GA 650 | Train It Before You Trust It with David Osagie-Igiehon

AI is a tool, just like we have the motor vehicle. Before now, people were riding horses, people were using camels, people were using cows. But now we have a locomotive. Look at the evolution of locomotive. When they first saw the iron horse, what did they say? They were scared. But today, humans are still in existence, but the iron horse is also aiding them in getting to their destination. So AI would be there, but humans are not going to get extinct.

Critical thinking would never disappear because people are Today, I'm excited to welcome David Osagyegihan to the show. Now, David was actually here in our studio here at Gemba Academy. We were doing some, some live events and he was here and we got to chatting. I was like, man, we need to do a podcast. And so we had a really good conversation. And really the topic today is something that was probably on a lot of people's minds. I know it's been on my mind and that is artificial intelligence. And so we get all into AI today and really how it sort of pertains to, you know, to the business world, continuous improvement, and everything in between. So it's a fun conversation. I think you're going to enjoy it. I know I did. So show notes can be found over at gembapodcast.com. This is episode 650. Again, gembapodcast.com and look for episode 650. Okay, enough from me. Let's get to the show. David, welcome to the show. How are you?

Thank you very much, Ron. I'm fine and I'm happy to be here.

Yeah, man. Well, we've had a good day. We're here doing a, um, kind of like a roundtable here with a bunch of different, uh, Lean, uh, enthusiasts from all over the country, really. It's been, uh, it's been a great day. So, um, yeah, excited to have you onto the podcast, man. So—

Thank you very much.

Um, well, as you know, we like to begin all of our episodes with our guests sharing a quote. So what do you got?

Hey, my quote for today is, nothing good comes easy. Keep up a positive attitude all the time.

What's that mean to you?

What it means to me is that everything that you have to do, you have to ensure that you keep an open mind in ensuring that things work according to the way it is planned to work. That's why I said nothing good comes easy. So it's planned to work in the right way, and so you have to follow that accordingly. I've had this quote for more than 30 years, and it's been in my personal email For the longest. Nothing good comes easy.

Love it.

Keep a positive mindset.

Good stuff. Good stuff. All right. Well, hey, tell us a little bit about yourself. Tell us about your background and maybe how you got involved in all this continuous improvement stuff.

All right. So, um, I, my first degree is in accounting. I like to always state that so that everybody understands the deviation.

Okay.

So, and, um, I've been in risk management, internal audit, and internal control and compliance. But pivoting from that field of finance and non-financial institution, I found myself in manufacturing in Canada. So I find myself in continuous improvement in S&C Electric Canada. And today I'm the lead LPS consultant for the company. Ours is to drive the system to achieve the desired objective because we want to empower people to transform the grid. So I started as an assembler. From there, progressed into a Lean facilitator, to a senior Lean facilitator, to a lead, then from there to a lead consultant. So today we try to drive the system, and Gemba Academy has been part of that journey in making that become successful. I'll be wrapping up my master black belt shortly as part of the journey to becoming one of the Lean experts. And when it comes to Lean, I'm crazy about Lean.

Yeah.

Yeah. And I can speak about Lean even when I'm sleeping. Oh, that's great.

That's great. That's great. All right. Well, so today I guess we're going to kind of dive into the whole idea. I know you're passionate about artificial intelligence and Lean and how they sort of interact. And I mean, I just want to open it up with like, what, what, how are you using artificial intelligence and, and making it to sort of, I don't know, supercharge and power your, your Lean efforts?

Yeah. So currently now I'm actually working on becoming an AI consultant. So you can see infusing AI and Lean together, it's like driving the entire improvement system to the direction we need it. Like I keep saying, we are flying right now. So what I've been able to use AI to do is to transform the system activities or problems that would have taken us years and years and years to come up with proper documentation. We've been able to minimize all of that timeframe and shrunk it to little. So we are not using AI tool as another system. We're using it as a collaborative tool, not a driving system for our business. No, it's like, it's like another tool that supports our lean tools to be able to derive better problem-solving activities.

Got it. So, I mean, I don't want you to share anything that you're not supposed to be sharing, but I'm just curious, like, can you give some examples maybe of, of of something that's maybe unlocked because of AI that you perhaps weren't doing, say, 5 years ago or something like that?

Yeah, something personally I would say there are some things that AI has actually unlocked for me. Now, asking questions in terms of doing research from different providers in terms of different authorities has been like a heck of a work for me. Most times if I have to do research, I need to read like 6, 7 different books to be able to get an idea. But right now with AI, I have developed an agent called Dominic. Now I created Dominic to be a super agent. So when I ask it questions, Dominic can give me that answer right away because why? I had to create Dominic like an authority when it comes to lean, when it comes to strategy, when it comes to quality. That's how I created Dominic.

Okay. So, okay, let me stop you there. So, um, Are you comfortable sharing what AI platform you're using primarily?

So individual, as a person, I use Perplexity, I use Lovable, I use Claude, I use Jimmy. Yeah, Jimmy.

Gemini.

Gemini. I keep calling it Jimmy. Gemini. I use Meta and I use Copilot. So these are the 6 AIs.

No ChatGPT.

ChatGPT is there, but I don't— it's not a tool that I use.

I see.

regularly. I use— I'm heavy on Claude, I'm heavy on Meta.

Okay.

Yeah. And I'm also heavy on Perplexity. For Copilot, it's just because it's an Office tool, I have to use it. And I've developed it to the point where it can actually spill out information the way I need it, because you have to train your agent to be able to understand what you need.

Okay. So Dominique, let's talk about Dominique. So in This is with Claude, I guess, or?

No, Copilot.

Copilot. Okay. So dumb question, 'cause I mean, I'm not, I wouldn't call myself an AI expert. I mean, I just get in there and ask it questions like everybody else. But when you say that you've sort of trained it or you've developed it or whatever it might be, well, what does that mean? Like how do you do that?

So basically you, when you say, we'll say you're training an AI tool, it means to the activities that you want it to engage in, the areas you need it to focus on. Like the lean tools that you use, the methodologies that you adopt, the type of project that you want to engage in, and the type of project that you're currently solving problem on. When you train it to understand how your structure of work is, it begins to understand you when you ask questions. And most times it has to be by the type of prompts that you write. When you're prompting, sometimes some people ask questions like they're asking Google, regular questions. AI does not work like that. The quality of a prompt will determine the quality of the output.

So give me an example of a good prompt.

So a good prompt would be, um, if you ask, like, if I'm asking Dominic, hey Dominic, I want to write a strategic document to deliver to my board, and I need you to address this as a master black belt in a strategic approach using economics and accounting figures. to determine how this should be. Your structure should be classic to the 6 approaches strategy and also to planning, uh, approach. Could you please develop this? Then before, before you put up the statement, you say, make this a mega state, a mega prompt. And it wants to make it a mega prompt. It's going to create something that is out of this world. That prompt is what you take out, like what I do. I'm going to share my experience here. So what I do, I take that prompt and I put it under Claude. Claude is the crazy one, and Claude is going to provide for you that particular detailed document, how you want to present it. That's why today consultants, they don't stress themselves anymore when they need to present PowerPoint or maybe any proposal. When they— you use different tools, right? For me, as somebody who's knowledgeable, before I go fully with what AI has given to me. I also have to do my own review to ensure that whatever is being delivered is exactly what I need it to be. Yeah.

I got a kind of a silly question. People are probably like, Ron, why did you ask that? But I'm genuinely curious because you mentioned that, you know, said, hey Claude, could you please do blah, blah, blah? What do you think about like, should we be polite to our chat agents or does it not matter?

Being polite is subjective, right? Depending on what you're trying to ask it to do for you. Yeah. If you ask the wrong question, one thing I've realized, if you ask an AI the wrong question, it's not going to respond to you. Right. Like Meta, Meta is very strict. If you ask anything about a specific family issue, how do, how do I, how do I make my wife happy? It's not going to respond.

Oh, I didn't know that.

Oh no, it won't respond because I think it has some restriction.

No kidding.

Yes, it would not respond to you.

Oh, I didn't know that.

Oh, I'm having this challenge with a friend and I want to deal with the issue. It would not respond to you, but you can reframe that and say, if you find yourself in this situation and you need to address it in this manner, how would you go about it? Then it's going to give you a cue on what to do.

Okay.

Yes.

Okay. Um, well, let's talk a little bit more just, I mean, obviously most of the people listening now are in continuous improvement and things like that. So maybe give us some examples of some of the other things that you've done with, uh, with, and be as specific as you want, you know, as far as like, what did you prompt and what was the result? What did you learn and what were the benefits of it?

So one of the projects that I would say that I've done has to do with governance framework, because governance framework, if you have to sit down to actually design one, you need a lot of input from different authorities. Now, what have I simply done? I have been able to write a proper prompt that allowed Claude to draft for me 125 page governance document.

Wow.

Yes, because you want to see a framework, you want to see a policy, you want to see a standard operating procedure all in one that allows you to drive a system the way you want it. But it doesn't stop there because your prompt has to guide it. And because of the way you have designed your agent, your agent will be able to take information from authorities. I made sure that I created my agent to be able to dig deep into authorities that have created that. So like Claude can go into histories of papers that are out there in the public domain. Like you have a write-up from McKinsey, like Deloitte, like KPMG. All of these things are in the public domain. So it goes into all of these searches across the globe. The best is what it's going to put based on your prompt. I'm very, very specific about that. Based on the prompt, the quality of your prompt, determines the quality of information that you generate. I've also used AI to write movies. Yeah. And that's how interesting—

Like a script, you mean?

Like a proper script from start to end. But it has to be to your prompt. What do you want it to be? I've also used AI to develop a video of myself cooking in the kitchen with just my picture and giving it a prompt, guiding it on what it needs to do. Majority of the reels you see today on the internet space, majority of them are prompts, proper prompts written to generate videos.

How long does it take you to write a prompt for something that's going to be, I don't know, you're writing a movie or something like that?

Depends on your level of AI knowledge. You can write a prompt in 3 minutes, a mega prompt.

Wow.

Yeah. Depending, because You're using— that's why I use it.

What's a mega prompt?

A mega prompt is like— I'm going to describe it in layman's terms. A mega prompt is a prompt that allows you to get the best out of your idea.

Okay.

But you could not do it without the help of AI.

Okay.

So I'm like, if I ask a generic question, I want to go from here to Illinois and I need to know every single lake that is on the way so that I can get to my destination, or rather enjoy my destination. I say, make this a prompt. Now it's going to rewrite that. It's going to say, act like a travel— it's going to now return back, act like a travel agent who understands the geography of the region, determine what bus stops are along the way, how many lakes, the size of the lake, and all of that. Now I can take that prompt that it has given me.

And give it back to it.

And give it back to it. I say, thank you, or awesome. This is a very good prompt. I want you to be very specific. I need you to tell me what's the population size in the different areas and what are the restrictions in this area. Then I say, I put before you, I say make this a mega prompt, and then it's going to write a mega prompt and be detailed. Now when it writes the mega prompt, based on the quality of the different tools that you're using, you take that mega prompt and put it in another AI tool. But you have to know what these AI tools can do. That's the most interesting part. So not all AI tools that you put in a mega prompt can give you exactly what you need. I, I, like I say, I find interest in Claude.

Mm-hmm.

It has helped me reduce time in delivering quality work. Quality work being that I actually know exactly what I am looking for and properly structured the way it's going to be appealing to my customer. My customer being either the board, being senior executive directors, or all of that.

Okay, so that leads me to my next question then. Like, how do you Fact-check it? Like if, let's say you're just, it's doing something, it's sharing something that you don't know everything about, right? Like, how do you ensure that it's not hallucinating or it's making a mistake or something like that? And next thing you know, you're up in front of the CEO and he's like, no, that's not right. Sorry. You know what I mean?

So that's the additional piece that you do. That's a human interface. Like for every prompt that is written and the feedback that comes, there's a reference that it actually tells you this particular session is gotten from this, this particular session is gotten from here. If you want to do additional work, you have to go verify. But I always say, before you ask the computer or the AI what to do, it means you too, you are knowledgeable.

Yeah.

That's why you have to read your prompts to ensure that your prompt actually is aligning exactly what you need, not what you want, because sometimes what you want is confusing, but what you exactly need. And once it delivers that, you have to do a review. Without— you don't just take it hook, line, and sinker. No, you have to do a review, and with that review, you can correct its errors. And that's the beauty about AI.

Okay, so next question, um, what are your thoughts on Like one of my fears, and I don't know, maybe I'm wrong or right, I don't know, is that I think about my children, they're growing up and going to school or whatever it might be. And obviously they're going to leverage AI for various things or whatever. Do you think that there's a risk that the human species is going to forget how to think, forget how to learn, forget how to problem solve? Because right now, I mean, we're going to look back at Claude today and 5 years from now and laugh at it. It'll be like, you know, how, you know, like the original PCs or whatever, you know.

The risk is there. The fear is there. But this is the way I will ask— I'm going to answer this question. Where is BlackBerry? Nowhere.

Not around.

Where are the pagers?

Mm-hmm.

Evolution must always happen. People would always think. We cannot say because AI is in existence, critical thinking is going to disappear. No, it's a phase in the life of the humans. It will come in, people are going to stabilize it, just as people are trying to make AI better. That's how people are behind the scene putting in controls because nobody's going to let it run wild and not be able to control it.

Well, there's people who build bombs and are terrorists also, right?

Still, still, still being controlled. Yeah, right. So I mean, it's not going to extinct humanity.

I hope not.

It would not, at least not in our lifetime, right? See, I, I tell people, kids of these days may not remember the movie called Space: 1999, but today, are we in space? We live in space.

Yeah.

We fly to space and come back like it's yesterday.

Yeah, yeah.

So that's the same thing with AI. AI will come, it has come, it's going to stay.

Yeah.

But how, as a, as a people, The thought of how to control it is what is important.

Hmm.

And there are lots and lots of people behind the scene working on that to control it so that it doesn't go wild. So in my opinion, our kids are there because there are other curriculums that are going to help them think critically. AI is a tool, just like we have the motor vehicle. Before now, people were riding horses.

Yeah, yeah, yeah.

People were using camels, people using cows.

Yeah.

But now we have a locomotive. Look at the evolution of locomotive. We would say when the iron— when they first saw the iron horse, what did they say? They were scared. But today, humans are still in existence, but the iron horse is also aiding them in getting to their destination.

Yeah.

So AI would be there, but humans are not going to get extinct. Critical thinking would never disappear because people are superior than machines. AI is a machine. It's just a tool. My thought is everybody use it right. Just like you're driving your car, the speedometer says 300. You're not going to drive 300. You're going to drive within the speed limit. So those are the controls in place. Just like that, AI too has a control.

So let's say we've got some continuous improvement practitioners listening right now, and, you know, like everybody, they're dabbling in various AI tools. Are there any, like, I don't know, specific tricks or techniques or whatever? I don't even know a way to ask the question that they may not know about that you think that would add value to their life?

Yes, there are lots of tricks when it comes to using AI as a tool with CI work. Reason why I say there are some tricks is because the quality of knowledge that you have on how to use AI is also very important. If you ask any AI tool like you're doing a Google search, it's going to give you the minimum information and basically wrong information.

Mm-hmm.

But what is the best way of using the tool is being an advanced person in AI, you should be able to create your agents. Today, there are some people who have created like 6 agents. And when you call in, it's actually an AI that is answering you. But because you have built the agent to know your product, To know your schedule. It responds to you and it answers you like it's that person. The voice prompt and everything you're thinking is the person that is answering you, but it's because the AI has been trained to speak like you, to act like you, to respond like you.

How do you train an agent then?

So this is the interesting part. Training an agent Is based off of what— who you want the agent to be. David today is an accountant, is an internal auditor, is a risk analyst, is— he has a financial idea, he is working in CI, he is an assembler. Put all this information together, consider David as an agent. Then you ask David, I would need you to solve a problem in assembly. that requires a reduction of time. Make sure and ensure that the time that you're going to be saving can actually be converted into cost saving. Run this over a 10-year period and generate the report.

So you're basically, while you're creating the prompt, you are building their persona, so to speak.

First thing, you create the persona.

Okay. So I log onto Claude right now. and I just start typing in there, you are—

So that's not an agent.

Okay.

That's Claude's own agent.

Okay.

I'm ignorant.

So, okay. Maybe everybody's annoyed with me right now because I don't know what I'm talking about.

Sorry, listeners. So, and you create, where you create the agent is that there's a portion in every tool, right? That says create an agent.

Now, I've never done that.

Yeah. Okay. We'll get to that.

Okay. You're going to teach me this. We can do it.

So when you try to create an agent, you give the agent some specific qualification.

Okay.

And it's going to ask you, where do you need me to generate the source of my information?

I see.

Then you can specify where you need it to generate information from. So take for instance, if I need to design a super CI specialist, I can create that agent and say, take information from Gemba Academy, take information from the Institute of Engineering, take information from ASQA. Now, and I say, I'm creating this. Who should this agent be? This agent should be a doctorate— should hold a doctorate in engineering. hold a doctorate in business administration, hold a doctorate in commerce, hold a doctorate in supply and logistics. Now, with— then what skills, or what skills should this person have? Empathy, uh, whatever you need to create it to be. Once you divide— you design that behind the scene, you've created that agent. Now the question you're asking it is a complete different thing.

Yeah, okay.

Now you see that you have built something that has a wealth of knowledge. When you ask it a question, it's going to dig deep into its qualification, into the search engine that you have referred it to look into, and the tools that it has to develop that.

Why is that better than just logging on to Claude and asking regular Claude?

Because it does not know how you structure yourself. Because I can also create another agent that is a salesperson. If I ask a question, it's— I'm confusing it, right? It will respond to me what it sees in the cloud.

So you build an agent that's a super continuous improvement expert, knows everything Gemba Academy's ever taught, everything the whole world of lean has taught inside and out. And then you say, now it's built. And now you say, okay, help me analyze this value stream map or whatever it might be. Something like this.

I have now, when you create a prompt, you're asking it specific question.

Okay.

Assuming you want to say it's a, let's assume it's a new company and you want to build the structure for the company. You're going to say, I want to create an engineering company that produces XYZ. and I need it to have all the value stream and all the processes and everything designed for me, the system. Now, without making it a prompt, it's going to give you generic what it sees from— let me use Google.

Yeah.

It's just going to give you generic. Now, when you make it a prompt, it's going to specify exactly how it wants that particular agent to think. That's the prompting. Because it's just like, David, what is your first name? I tell you, right? What— David, could you tell me your origin, how your name came about, where you lived, where you were born? You have prompted me. That's how the prompt works. The way you want to make it a mega prompt David, tell me where you were born, how you were born, the school you went to. Structure it so that I can use this to speak to your profile from the day you were born to up till today's date.

Hmm.

Now, when you make that a mega prompt, it's going to structure your profile the way a profile needs to be structured. And then when it delivers it, like I say, it's a tool. It's not the final habitat. It's a tool. You review it and ensure that everything, its structure makes sense to you, and then you can make corrections where you need to make corrections. So that's how AI tool can support our CI. You can't rely on it 100%, but to a large extent, the AI tool can help you reduce over 100 minutes or 100 hours of work that you need to do, and it can reduce it to 1 hour.

Oh, I know. Can you— well, I mean, you can probably just do things you literally could not do.

100%. Yeah, that's, that's a fact.

Yeah.

But the good side is this. This is what I tell a lot of people. As much as you know that this is what you need, because it's important enough for you to really know what you're doing, you can't ask a question and present it. That's the mistake a lot of people make today, that they use AI in producing their resume. Then you go physically to go defend what you say you are in your resume. You can't even stand the chance with one of the questions.

Yeah.

It's just like we're talking now.

Yeah.

If I don't understand how to use AI tool, I won't be telling you this. I have used it, I have seen the benefit, and I'm still using, and I'm still going to use it. The first time I told Meta, I said, hey Meta, this is my qualification, and I'm also going to be an AI consultant. What do you think? You know the response? It says you're going to be a crazy interesting person combining continuous improvement and AI. So you're going to be an exceptional person. That is how good AI can be.

Mm-hmm.

Because it knows that AI can actually generate information that you have never had opportunity to lay your hands on. And that's the information you are seeking. So that's how good AI can actually help us improve our processes because it's not just going to give you exactly what you need. It's going to give you best practices, but it's based on your prompts.

Yeah.

What type of prompt are you writing for the AI tool to generate the information that you need?

Last question, 2-part last question.

What are you most—

what gives you the most reservation, if anything, about AI, and what has you most excited?

My reservation about AI is about, especially for companies that have proprietary, it's something that we really need controls on so that people don't lose their royalty, right? I have seen— I've actually written a prompt and the information that came back, I was never meant to see it. That's how dangerous AI can be.

Hmm.

But what excites me the most, like I'll tell every professional, is this: AI can make your life simple and also difficult. Why did I say simple or difficult? Simple in the sense that you can use it if you understand it. Difficult if you use it and you present it and you don't understand it, then it's going to make your life very difficult.

Yeah.

So the most important thing is before you use an AI to generate anything, even if it's a report even if it's a presentation, even if it's a paper, even if it's a book, please be trained properly in that field. I, I can say I'm proud that I am a student of Gemba Academy, and being that I'm writing, I'm finishing my master black belt, I understand how continuous improvement can actually work closely with AI tools and how it's going to help organization. Because every organization want to be— want to get to market on time and to satisfy the customer. And in order to do that, you want to be able to solve your problems. And solving problem takes time to do.

Yeah.

But when you have— when you ask specific questions Normally, if you have a team of 6 people working on the problem, they need their personal time. They have other tasks that they need to be engaged in. It's going to take months and months and months to come up with a structured plan.

Mm-hmm.

But if you pick these ideas in one meeting and ask everybody how they're going to engage this, and it's properly documented, and you put it into an AI tool, and write it, tell it to give you a proper prompt, taking each of these ideas to formulate a plan for you. In less than 3 hours, because of your review, you have a plan document on how you want to engage. That's why I'm not— I don't take away the human piece of it, never, because the human piece is very, very important. But what you are trying to do is time management. You want to manage that time that is not available.

Yeah.

And that is the value that AI is bringing into CI.

So when's your book coming?

I don't have any book for now. Come on, man.

Well, AI could write it for you, right?

Well, you see—

But seriously, like, you have so much knowledge, you know? Like, I've learned so much just listening to you. Like, like, I'm ignorant really when it comes to this. And I imagine most people many people listening right now are similar to me. You know, we just get on ChatGPT and ask it questions, you know. And so if—

I, I always say use different tools, understand the tools.

Yeah.

My trainer, she is vast. They call her the queen of AI. I'm not going to mention her name because we're not trying to promote her. They call her the queen of AI. She's vast. And in our training, she has brought in other people who are experts in the AI field to speak to us and guide us on how we go. I was, I was okay. Now I'm going to state this here. So I'm actually writing a book on OPEX because a lot of people, if you've never engaged in problem solving, you would never understand how well operations should become successful.

Mm-hmm.

And if you are in operations, there are some tricks you must know in continuous improvement that allows you to become very, very successful. So the direction of my book is actually on OPEX, integrating more of continuous improvement thought process into every single phase of it and ensuring It actually speaks to real-life issues, not an assumed work or a theoretical work. It's based on practical approach, just like you see me talking today in the seminar. It's the same thing. Yeah, you cannot— you can't— you can fake it if you've never done it. Yeah, it will be obvious.

Oh man, this has been a lot of fun, man. I'm glad we did it. Thanks for, uh, For carving out some time after a long day. Thank you, bro. But, uh, um, if people wanted to connect with you personally, what's the best way? LinkedIn, something like that?

I'm on LinkedIn. It's my full name that is on LinkedIn.

We'll put your LinkedIn and, uh, you know, any other contact information.

Yeah.

So everybody can scroll down and, and, uh, go, go connect. So.

Thank you, bro.

All right. It's been a lot of fun. Let's go have some dinner now.

Yes, it's time. It's about time. Thank you, Earl. I appreciate it. Thank you.

Thanks for joining us today. We know the path to continuous improvement can be challenging to walk alone. That's why Gemba Academy exists, not just to provide training, but to be your partner in problem solving. Whether you're just taking your first steps or leading a global transformation, our goal is to help you bridge the gap between knowing and doing. Beyond our library of Lean and Six Sigma resources, We're here to coach and guide you through your journey, whether that's through personalized support or joining us for a live training to get that immersive, hands-on experience. To learn more, go to gembaacademy.com.