such things via the visual thing, via the picture is via the olfactory or smelling thing. And I find that mind blowing because I haven't thought about it before.

Andreas
Yeah, I can also give a bit of background here because I think the challenge is quite clear. If you want to, in the end compare even two different apples and which type of apples they are, there's not so much difference. So now, if you use the picture, how does an AI know which apple it is? So in the end, it also needs to learn like a human. It needs a lot of samples, it needs a lot of examples, it needs a lot of training data, which means in the end, it needs millions of pictures of the one apple and the other apple. And it needs a mapping. This is the one apple and this the other apple. And then it basically learns. Okay, if it's looking like that, it's going into this direction. But now we are having challenges. So the more closer it is, the more unsecured the system is. So the more data we need, the more examples we need. And the examples need to be correct because someone actually before needs to have mapped this information. Oh, that's a Granny Smith. And that's Mira Bell or wherever it's called. Apple. And it needs to be mapped to each picture and it needs to be correct because if this training data is not correct, I will learn wrong things.

Andreas
And that's also for all the use cases that you have, if what you base it on, if the data is not having a good quality, your AI will also be stupid like hell, I'm sorry to say that. And you can even see that. I think that's something you really face every day, capture, maybe you know that when you're in the Internet and you take, I'm a human, and then you get pictures of a kind of a crosswalk or about cars or buses. Actually, think about it. You are training a system in behind because you are basically giving a feedback. Yeah, this picture there, it's a traffic light, this is a bicycle and things like that. And that's actually what we need in millions and millions and millions of examples later for autonomous cars, for autonomous driving, because there's lots of sensors necessary in the driving because the car needs to see, oh yeah, there are other cars. There is a bicycle, there's a traffic light, it's red and things like that. So actually you're already part of that game and part of training and AI if you're in your daily way in the Internet.

Nico
Yes, even I saw, to wrap it up with something funny, there was a LinkedIn posting about, we discussed about this, that you can use these tools to create art out of words and someone typed in "salmon in the river". Have you seen it?

Andreas
Yes, I've seen it and it was basically packed salmon. It was the filet of the salmon swimming in the river and not the fish anymore. The system learned  salmon, that's this red kind of fishy meat in the packages. Oh, and river is that so I put the meat in the river. Absolutely.

Nico
Now, if you say if you want, then AI is basically decisions. A computer does all it's a lot about decisions.

Andreas
It's basically using data out of the past to take decisions in the here and now. If you think about it a selfauthonom car, it's decisions. It's a decision to steer left, to steer right, decision to accelerate, to deaccelerate, things like that. So it's lots of micro decisions and it's lots of small AI pieces because there's not one general AI that is not the Elon Musk robot with one brain that's sitting in the car and driving the car. It's lots of sensors and lots of micro and mini decisions working together to keep the car on track and not hit people, basically.

Nico
To get the connection now, what I would find very interesting is, if you could answer us and the listeners, or me and the listeners the question, what do you think, how were decisions made in the past? Also in the business context, how is it today and how is it changing?

Andreas
Yeah, I think decisions were there forever. So even when they're in the Stone Age, basically there were decisions. If we go hunting or farming or yeah, all the kinds of decisions were existing and how were they taken. So in a huge amount of time in the world and still in parts of the world today, like that, it's the right of the powerful [stronger]. The more powerful you are, the stronger that you are, you can take decisions based on your own favor. And in companies, it's a lot like also the hierarchy. When you fire up, you take more decisions and it's based on your gut feeling. And I say experience, because experience means yeah, it's also a gut feeling, it's also some kind of an opinion. It might be sometimes a stronger opinion and then the weaker opinion and don't want to touch on it. But it's still very much about opinion and it's still very much about the experience. And the examples you base the opinion on is 20, 30 years old and it's maybe partially applicable only today. So you try to solve today's problems with 30 year old data and experience and what we are trying to do now, and I think that's where it's changing.

Andreas
We tried to rely more on real information and data, so we tried to collect a lot of data here today and yesterday and in the close time frame where we are and then take decisions based on that. So we try to get insights. For example, I always call it visual analytics, for example. So BI, we highlight data, we show it, we want to show, okay, there is a pattern, there is some yeah, this is my numbers, this is my KPIs. Same with Process Mining and try to derive decisions out of that. And that can be first of all the transparency. So it can mean I'm supporting a human decision by visualizing data. It can in a second step be see me automation. So Nico, I suggest you for our next podcast. Our data says that's the best topic and if you click here, we are scheduling the reporting, the recording for next week. So you already have the suggestion and now you can take the suggestion or you can overrule the suggestion as a human. And the third form would be the full automation where you have in the end a decision in the back and where you say I need to classify a financial booking and I need to do that in a more I would say the traditional ways are not working because it's not an easy A or B rule but it's a very complicated environment.

Andreas
So I basically use some kind of the data of the past to classify the document in the here and now. A good example would also be, I think this is a good application in companies, is the whole planning, the whole in the end planning about how much goods will I need to sell, how much do I need to procure, how much do I need to manufacture, how much in the end financial support do I need to do all of that? So all that kind of integrated business planning that I need that can really be enhanced by data and can really be enhanced and that decisions can be improved I would say. So it's better decisions than if it's only human based. I think this is where we are.

Nico
Going to what I find interesting because you started with this gut feeling kind of decisions and I was just wondering where does a gut feel come from? It comes also from data to some extent, right?

Andreas
It comes from your personal experience, I would say. And this is I think it's even comparable because you have made some experience and based on that you think that's the right way how to go forward. I think the big difference to the data driven way is that you might have only three data points and they are 20 years old and the AI that's helping you to take the decision has 3 million data points. And I think that's the tricky thing about the human experience. It's always based on a very low amount of data points and when you look at your statistics that makes a huge impact because few data points is a lot of insecurity. High amount of data points is way more predictability.

Nico
We want to also go into, because our podcast is around automation, of course we want to also talk about what capabilities are on the top automation tools. For example, Celonis or your UiPath. Before we do it, should we do a short break?

Andreas
Yeah, let's get a glass of water. And I think there is some news besides the podcast that will fill the break. And then we are back with Celonis and UiPath.

Advertisement
Why buy process automation at a high price when it can be built internally and sustainably? Surveys show that a lack of knowledge and skills is the number one cause of death of all automation initiatives. More and more companies realize that automation success is not only about expensive tool licenses, but the perfect combination of bots and people. Therefore, the Bots & People Automation Academy offers over 50 learning modules to train employees and managers in topics such as automation strategy, citizen development, tools and technologies and many, many more. Visit www.botsandpeople.com or click on the link in the description to read all about it.

Nico
Okay. Hope you have a glass of water on you.

Andreas
I actually have a bottle of water. But it has two liters so it's covered.

Nico
Okay. How many liters do you drink a day?

Andreas
Actually a lot. Unfortunately also, some I would say, unhealthy stuff. I always drink some. I don't want to make a retrievement out of it, but I drink some Coke Zero and Red Bull. Basically my former customer. Since I worked at Red Bull, I'm drinking a lot of Red Bull. So it had an impact on me. Not a lot, it's okay. Sometimes I would say employees say it's too much, but it's actually not so much. The most things I drink is water and I drink about two to three liters water every day. And then a little bit of something else that's flavored. But I actually drink a lot because I personally get a headache if I don't drink enough.

Nico
At least it's not a Gatorade or Powerade or something like this. Okay, but let's dive into your UiPath and Celonis. Maybe also we can talk about automation anywhere, blue prism that are all tools that don't call themselves anymore process mining tools, but also especially the RPA vendors. RPA and AI tools or automation suites. And that's why. Let's check. Is it really AI based already? What is your opinion?

Andreas
I like the Automation suite naming more because for me this RPA plus AI, it's a lot of marketing, sorry for the word, bullshit bingo. There is for sure something that we would in the end say is artificial intelligence. But it's way less than on the slides. I think that's something that I can say and that's across the tools. So for example, let's use UiPath as the example and then scale it to the others what is in the tool and what's also the approach around. Because I think for UiPath it's very important to mention that they are really clever in their approach regarding their ecosystem and from the beginning. From day number one. They were a company that said: Okay, we have a capability yeah. We want to also grow that in the future but we're not able to solve everything only with UiPath today so we have tons of partners that help us we have integration to Abby OCR and to other OCRs and for some of the capabilities they could basically integrate to they developed an internal capability so there are some internal capabilities that you would say are humanlike or more intelligent but the base product we need to be honest it's very much rule based in traditional It with the sense of yeah.

Andreas
It's mimicking human behavior and it's logging into a system and there are then some. I would say. Key topics and there would say UiPath is progressing very well and has some intelligence and this is all around. I would say the access and I would say if you look basically the computer vision so sometimes you can identify the object in the login and you can basically go through the object sometimes you really need to do it through computer vision so the robot will see the screen and he needs to identify only from the visual like with basically the apple that we had. That's the login and there I need to click now in and insert this kind of data yeah. I think with UiPath there are two main topics. And the number one is, as mentioned, the computer vision and the number two is everything around unstructured data to structured data and there you have basically the OCR capabilities and the NLP, so the natural language processing and it's basically the way if we go the whole way from you have a document. It can be an invoice. It can be a contract. It can be anything now you need to basically identify what kind of document is that so what do I expect in that document on data and information and then you need to find in the document.

Andreas
In the text. In everything. That's the date it was issued. That's the amount. That's the tax rate. That's what it's all about. You need to extract that information because it's hidden in the document and basically fill it into. I would say. Fields so you need to bring it in a relational data model that a normal system can use it and progress with it and I think this is where most of the intelligence is needed and where it is there will also be nice features in future with text to speech and things like that so we have the Microsoft example but it's not that everything in there is AI so I think when we have the expert talk. The experts know exactly where we need AI and this is because there are major challenges in automation that we need to overcome like having the invoice on paper in front of you and automating. It. That's a challenge. You can imagine that if you are a person you are getting a letter home and you want to automate to respond to that letter. That's a hard task. And this is where in the end the human brain is when you do it yourself is working to in the end do that.

Andreas
And this is where we have AI today to be able to progress. And I would say that is the other RPA vendors are on a similar direction. I would say that when it's really about AI, automation anywhere also had a big focus on that. So they are pretty strong in that I would say I've not heard a lot of I would say value add from Blue Prism. That where they would have something that automation anywhere and UiPath don't have. So in the end I would say it's comparable. But even like when you look into the, we had the gardner magic quadrant, when you look into that I think it's quite good, reflected. The capability of the tool is quite good reflected in the market research and maybe a short last word on Celonis. So basic process mining is really standard IT functionality. Yeah there's an idea behind, because you have the timestamps and you bring that in an order and you build a graph and you have that kind of xray into your process. But that's not AI. So that is process mining. Like every process mining tool is doing it where the AI comes in is in interpretation of the data and this is where still there's maturity lacking.

Andreas
So there are ideas, there are first things that are happening but it's not mature. And to name maybe a use case whereas is using AI it's duplicate payment or duplicate invoice checker. So you get an invoice and you might get the same invoice twice. So if it's a million euros, you don't want to pay double because you lose a million euro. So you try to identify where are the duplicates and what can I remove. And sometimes it's a onetoone match so that's quite fine to solve. But sometimes maybe there is two numbers or exchange and instead in the invoice number of a zero two it's a two zero somewhere. So you don't have exact match but a fuzzy match. So you need to also compare that kind of closed documents and give some kind of an indication that's 80% a duplicate, that's 60% of a duplicate. And I would say there is certain things that you can do a rules based comparison but the more clever the system needs to get, the more AI you need. And what I've seen from Celonis in the last years is that they are in the stage between. So there's a lot of things that are happening in the machine learning workbench in Celonis, which is rules based, it's not real AI but there are also first steps towards, I would say machine learning and this is in there that's coming.

Andreas
But if you look into how sometimes it's on flights like oh, huge AI and process AI and RPA plus AI, I would say the reality is a bit more come down, it's a bit more relaxed and it's sometimes on paper, on paper it looks like everything is already AI. I think in the reality it's a lot of great software but it's not full of AI everywhere.

Nico
I remember one project where in my career as a consultant where we implemented SAP Concur and if you upload your documents there, there is of course an AI scrape. Of course it's great. There's an AI scraping the data field from an invoice or from check you have and you upload it and still you need to correct it or you need to look over it, of course. But there were a lot of people that expected that it will go all by a magic hand, you know, and that I found super interesting that the expectations from the business was very high. They did expect, okay, I just take a picture of that check from the restaurant where there's also red wine on the check and then it will recognize automatically everything and super to 100%. But this is not the reality.

Andreas
There is in the meantime really good software. So when you use maybe the app of your bank and you scan invoices, I think that made really good progress. But as we said in the beginning, there's no general AI. It's use case based. It's a lot of training. How much data, how clean is the data? It's also continuous training because invoice formats are changing, the world is changing. So in the end that also means that training needs to be continuously. You cannot stop it for most of the use cases. So in the end there are lots of other immature use cases and we are at the beginning of the age of AI. We're not at the end, we are at the beginning. We're still in the hype cycle pretty early. I would say most of the eye topics are not over the hump. They are still some kind of in the hype phase itself, not in the sustainable phase. And that's why most of the cases don't overestimate the impact that you can have. And on the other side there can be a huge cost because creating a new algorithm, doing data science, doing data engineering, having clarity on the use case, having an impact with it, that's not a small task.

Andreas
And I would like to give the advice before you move with your company, everything into an AI use case, think about whether there are simpler methods, how to achieve the same thing, or 90 or 80% of it. Because lots of things you can automate rules based, for example, or you even want to, if you are doing we want to do automated taxation in your company for, I would say a process that's not already covered by the SAP standards, then you can do that with AI but to be honest, it will be a huge effort. It will not be 100% reliable. If you have not done it correct in the past, you will automate all the errors you did in the past, also for the future. So it will be a pretty much a pain to get to the result you want to. But if you can map this into few single rules in a decision management, into decision modeling and notation and automate that, then that might give you a better output with less investment financial as well as time investment.

Nico
Yeah, that's very relevant point. And also what you always preach and preached that you shouldn't go with the technology first but look at the use case, look at the challenge and then see what is applicable and then also choose the minimum effort solution. Why choosing AI only because you can put it on a nice slide and say hey, we have done some AI use cases, it isn't worth it to be honest.

Andreas
It's always a decision between alternatives. And the same way that I want to say don't always use a robot, but there is valid reasons for robots. There's a valid reason to go for a quick fix solution. There are also valid reason to go for investment into the right use case, into artificial intelligence. And it's two different kinds of artificial intelligence apply a solution that has it embedded like Abby OCR or some capabilities from UiPath or automation anywhere is a different story than creating your own AI and algorithm and doing a full machine learning project inhouse. So be aware about there's a problem to solve, what are the solution options and then just select the right solution option that's feasible for you to do that is solving a problem. So be in a problem solving mindset and not in a I need to push technology into something mindset.

Nico
There's one example case I want to mention which I use since last week, two weeks ago maybe it's called it's not a business, not a B2B business tool. That's why I can also recommend it. It's also for free. It's called Quillbot and you can, it will rephrase your text automatically. Such a nice AI use case. So you type in an English text and I'm not the best English speaker of course, because I don't have the mother tongue. So if you are also not, if English is not your mother tongue, I can really recommend Quillbot. Then you type in an email or whatever you want to write and it will rephrase it to a very good sounding text.

Andreas
To be honest, there's a lot of really nice technology out there. I'm also always impressed by Deepl because I think my English is not that bad. I think my German is also quite okay. So translating between both I should be able to do. But if I sometimes look into what Deepl is offering me. It's really high quality language. So I'm really impressed sometimes by what technology is already able to offer.

Nico
All right, now we've covered the automation tools to which extent they use AI. We have talked about decisions, what happens in the future.

Andreas
I think in the future, we talked about it in the beginning shortly. On the one hand, technology will come closer to the people. So there will be a lot of intelligence solutions that will have the intelligence embedded. But it will not basically push to you so much. It will not basically be that kind of force. It will be more in a natural way part of your life. It will be there. It will be there in your car, it will be there in your eyeglasses, it will be there in the smartwatch that you use and things like that, but it will be in a natural way part of your life. I think that's it. And in, I would say, companies, the trends that decisions will be more and more based on data. If the data is there and it's clean, that is something that will continue. And from that I think that the trend is quite visible already. So I would not say that there's a huge surprise in it. And all the things that we talk today are playing a role in the big trends that we see, like in the metaverse and like in autonomous driving and things like that.

Andreas
So maybe we can do a follow up, maybe that's now that the call towards the audience. What's the topic that's interesting for you? So what should we maybe follow up on? Is it autonomous driving or is it something else? Is it a meta verse or maybe is it a very underestimated topic at the moment? Ethics in AI and governance in AI and there's a completely different view in China and in Europe and in US about this topic. So I think there's lots of interesting followup topics. But I think the future will be in a way that it's my big hope. I'm always positive that it's benefiting all of us as individuals and a society. That's my hope, that's my wish and that's my picture for the future.

Nico
There was also a reference again from Richard David Precht because he talks also about a lot about artificial intelligence and he said that you can see things often very black or white, but there's also gray futures. And the probability that the AI will use us as slaves and make us their own and AI will conquer the world is not at all realistic. But there is of course a great picture of it that we need to be careful, that it's not influencing us in a way that we don't want to. It already starts with spotify. Everybody probably realizes that it's very easy that you get caught in a bubble of your own music and it's hard to discover new music. And these are things you can also build in the AI that sometimes the AI shows you some music that you don't listen to or something new and life is just beautiful because there's also sometimes coincidence and nothing calculated by an.

Andreas
AI that's already part of our today's world. So that's nothing that's new. We have that bubble, we have it on Facebook, we have it on Instagram, we have it on TikTok, we have it everywhere. We have it maybe even in our private lives because we deal with a certain kind of people. So I think it's there. We could argue we're already slaves of our smartphones. So I think that's a topic that we already need to deal with that's not special because of AI and what's coming now. It's something where we might need a bit of a sensible way how to think about how strong is this trend going to be in the future. And there are also other trends. There are trends like putting your phone away on the weekend, not putting your phone into the bedroom and things like that. And I think everyone can take small decisions for himself to ease that pain a bit.

Nico
Do you think, Andy? That were good words towards the end of the podcast?

Andreas
I hope so, yeah, I think so.

Nico
Let's close with the learning of the week. What have you experienced and what is your learning this week?

Andreas
Yeah, so I think especially with thinking a bit about the AI topic today, I realized that we are also in it's by the way bubble. So I'm just realizing it now that we are very much in our Western bubble, that we very much talk about Celonis and UiPath and automation anywhere, and we are talking about Tesla and Microsoft, but it's all Western Europe and it's all us. And there's so much things going on in China. And I saw some interviews and some touch points with people in Taiwan this week because of the China Taiwan issues and things like that. There's so much of a development there, there's so much also technological progress, even if they might have a different culture. And I don't want to go into this discussion now. I think we sometimes need to lift our head a bit and look outside our bubble because there's also so much to discover. And from my perspective, that's a little bit the learning of the week, get the head out of my bubble and look into other regions.

Nico
It's a nice learning. Sometimes we forget that the world is bigger than just the Western world.

Andreas
And yours, Nico? Any learning for you this week?

Nico
For me it is tough because this week was a lot of work, which is not a negative thing, and probably the learning is that it's not a negative thing. So sometimes there are weeks, there are times where there's a lot of things to do and it's also hard to get your head up and think about strategic things and think about bigger decisions. It's rather than getting your stuff done. And my learning is that it's also okay to allow yourself to be in such phases. If you manage, then after a while to get on top of things again.

Andreas
And I think that's something that you're not alone with. I think that's something other people have experienced. That's why I think it's a good learning.

Nico
Yeah, cool. It was a very good episode. A lot of fun to talk to you. Like always, Andy. Have a good day. Greetings to Munich and see you.

Andreas
Yes. Greetings back to Berlin. And I think next episode we record in Berlin together.

Nico
Oh, nice. Yeah. Looking forward. This is the first time.

Andreas
Thank you. Bye bye.

Advertisement
Thank you very much for your time. We hope you enjoyed today's episode of the Bots & People Automation Insider podcast with Andreas and Nico. If you're interested in listening to experiences and opinions from people directly shaping the industry, feel free to tune into our tandem format, the Automation Talk. There we talk to inspiring personalities such as the inventor of process mining, Will van der Aalst or Celonis founder and CEO Bastian Nominacher, or Walter Obermeier from UiPath, just to name a few. Do you want to dive even deeper into the world of process automation and digitization and upskill yourself and your team? Then take a look at our Automation Academy. You can find all the necessary links in the description. See you next time. Your Bots & People team.

‍