157. Lean Coffee Duesseldorf - Special - HUMAN ROBOT AGENT with Jurgen Appelo

6:28AM Apr 11, 2025

Speakers:

Keywords:

Human Robot Agent

AI co-pilot

psychological impact

sociological impact

cognitive load

organizational structure

modular team

value streams

vibe coding

software engineering

beta readers

feedback efficiency

AI in healthcare

AI in aviation

agility.

I like, who started Juergen, did you start the recording?

Oh, I will do

yes. Okay,

let's start again.

So and I get the chance to preview pre reads the book. And I enjoyed doing that. And we have a discussion. Say, Okay, we do this link, COVID, discipline of regularly, and then two base, two weeks base. And perhaps you like to explain something.

And I think I

by yourself, because, as I saw the US first slide where you say, where coming from, which kind of books you read already, or read or write already, and Jurgen, salam, all the other please go on mute that we have a chance just to listen. We decided to start with a introduction speech for 3040, minutes, and then we start with a discussion. And we can discussion with Juergen and with us. What do you think about the topics and so on and so on, what we can take from this presentation, and so on and so on. Jurgen, it's your speech.

Alright. Thank you so much. I assume you're talking to me because we have a Jurgen, a Jurgen, a Jurgen, a Jurgen this call. So it's, it is almost confusing. Um, anyways, great to see everyone here. I see 33 participants. 34 awesome. So yeah, early morning, I'm barely awake, but

ready to go,

I'd like to talk for a half an hour on this book, human robot agent that came out a month ago in various formats. Very proud of it, and I have some fascinating insights to share. At least. I find them fascinating, up to you to judge if you agree with me on that. I wrote other books before, as you know, perhaps I wrote manage theater. Oh, managing for happiness. Startup scallops grew up. The last one was actually a novel glitches of gods. I spent about 4000 hours writing that one, a bit, if you ask me. But when I finished the manuscript and had not published the book, yet, it became

good feed. FC, is very useful. And there I saw the power of using AI as as a co pilot in in in writing. So human robot agent is the last venture. And I have to be honest here, I'm not an AI expert at the technical level. I know a bit how llms work, about neural networks and so on. But I don't go deep there. My interest is in the psychological, sociological angle. What impact does aI have on organizations, on how we work, on teamwork, productivity, and so on. And I noticed myself that basically it is the bottleneck of the of the human brain that we try to address with AI, because the human brain is not very good at calculating, memorizing, repeating things, executing things consistently, etc. That's what AI is good at. One example that I like to give is that I fired my beta readers because, well, how it usually goes with beta readers is that 100 people sign up, and then each chapter, more and more people drop out because they're simply unable to to keep up the discipline. It's a bit like going to the gym or in in January, the gym is full, and then month by month, you see the attendance going down, and by the time I get to the end of the book, there's only one beta reader left who is probably just as autistic as I am and simply unable to complete, unable to stop doing the project. So

for me as an author, I.

That is useless. I have two feedback on the first chapter, no no feedback on the last chapter, and it takes weeks to all of those problems are solved. I have multiple big readers. They are available instantly, Gemini, chat, TPT, you name it. I can configure them to look at specific things, and they do a marvelous job at that. One specific example in software engineering is that we use AI to address the bottleneck of coming up with prototypes with MVPs, some of you may have heard the term vibe coding is all the rage these days. That's the whole idea of having a conversation with an AI about an idea you have, and then having actual working software as a result of that vibe coding session that you play with you

the oil code is supposed to be tossed out after that. That was the original idea of vibe coding. Of course, people take it to other levels, and now we have already, we're already seeing software engineers trying to vibe go to real applications with all the

consequences.

Is that the

architecture in

for a code,

and for me, that is

a increasing more.

This is a model in the focus on the

quality of the environment, and I see that A

because no problem maintaining that

career for some Engine

everyone else was crap type code.

So this

is that,

and they don't do any trust exercises on their way to the to the fire. So why are they able to do that? Also in the airline industry, where there's a lot of re teaming across airplanes, they are able to reduce cognitive load thanks to technology. In hospitals, as you can imagine, when a doctor moves from one room to another, that doctor doesn't need to have all the knowledge of the patient.

I think we have an internet problem.

Can't hear you internet problem. Okay, I don't see a problem here. Okay, they the doctor sees an entire dashboard of of metrics. What are, what's the previous treatment of the of the patients, the previous medicine, what is the current status of heart rate, blood pressure, you name it, and and then they can do their thing. So as as in this article on cognitive load, it says, while the human to human interactions are shaped by cognitive limitations, collaborating with AI opens up new possibilities that challenge our traditional understanding of of cognitive constraints. So, yeah, we're able to organize ourselves in a different way very soon, because of AI, because AI is reducing cognitive. Load,

as I said, has advantages in terms of agility, because you're making your organizational structure itself more more agile, again, that links to the red part of the of this model, more modular team structure, a looser structure. Okay. So another thing that I've noticed is the disintegration of value streams when we enter into.