How to apply Artificial Intelligence such as ChatGPT to rethink processes with Benjamin Dehant
#029 Benjamin and I talk about various AI topics and the opportunities they present for BPM.
In this episode, I’m talking to the AI expert Benjamin Dehant about how to apply AI (such as ChatGPT) to rethink processes.
Therefore, we learn how AI can help us rethink processes, create art and support BPM use cases. However, he cautioned me about the limitations of AI when it comes to emotions. We discuss how AI can improve processes by asking questions and generating ideas and how it can be used to create art and relate it to process modeling. Additionally, we talk about how AI may change the work of process professionals in the future.
Today’s Guest
Benjamin Dehant
Benjamin is a freelance creative designer and content creator from France. He works on different topics such as virtual reality, NFTs – non-fungible token, design thinking workshops, and AI-generated art. He is blogging about this on madebyai.xyz. He also recently co-founded a service around AI transparency called aitag.ai.
He likes learning new innovative trends, he is learning fast, and shares his insights with the community.
Up to now, he has an exciting history of different jobs and lived for several years in Tokyo. Today, he is living in Alès, France, with his family.
You’ll learn
- What the hot AI topics currently are
- How different ideas to apply AI to processes will help us process guys in the future
- How and where to learn more about how to use AI tools such as ChatGPT to rethink processes
- What AI-generated art may have in common with the art of process modeling
- And What Benjamin’s top recommendations are to rethink processes
Resources
- Benjamin Dehant on LinkedIn
- aitag.ai
- madebyai.xyz
- futurepedia.io
- allthingsai.com
- Jim Kalbach on LinkedIn
- Mark Tippin on LinkedIn
Get notified about new episodes:
Transcript
Please note that the transcript was generated automatically and only slightly adjusted. It does not claim to be a perfect transcription.
Mirko Welcome to episode 29 of the New Process podcast. In episodes you may have been waiting for a bit longer. So thank you very much for your patience. Today we’re going to explore how to apply artificial intelligence, such as Chat GPT, to rethink processes. Therefore, I’m talking to Benjamin Dehant. Actually, Nelson Inno, my guest from the last New Process Podcast episode, recommended to invite Benjamin to learn more about how AI can be applied to rethink processes.And here he is. Benjamin is a freelance creative designer and content creator from France. He works on different topics such as virtual reality, NFTs, so, Nonfungible Token Design Thinking workshops, and AI generated art. He’s blogging about this on Made by XYZ, and he also recently co founded a service around AI transparency called a Tag AI. He likes learning new innovative trends. He is learning fast, and he’s sharing his insights with the community up to now. He has an exciting history of different jobs. He lived for several years in Tokyo and even worked for my favorite fast food chain, Subway, for several years. Today he is living in Alès, France with his family, and maybe you can hear his two month old baby in the background during the interview. So in the interview, you’ll learn what the hot AI topics currently are, how different AI ideas can be applied to processes, and how this will help us process guys in the future, and how and where to learn more about how to use AI tools such as Chat GPT to rethink processes. We’re also talking about what AI generated art may have in common with the art of process modeling. And Benjamin will tell you his top recommendations to rethink processes. So enjoy the interview with Benjamin Dehant.
Mirko
And now let’s start to rethink processes.
Yeah. Welcome to New Process Podcast, Benjamin. It’s great to have you here. As I already said in the last interview with Nelson, I asked him for his recommendation and he just recommended you to invite to the New Process Podcast. So I’m super happy to have you here. Welcome, Benjamin.
Benjamin
Well, thank you very much, Mirko, for having me on the pod. And thanks Nelson, for recommending me. I hope we’ll have a nice chat.
Mirko
Yeah, definitely. Let’s start with a check in. So what do you prefer in an aircraft aisle or window seat?
Benjamin
That’s a funny question. I would say to start. But the thing is, I always took the airplane when I was a kid and I always enjoyed it. And I don’t know, for some reason, like, a couple of years later, I started to be afraid in the airplane. So before I used to enjoy the window seat, but nowadays, I mean, kind of in the middle of the airplane. I would say I’m trying to fight that because I love traveling. But, yeah, I would say both in my life so far.
Mirko
Okay, that’s good. Do you still have a favorite airport, then?
Benjamin
So one big thing happened in my life is when I went to Japan. So I would say the Japanese airport, which is Narita and Hanida. I don’t know if you ever been there.
Mirko
Yeah, I’ve been to Narita One.
Benjamin
Narita is great because it’s the memory of landing into Japan. In Japan. But the best airport, I think, was Hanida airport as it is downtown, basically. So you can just land and in 20 minutes you’re already anywhere downtown with the subway. And of course, it’s Japan, so everything is clean and organized and so on and so forth. So, yeah, that’s my best experience for the airport.
Mirko
That’s cool. I still remember that journey, my journey to Tokyo. It was about, I think, 22 years ago, my first intercont flight when I was leaving home. So that was cool. Yeah, cool. That’s great.
Benjamin
I can imagine.
Mirko
Okay, cool. So what was the best process you have ever experienced?
Benjamin
I can say, maybe I don’t know if it’s what you would expect for that question, but so I got different careers already in my life. And when I think about process, I don’t know, there’s always this memory of when I was young and I used to work at Subway, the restaurant, maybe.
Mirko
I love it. Yeah, I love the subs, and I love the process. Tell us more about that. That’s cool.
Benjamin
Yeah, I used to work there for a couple of years, and I was young, and at that time, I was not really like the guy who followed the process, and maybe I’m still not. I was kind of creative. And I was also saying, it’s quite easy. We don’t need to follow process for dishwashing preparing, food serving clients, that kind of stuff. But of course I was following them because it was the rule. I was doing the processes of Subway, but I was not really thinking about that. And many years later, I worked in a restaurant as, like a part time job when I went to Japan, actually. And there it was, a small French restaurant. And I could really see the difference. There was no process at all. Nothing. It was kind of organic and chaotic, everything. So there was no way to find how to do something or how to, no instruction manual know how to and that kind of things. And then I was really like, oh, that’s a shock. I just understood all the work that was done in Subway and this kind of company where everything is planned, organized, and you follow the how to and checklist and things like that. So it was really like an aha moment in a way. And then I realized the power, actually, of processes. At that time, I had no experience in design thinking u weeks whatsoever. So it was brand new. Yes, that was a shock. So I would say Subway, their process is really great actually. Anyone can follow them and can be efficient and they put a lot of effort, of course, when they created them. So everything is kind of neat and clean and organized and yeah, I think that was my biggest experience with process, I would say. There’s also one thing that maybe can speak to everyone. Maybe you probably know Ikea, of course. And I don’t know, for some reason I love the Ikea instructions. I just find them really easy to follow. And then you feel kind of strong after you’re like, oh, I could build this thing just by following the instruction. And once again, it looks easy. But if you ever bought a furniture somewhere else and then you check the instruction, you are just like, it’s not possible. So maybe it’s kind of basic example. But of course they put a lot of effort also in this instruction. I really like them somehow. Just a couple of examples.
Mirko
Okay, cool. How did Subway document that process? Or how did you read how to proceed? There was the video or how were the process documented there?
Benjamin
Now the videos, maybe they have videos now, but at that time it was just you have huge manuals and a lot of checklist everywhere. You know, like, I don’t know, there’s a checklist in the kitchen, in the cooking area, everywhere you have a checklist or instruction or both. And you can just follow them. Of course, when you start, someone is teaching you the basics and how to do XYZ. But then everywhere you have indication, it’s almost like dumb proof. You cannot do a mistake, of course. You just arrive somewhere in the working station. You just follow what is written. And of course, after you remember by heart, of course. But yeah, this kind of checklist, same when I was a kid, I never had checklist. I hated this thing. And then I realized the power is very simple but very powerful. And even in the airplane, the pilot, they have huge checklist because of course, you don’t want to take a flight without checking every little thing. But Subway is the well designed machine factory enough to produce food.
Mirko
Yeah, I’d love to learn more about that.
I was visiting a Subway restaurant, I would say several times during the vacation in March when we’ve been to Florida, and it’s always super interesting to see how they are preparing the food. And I’m wondering if there is somebody out there who’s working for Subway and knows more about the process. We should definitely have an interview on that.
That’s great. Okay. But let’s get closer onto the topic we want to discover today. How would you describe your relationship to processes?
Benjamin
So you probably saw this on Facebook many times when you say, are you in a relationship? And I would write it’s complicated, I would say. Yeah, because as I was saying, when I was a kid, I really didn’t like the processes and I was never following them and I was just always trying to do something by myself. And only if I was hitting a wall or I couldn’t do better than the actual process, then I was following it, something like that. And I think I still do it a bit. So, on one hand, I don’t really like the process. On the other hand, I love when there’s a great process and you just follow it and it’s easy to execute and you get the results you expect. So, yeah, I would say it’s really half.
Mirko
Okay, it’s complicated. That’s great.
Benjamin
When I started working using Mural, the virtual whiteboard tool, at that time, I was jumping into like UX workshops, design thinking. And the thing is, I called my workspace in Mural trust the process. This sentence, of course I call it that way because the very first workshop I facilitated in my life, it was remotely it was with French people, always complicated with French people. I had this great framework, like problem solving framework, whatever. And I was saying to them, oh, this is a great framework, we should follow it and we will get the result we want to get. No worries, let’s do that. And during the session, every single step, they challenged me, why we do this, why we do that? I don’t think it’s great, and so on and so forth. And it was really hard to facilitate as first workshop. And then I was like, okay, we should trust the process. And I will just write it in my workspace and think about that all the time. Because there are so many great frameworks in design thinking and related to that, that you can’t just follow them. Of course, sometimes you should go out and find new ways, but I think they are really well crafted and yeah, trust the process.
Mirko
That’s perfect. Okay, I already introduced you as a person being very interested in learning new things, new innovative trends, and that’s why Nelson recommended you to talk to. What are you exploring at the moment? What are you learning? Right now?
Benjamin
Yes, right now. I mean, I started probably last August, I would say, but I’m really deep into generative AI. So you probably heard already about AI things. And I’m quite deep in this. It’s not really a new field, but in this new field of generative AI, I would say it takes actually a lot of time to check the news every day and to do some intel research, I would say. So I have a newsletter where I try to register all the news articles and things like that that I discover about the field. I try to experiment a lot and yeah, I’m really deep into generative AI. Like, what are the use case? What it is already good at? Not good yet. Where it’s going to go. I’m really exploring and everyone is exploring the field because it’s still brand new, I would say. So that’s the place where I spend most of my time, I would say, outside of client work, of course, that’s one thing. But also now I want to go back to learning a new thing. So I have background more in design, visual design and design thinking. I want to learn how to code, but probably not for normal product. I want to learn how to code for games, I would say. And I used to take some lessons, private lessons with a guy. It was a bit expensive and it’s hard sometimes to find the right moment to learn. And then you have to ask questions to the teacher. So it was great. But now with Chat GPT, you can do this much more powerful. Like you can have a private teacher for $20 a month, available 24 hours, basically. And you can ask him the same question 20 times if you want, and he never gets bored. And so, yeah, I think I’m going to leverage AI to learn new skills faster, I would say. So, yeah. Learning everything about generative AI, like text, image now, video and so on and so forth. And I want to learn coding. I always wanted to learn, but I’m kind of stupid when it comes to code. So adding this kind of new AI private teacher is going to be great. That’s the things I’m learning right now, I would say.
Mirko
Okay, that’s super interesting. I think most of our listeners have already heard about Chat GPT, but what else is going on in the AI space right now?
Benjamin
So, it’s very wild, I would say. Some people compare it to the Internet bubble even. I think it’s bigger than the crypto bubble, of course. I mean, when I say bubble, it’s just like the hype. I don’t think it’s going to be a bubble like go down for the month. It’s really huge and it’s really fast and everything might be disrupted, for example. So you already witnessed probably like the text, the chat bot, let’s say Chat GPT, you have the visuals with Midjourney Dall-e, stable, efficient, where you can create pictures, basically images. Now you have the video coming. There are some tools already when you can prompt a video. Either you put a video and you can kind of have an AI filter on top, or you can just prompt the video out of few words. You can prompt music already. You can prompt soon full video games. You could say like, make me a Pong, and you would get a Pong game. Movies, it’s coming. Because you can already make short videos, so soon you’ll be able to make long form videos, then you can make movies. It’s coming for almost everything, I would say right now, if you are like a manual worker, you’re maybe more safe for the moment. But some companies are building like humanoid robots and then you can power them with Chat GPT and if you have some Chat GPT into a humanoid robot who can perform like manual action. So everything is going to be disrupted and it’s happening right now. I think in five years it will be crazy where we are and unless we do regulation, unless we add some Ethic and things like that, otherwise big disruptions are coming in the field. So I really recommend you to try to catch up at least what’s happening so you know what’s happening and then you are better equipped kind of against the AI or with the AI depending how we want to see it, definitely.
Mirko
And we’ll explore that a little later. But what is from your perspective right now the most promising AI activities going on there?
Benjamin
Well, I think for the moment, they are the startups who are launching product and services using AI, that’s one thing. But I think for most of the people who might listen to this podcast, I would say Chat GPT basically is the most promising thing. I mean not just Chat GPT but the LLM and the last language model and GPT four and maybe five, six, seven later. That’s probably the most promising piece of tech at the moment because you can really do a lot with it and so many people don’t explore it yet or don’t enough. And I think you can go further when you have some skills or knowledge in the field, you can go even further with the bot. So I think that’s the most promising piece of AI that people can use right now. So there is also another thing that I think is promising. It can be related to Chat GPT but it goes further. It’s just the concept almost of life copilot kind of because in the near future I think everyone will be able to have the power that usually a company would have or like maybe a super successful solopreneur would have. Like you would have a VA, you would have your own developer, designer, marketing guy and a full team basically for you. In the future I think everyone will be able to have this maybe inside your pocket or on your computer or whatever. And you will have a full army of little bots that will do a lot of things for you. For example, you would be able to say, okay, can you book an airplane ticket for my daughter who is going to study abroad and can you get her the Uber from the airport to the airbnb and so on and so forth. You are going to be able to prompt everything almost. And I think this is promising because it will give more power to everyone basically. So if you have that kind of power you can probably do great things. Maybe we will see how people use it. But I think that’s the most promising thing, at least from my point of view because I never been able to have all of these people working for me, even for daily life. I would say it was something more for the CEO or these kind of people. Now everyone will be able to have this. Of course not everyone is going to use it that way, but that’s something coming and I’m really excited about that.
Mirko
I would say yeah, definitely. That’s super interesting. And I’m always trying to apply these methods to processes and I already had some ideas on how to use like Chat GPT for process. So maybe it can help the people out there to improve their processes by asking the AI on questions regarding a process, how to improve or innovate a process, or even with regards to employees if they have questions concerning a specific process, on how to execute an activity that they could ask a bot. Explain me how to do this in detail. What else? What would you imagine as ideas on how these features could help to support us process guys?
Benjamin
Well, I think as you mentioned and we should start there. Probably the first thing that you can already do with Chat GPT is for example, you could input a current process, try to go deep in the details and everything. Maybe there are different stakeholders, maybe there are many things. If you are already creating processes, you are the best person to speak about your process to GPT. So the more information you give to the bot, the better will be the result. So you should give it your process and you try to ask it to can you make it more efficient? Can you challenge it, for example? And maybe you would be surprised, you would get some interesting answers, maybe not. Maybe it would ask you to give more information also. That would be the first thing. Then you could say okay, this is my process, can you give me like five or ten ideas to make it better? And once again, I know it sounds really simple, but the more information you will give to the robot, the better the answer will be. Because if you just give process and say can you improve it? It might give very generic answers. But then if you start to give data, user testimonial, whatever, any more information, you can start to get really interesting answers, I would say. So I think there is a lot to do already for all the process guys to explore this first level I would say of using Chat GPT to improve their work in a way. Then there is also when you mentioned like the employee who have questioned about processes, for example, soon, if not already yet a company can train longer edge models like Chat GPT with their own data wiki and everything. They could train this and then they turn it into maybe a Slack bot or whatever tool they use. And then every employee could ask to this kind of chat bot who knows everything about the company and the processes I forgot how to do this thing or I forgot how to do that or can you help me with this task? And I think this is something that is going to happen and be in almost every company in a couple of years because it’s going to be really powerful. Also, of course, you are going to save costs like that. But I think it’s going to be really powerful. So that’s probably one logical next step for the companies. They will need something with data privacy and so on and so forth. But it’s already coming. That’s a good way. I think to use Chat GPT and this technology for process guys and company and also it’s back to this idea of life copilot something copilot I think it’s just the process guys should use this tool more often during their work or when they are creating a new process or when they are brainstorming when they are anything doing some research they should try to use this new tech each time maybe ask questions and do some research and challenge. And brainstorm with the robot so they will see what it’s possible to do with it. And I think you can give it like tedious tasks and things, that maybe you will struggle. I don’t know. Maybe you have to do some data analysis, research, whatever. You can give this task to the robot, for example, now. So you should do that. The only thing I would say to be careful is it can be weird. Because if you ask Chat GPT, for example, can you follow the human center design approach, blah blah blah, or empathy or that kind of things, it will say that, yeah, I know about it. And maybe your user feels like this, feels like that, it might give you answers like that. And you think, wow, maybe even knows about empathy. But I would recommend to keep your hands on this part of the work I would say because it can give an answer that really looks smart but in the end it’s totally off the track. So yeah if you are creating process keep maybe this part like maybe user test like user research, empathy and that kind of stuff keep it for you for the moment I would recommend it can give you ideas you can say. Can you give me ten questions that I would ask to my user to see if my process, I don’t know, is efficient or not? For example it would give you questions and maybe that would be almost the limit of what you should ask him. But I don’t know I explored a bit like a design sprint process with AI and fully done by AI kind of and I even said generate me some personas generate me questions for the user test at the end and can you guess how these personas would react to this thing with this question? But at some point you are not sure if you’re still in the reality or it’s just kind of fiction. Just try to be careful at some point, but otherwise you can do a lot already with the AI.
Mirko
Yeah, that’s super interesting. I think. Good news for the human centric approach there that AI will not replace these emotions there, at least not yet. Okay, at least for this year. But I already played around a little bit with chat GPT, but with regards to processes, I don’t know. Just to give you an example, when mapping processes, there is a standard notation which is BPMN 20 which can be exported as XML file. For example, can I import that into chat GPT even as a normal user to tell the AI about the process and then ask questions or how do that really look like.
Benjamin
So for the moment, you cannot yet export XML inside Chat GPT, but I think it’s coming pretty soon especially they start to release. So there’s GPT-4, that you can already access if you’re a paid user, but you don’t have the full GPT-4 power, you need to request an API access and then you can play around and do more things. Already you cannot input yet an XML file. But if somehow you can extract the content and just use a plain text, somehow you could give it even if it’s like an Excel that you would export as kind of a word document would be a bit weird. But you could even give this to the AI and it might understand everything. So you can already try to play around. For example, you can prompt the AI, can you create me a table with these kind of columns, this kind of line, and it will do something. So they are workarounds for the month, but I think in a couple of months you’re going to be able to do that. And also some people are working on the same technology, but maybe more open source or just more open than Chat GPT. So probably you can already access an XML file and give it to a chat bot to do something. So yeah, it’s almost doable already and probably some people manage to do this by hand for the moment, but yeah.
Mirko
Okay, so if our listeners would like to learn more about that, which resources would you recommend? Where should they go to to learn more about how to yeah, how to operate or how to prompt, whatever you want to know there.
Benjamin
So you might have noticed, and maybe your listeners as well, there is a million of content created every day about AI, Chat GPT, Midjourney, nowadays. So it’s a bit hard to navigate, I would say, because every Medium article, every LinkedIn post, every whatever will speak about that on YouTube. It’s terrible. You have like how to do a billion things which are debated. There are some gold nuggets inside this content, but it’s hard to find. Maybe I would recommend to try the opposite, which is there is a website called I think Futurepedia and there’s a second one. I will give you the link maybe after if you want. It’s just basically a website where you have all the AI tools that are being developed for the moment, whether it’s a full model or maybe just an app using AI, for example, I would recommend to go there and to scroll a bit and to see if there are some tools that resonate with your listeners. Maybe this use case or this tool or this tech might be interesting for me. I would start by there maybe because if you just try to read some content or you can of course use still the old Google fashion and say what can I do with AI about this or XYZ? And you will find some good result, of course still. But I would try to recommend just check the use case, what they are creating, start from there and maybe then you will see some Twitter posts, some medium posts about that specific use case. Otherwise I think there is still a lot to do by just playing with Chat GPT. Many people don’t think yet that you can do this, you can do that, so just try. Maybe they should just ask Chat GPT. You can even say oh, this is my current job title. I don’t know, maybe I’m an agile coach, for example. Whatever, what can you do for me? And you just start like a conversation and then little by little, okay, I have this kind of task problems, these needs, and I think it’s a good way to start. Just start speaking with this chat bot and see up to which point you can go also.
Mirko
Yeah, that’s super interesting and I think I have to try it again and learn more about that. I played around a little bit in the past, but actually I think I haven’t really grasped the potential right now. And yeah, I must go deeper into that. Thanks for that recommendation. That’s super cool. So you are working a lot on AI generated art. That’s right. Or so modeling processes for me is also some kind of art, like an art piece in the end, if you have a process model there, what are your key learnings about AI generated art? What are the risks and chances? What are you doing right now there?
Benjamin
Yeah, it’s a good question. So it’s a bit complicated in this field and for some reason there were more protest against AI art, I would say, than against Chat GPT for example. And I’m still not sure why. I asked a couple of people, for example, they didn’t want to say that they were using Chat GPT for writing a piece of text. Maybe like this is my text, not the robot. But they are okay with that. Whether when it’s AI art, many people want to say. They use AI and many people want to know if it’s made by AI otherwise it’s not really a piece of art. So it’s kind of funny for the visual, people are kind of against and more picky about it and when it’s just text generated it’s okay. So it’s kind of interesting. But anyway, what I discovered is that maybe first how I use it, you have to write a prompt to get an episode visual. Let’s say whichever tool you use, you have to write some kind of prompt. And what I tend to do is I try first with just a few keywords, then I generate some result and then I start to add more and more because maybe you saw somewhere sometimes you have like maybe five or ten sentences to get an image and of course you can get great results. But it’s going to be really hard to get these ten sentences like from the get go. So I prefer to just go with a few words first. Maybe like a flat vector art or of something, I don’t know, like a flat icon of a locker, for example. And then I would say, okay, maybe I want to get this kind of style, or maybe I want it 3d. Or maybe I want to add, like, blue and yellow for the colors, for example. So little by little I would add more keywords to my sentence and generate to try to control a bit the result that I will get otherwise it can be just by luck that you get a really nice result and that’s not what we are looking for. If you want to do something consistent or even something for clients for example so I will try to do like that. The thing is the more knowledge it’s like Chat GPT, the more knowledge you have in some field, the better your result will be and so for example, if you’re a 3D artist you will probably be able to get really nice 3D visuals made by the AI. I’m not free artist but I used to do some like vector illustration and that kind of stuff for web design, for example, so I can prompt quite good images for vector art, for website, for example, or for app or whatever. And the good thing is then after you can even turn them into a vector file and you can play around to adjust the mistake that the AI will do, for example. What I discovered also is that if we think about client work, for example, some people really want to experiment what is possible with the AI. And so they would order me visual, maybe for their website or for social media, or just for mood boarding or ideas from other companies. In the end, they didn’t want to use the AI because you cannot yet copyright a result. If it’s 100% made by AI, you cannot copyright it. If I use it a bit and then I do a lot of work on the visual. It’s still copyrightable but for the month you cannot kind of protect your AI out. So for some companies, of course, if they try to use it for their core branding or whatever and they cannot protect it, well, it’s a bit of a problem. So that’s maybe the limitation for the month we don’t really know who is the creator and the owner in terms of the law and IP and everything. So yeah, that’s the main challenge, I would say with visual AI for the moment you cannot really protect it, you cannot really own it somehow. So some people are okay because maybe it’s okay if you use, I don’t know, a blue dog made by AI for social media posts you kind of don’t care but if it start to be your own logo and then you cannot copyright it or that kind of things can be tricky. So what people do usually it’s they mix AI tech and traditional tech like Photoshop and AI and then it becomes something that you can copyright again. So yeah, it’s a bit like that but there is a lot of trial and error still. Many people experiment, many people share. So you just try to see the best practice, the best result. There are even prompt marketplace, I don’t know if you heard about that. You can go to some marketplace and you can buy prompt like for two euro you can buy a prompt that will the promise is that it will generate you consistently something. Maybe it’s a prompt to get, I don’t know, a flat logo for example, or maybe it’s a prompt that you can use into Chat GPT to get this kind of consistent results. Like the perfect prompt to get LinkedIn post about marketing for example. So there is a marketplace selling prompt and we can buy and sell prompt. So that’s really interesting. Also the key message prompting is the new chic for the moment, if you want to use AI best you should learn how to prompt and fast, I think, but maybe we can speak about that a bit later.
Mirko
But yeah, I’m wondering if I could prompt a process model as well there and in the end it’s a logical process but it’s also a picture. So maybe that’s a way to map a process or to design a process.
Benjamin
I think you can really try and so maybe sometimes you don’t get good results just because you are not lucky with the prompt or just you didn’t make the right prompt from the get go. But just to give you an example, like recently what I tried to prompt in Chat GPT, it’s like creating a full so it’s on the foresight project. I’m trying to make a Visual Novel Point clip game. Basically you would have a murder, you have suspect and you can click on some clue to see if this is the clue of the murder, whatever. But then it can be a bit intricate and if you start to have free suspect and you ask them question maybe they are lying, maybe not, maybe there is this clue and so on and so forth and you can kind of map a scenario of a murder. But still this starts to be a lot of information and everything and I start to be able to get Chat GPT to generate me in one time a full murder with the protagonist, the place, the police report, the clue, what does it mean, who might do which thing, and so on and so forth. It’s still a bit hard because sometimes you just say something that’s not coherent. You would say like this might be the murder and so he used this thing but it doesn’t make sense sometimes. So it’s still a bit hard, but you can go quite far. Actually I was really surprised. I was like okay. And I was saying something like generate me the list of the free suspect, the free question that I would ask the first time, what would be the answer? Tell me if it’s going to be a lie or not and if it’s useful or not, who would be the next person. And it’s quite complicated, quite intricate. So you can already do lots of things and that’s just an example for fun, I would say. So you should definitely try again, I think, and you can get interesting results. I saw recently on the internet some people, they start to generate some kind of graph and chart and stuff like that with the AI, they give some data and see can you give me this and this and that? So it’s coming, it’s coming and maybe not yet and that’s why we need your knowledge and expertise to make sure that it’s following the process or the theory or whatever. But it’s really coming soon, I think.
Mirko
So just to double check that, do you think in the future it will be possible to just prompt design the perfect, I don’t know, goods receiving process for me and then I can redefine my prompt a little bit to get the state of the art goods receiving process design.
Benjamin
I think soon you are going to be able to prompt that kind of things. And so just a word about prompt. You probably heard about prompt engineer and AI whisper and that kind of stuff for the month. It’s one of the hottest skill to have to be able to prompt well to the AI because that’s the only way to get some nice result or expected result or whatsoever. And I think it’s going to stay like that for a couple of months, maybe even a couple of years. But I think at some point it’s going to become so easy that you would not need to be able to prompt. You can just say some kind of vague question create me to do app, tailor it for my needs. This is my life and my needs and make me an app. And you would just get the app perfectly tailored to your needs, for example. So I think in a couple of years it’s going to be that easy to prompt anything. Even more complicated things like you could prompt a full product, you could prompt whatever yet if you want to create some kind of process, of course it depends on the needs of the user of the process, the company and everything. So you still need to input a lot, a lot of data into the chat bot. But the more we go, the more data it can take, remember, and analyze and use. So you’re going to be able to make more and more complex things also. And there is the chain prompt when you start prompting something and then you ask another question about the initial question and the answer and then you can have a full conversation. It’s remember. So you can also prompt in several pieces and maybe you heard about like Auto GPT or there are few words coming, I didn’t try yet this thing, but it’s basically the bot is going to do some kind of loop. You would say create me, this thing and itself. It would ask questions and answer by itself and it would do the task. And at the end of the day you say, okay, I created this huge piece of thing for you, for example.
Mirko
Wow, that’s super interesting. I definitely have to look into that more deeply. That’s cool. Okay, but transferring your overall experience to processes, what are your three top recommendations to rethink processes, to get to more human centric BPM approach and to inspire people for processes?
Benjamin
First, I would say that I’m not an expert in processes, so I just want to say this. What I would say is that, yeah, I think it’s basically back to some design thinking principles and things like that. It always starts, of course, with the user empathy, the needs. I would say always keep that in mind. Then what I tend to do, and maybe it’s not the best thing, but as I was saying at the beginning, when there’s a process to do anything, for example, I usually try to do it by myself before even checking the process, maybe. And then at some point of course I hit the wall or I cannot do it or it’s not efficient. Then I go back to the process and I’m like, oh, that’s how they do it. Of course they spent many hours to design the process. That’s what I tried to do. So I tried to start from scratch and try to solve the problem, to perform the task or whatsoever. And then I go to the process or what is currently created to see where it is and how they thought about it, if there is a way to improve or to challenge it or not. So I would say always try to challenge again a process, because maybe it’s the best process for something, maybe it’s outdated, and maybe you can use new tools also to revisit it. Maybe now with the AI tools, you could check again if there is not a better way, a faster way, more efficient way to do the thing. But yeah, the key thing probably is to keep the user and the empathy in mind at the beginning of the challenge, if you can make it better. I just like to do this kind of design thinking, like empathy ideate, select go narrow test and again and again for everything. Not only product or service, I think for process also it works. Yeah, that would be my recommendation. But once again, I’m not an expert in process, I’m more like on the design side, I would say. But that’s how I would approach any kind of process. Like if we go back to Subway, I think that’s how they did. Basically, they checked all the tasks that you need to do, what’s the best way to do them, how much time it takes, what would be the sequence of action to perform it in the most efficient way, fastest way, and so on and so forth. And I think you can apply this to most of the things, of course.
Mirko
Yeah, cool, that’s good. And to wrap it up overall, what would be your key message to our listeners to rethink processes from your perspective?
Benjamin
Yeah, I think it would match a bit. Want my previous answer? I would just say never assume that you know 100% about something. I always try to challenge it, try to be up to date, try to revisit the things again, go back to the user or the person who is going to execute this process or flow the process. Try to go wild and to go into new fields, maybe other categories or the field of knowledge or activity to see if there are some things you can pick here and there and to bring back to your own field. That’s the kind of creative way of thinking and how designers do as well. Just try to go broad, do a lot of experiment, research, and try to get the best out of everything to put it back together. And always also test, test, user, test, user, test. We cannot say how many times it saved time, energy and everything to just test the things. Yeah, that’s kind of the design thinking loop. But that works for everything. I would say that would be my main tip. And also try the AI tools because maybe it would be a shock that you can already replace or do something much better, much faster with them already. Because Chat GPT was the fastest growing tech actually in the world. Like, faster than Ethereum, faster than Facebook, Instagram and everything. Like they went to a million user, I think, in a week, and like 100 million in one month or two. But still, it’s just 100 million user in the world. It’s not that much and so many people are still not using it. So I think it’s probably time to use this kind of tech and see what is possible.
Mirko
Yeah, definitely. And I was surprised. I visited my parents last week and my father, who is 70 something, told me, have you heard about Chat GPT? I tried that out. I was really surprised. So that was way faster than everything else before.
Benjamin
Yeah, definitely. That’s why I think it’s probably the big disruption is coming because these kind of numbers are interesting and yeah, for crypto, I know it’s a bit different. Many things, many things. But I think with this AI, I don’t know if it’s good or bad for the humanity, but at least it’s here and it’s really efficient and yeah, it’s time to jump on this, I think.
Mirko
Yeah, definitely. I will have to go into that more and more. So, thank you for all these interesting insights. Benjamin, where can our listeners learn more about your activities, what you are doing?
Benjamin
So, my main social media channel is LinkedIn for the moment. So if they want to see my experiment, what I’m sharing and posting and so on and so forth, they should definitely check my LinkedIn. Benjamin on LinkedIn. And I love to connect with everyone. So if I get some request, of course I will accept and I’m happy to chat with everyone. I try to post more and more also on Medium, I try to write some longer articles, mainly around AI, sometimes even fully generated by AI. But of course I mention it because I like to experiment what the AI can write or can say about any topic. And yeah, that’s my main channel. I’m a bit on Twitter, but Twitter, it’s a hot topic by itself recently, but I try to be there a bit also because you can find a lot of information actually about AI. Many people building product with AI on Twitter. And finally, if you are interested by AI, I have a blog called MadeBy.XYZ. Basically, I interview creators and founders in the AI field. So if you are interested by AI, it’s probably a good place to go to read about these people who are building with this new tech.
Mirko
I would say, okay, yeah, super cool. Now I’m really curious because I asked Nelson and I ended up in this interview. So what would be your recommendation with regards to topic? A method, a tool, an expert I should talk to to get more ideas on how to rethink processes?
Benjamin
I would think about a couple of former colleagues because I used to work at Mural. So this digital whiteboard and I have a couple of colleagues there that I think might be interesting. I can send you their LinkedIn profile after. There is Mark and Jim Kalbash. I don’t know if you know them, Jim, I think he was the guy who wrote the Jobs to Be Done or is working a lot on Jobs to be Done concept at least. And Mark Tippin is a great guy from Mural also, and he’s half creative, half analytic. I mean, these two people I think would be super interesting. I can try to put you in touch with them. For example, I would think about, yeah, cool.
Mirko
Yeah, let’s do that. That’s super cool. Great. Is there anything else before we leave this flight today you would like to share with our listeners?
Benjamin
I think it was a pretty good overview about AI and my take on AI and process. But once again, I would just say twice, three times, four times, just try the tool. Try to not judge it at the beginning and try to go further and really push the limits of what you think is possible to do with the tool, at least, because maybe it can do something that you don’t think it can already do. So you just experiment more and more and I think you will be surprised. Try to not be afraid and think about how it can help you make everything better. Instead of, it’s going to replace me, therefore I should not think about it, I should not use it, I should go against it. I think it’s here to stay, at least for the near future. So, yeah, just try the tool. Experiment like you would do with everything else. Usually don’t fall in the trap of I’m afraid of this because when there is a revolution, usually it goes through these stage. I would say first, it’s ridiculous, everyone makes fun of it, it’s never going to happen. Second, it’s dangerous. Oh, actually it can take my job, it can do this and that. So everyone is afraid and then it’s obvious everyone use it and it’s just normal. I think right now we are in the dangerous phase. Some people are still in the ridiculous phase. Just keep that in mind and just try to understand and maybe embrace it.
Mirko
Yeah, cool. Benjamin, thank you so much for all these super interesting insights. Before we leave the aircraft, how would you describe your flight experience with just three words?
Benjamin
It was challenging, interesting, and of course, pleasant. Yeah, I really enjoyed the conversation with you, Mirko. It was challenging because, of course, I’m not like the normal process guy, I would say, and interesting because yeah, we could speak about many topics that I experienced in my life and I could connect the dot and share about that. So I think it was interesting.
Mirko
Yeah, definitely for me too. So what I’m going to do is next I just go to Chat GPT and try it out. So thank you so much for being my guest, Benjamin. It was great talking to you and I’m looking forward to further interactions in the future. Thank you. Bye bye.
Benjamin
Thank you, Mirko. Bye bye. Thanks, everyone.
Mirko
Let’s recap today’s new process inspiration.
Yeah, honestly I just had to stop myself from testing Chat GPT any longer. I directly headed over to OpenAI.com to log into Chat GPT after the conversation with Benjamin and I started playing around with Chat GPT. So I asked for as already said in the interview, I asked for ideas on how a state of the art process for goods receiving looks like and that was super interesting. So there were great ideas on the different steps and what to take into account for a state of the art goods receiving process. So Patrick, that may be something for you. I also asked to list ideas on how AI can be used to rethink processes or to rethink business process management. Actually, results were not really satisfying. There was a lot of AI in the answers, but not so much human centricity. That’s what I expect with regards to rethinking processes. Also, that’s why I then redefined my prompt a little bit and asked for ideas on how AI can be used to get to a more human centric business process management. Unfortunately, also these answers were not perfect. Was talking about personalized customer experience, predictive analytics, chat box use of virtual assistants. Yeah, that may be something. Automated workflows, predictive maintenance and also accessibility. That’s an interesting point. So how AI can be used to make products and services more accessible for people with disabilities, for example, such as speech to text or text to speech technologies. And yeah, there were also some ideas with regards to continuous improvement, but I think I definitely have to work on my prompting skills. So obviously there are some limitations in the knowledge base of Chat GPT with regards to the new process approach. Yeah, let’s see how we can change that in the future. I also asked Chat GPT for the elements of a good podcast recap and I even tried to create a recap based on the transcript, but I failed. Maybe my prompt was not that good. So I did the recap on my own. And here it is. Before the flight we started with the check in. We learned that Benjamin’s relationship to processes can be described as complicated. But based on the subway experiences he told us about, I would say he learned to appreciate processes in the end. And during the flight at cruising altitude, Benjamin and I talked about how to apply AI tools to rethink process, or better said, how to use AI tools for different BPM use cases. And that was super interesting. We also talked about AI generated art and the art of process modeling. And finally we approach our destination with his top recommendations on how to rethink processes from his perspective based on his experience. So for me, the key points we discuss with regards to processes are the different ideas on how to apply AI to processes. So asking tools like Chat GPT for ideas to improve processes like I did with the goods receiving process or just ask the tool how a state of the art process could look like. So these are some interesting aspects which I haven’t taken into account up to now, and I think it’s good to have a browser window open with Chat GPT all the time to put in some questions there and get good answers or new ideas. I also learned that right now there is no AI, at least we were not aware of an AI for process modeling. But uploading BPMN 20 models as XML might be a good way to let the AI bot learn about the process and then the user can ask questions to be answered by the bot, so that may be a promising use case for the future. I also love the idea of having a virtual copilot for employees so to ask questions on how to execute processes for example, but also for process owners and process architects on how to manage and improve processes. And that would be super interesting to have a good knowledge base there and then use AI bots as a copilot. Actually, I’m wondering if this idea was already picked up by BPM vendors so we’ll see what the future will bring there. After playing around with Chat GPT, I think we can at least use AI to generate ideas for research and brainstorming to generate answers there. I also have in mind that Benjamin warned us with regards to emotions, so AI can very well list ideas on what to ask people about their emotions, but it is really bad in being empathic interpreting emotions and feelings. So maybe this is good news for all fans of the human centric BPM approach out there. Yeah, as a personal reflection, I’d like to add that I really see the need to explore AI tools and their application to process in more detail in the future. And I see the potential of having copilots that this will definitely change the way we work as process architects and process owners, but also the way how employees retrieve process information. I’m a little bit skeptical with regards to change and transformation, so all these softer aspects with regards to emotions and so on. So I think an AI generated process model will miss the joint process of creating a common picture, which I think is most important in the process of process modeling so that the people come together and they develop a common picture of a process and then use this for the transformation afterwards. But I will definitely explore this in more detail in the future so you’ll find all relevant resources and links in the show notes. Easiest way to get there is to go to newprocesslab.com, episode 29, and for sure check out Benjamin Dehant’s LinkedIn site as well as his blog madeby.XYZ, but I’ll also put the links into the Show Notes. Finally, by the way, Chat GPT recommends to end the recap with a call to action. So I’d like to ask you to try out Chat GPT to rethink process and then share your learnings with the community. So share the results, share your learnings, share a screenshot of the conversation on LinkedIn. Feel free to tag me and I’ll take it from there. So I’m really looking forward to get your feedback on this and get your ideas on how to use AI to rethink process. But for now, thank you very much for listening. Have a great day. Bye bye. And auf wiedersehen.
Mirko
Before you leave, chat GPT did not recommend to provide an outlook, but I can tell you that there are more inspiring interviews with method experts are coming up on how to rethink process and some more tool related episodes are also in the pipeline. So don’t forget to follow New ProcessPodcast or even subscribe to my newsletter by going to newprocesslab.com update. Yeah, that’s it. Have a great day. Bye.
