Stop Using AI as a Toy (w/ Alexey Prokopenko)

AI becomes like it should be like a teacher basically treated as a teacher to teach themsel to program properly before coming to work within the organization. So have you tried you know writing code yourself with these tools.

I came to a point when I would say AI systems or something. It's not like for playing or you know like just for me to generate pictures just for because I want to like like a toy basically to play with a toy. It's not like that. Developers must understand AI or how to generate because it helps to progress.

Is AI hype or help? Good day and welcome to the show. I'm your host Pete Cooper and our topic is AI hype or help. Today's guest is Alexey Prokopenko.

He's coming to us from Australia. Like me, he's an IT professional working in the field of developing softwarebased solutions for companies all around the world. Now, he actually comes from Russia, as you'll hear from his accent. I'm actually based here in the US. So, with our common interests, without further ado, let's get into it. Good day, Alex, and welcome to the show. Alex, thanks for joining us. We'll give we'll get into a brief introduction about yourself, but Alexey Prokopenko, you um you have a business in Australia called Kite Union and that's being impacted uh by these big changes in in AI.

Um and as soon as you start talking, everyone's going to know you're not naturally from Australia. So tell us a little bit about yourself, your business, and and how you got there.

Hi Pete, nice to see you again, and thank you for inviting me to this show. Uh yes uh my background is Russian and I came to Australia about 2008 yeah 17 I think or 18 years already.

Wow.

Yeah. So quite a while back. Uh when I came I as all migrants had a job had something and before uh coming to a point to develop my own business it's specifically after MBA finishing MBA I I wanted to do it later on. Mhm.

So I started this year it will be 10 years when I started this kite business kite union.

I call it kite because of it like letters in it and that's a pure reason why kite because of it and my business is just focus on it and was always focused and no other intention. It was just focusing to doing something but

flying high understanding resolving somebody else's problems. It's all about that. And

do you come to this from the business side or did you come from more from the development side?

Uh from two sides from a business and development in Russia I had in like many years back I had clients in Russia and I sort of already get a sense how to deal and how to do business how to run a business.

So it's a just natural growth or progression for my business to come to conclusion to open a company in Australia

and and yeah so the major focus was on custom app development and uh I found it's very interesting industry because PTO as well in IT but uh in in in IT industry everybody like there are so many solutions out there so like all problems are resolved in in many different ways. However, the big problem on the market, they don't resolve just for your business. Whatever business they are, they don't resolve 100%. You need either to mix, you merge or adapt or find out or doing experiments to get it things together. That is was a big problem when I it's still there after 10 years.

The problem is still there

but in a sense of linking these two people need some sort of custom solution. It doesn't mean that it has to be one like more than you know like more like better than any other term system. It has to be something tangible on top of other solutions which will give you specific need like custom dashboard custom adaptation one or another hardware or tracking devices or specific we did some streaming application for streaming TV. So it's a for streaming for specific um retirement villages it's everywhere there is a little trouble. So this is how I came across tokate union and started doing this business.

Yeah. Okay. So you and I have something in common.

I mean you've you uh come from Russia to Australia. I've come from Australia and I'm now in the US. So uh I think your challenges coming across you know culturally and uh language as well would would be greater than mine. Although I had my own set of challenges coming to the US. That's for sure.

Yeah. Um, so can you just illustrate a little example so that everyone can really ground down without sharing any um any trade secrets or anything proprietary. Give us an example of a project that you that you've worked on, one that you're proud of.

Uh we I'm proud of many of them. We've done many different mobile application during this time. We done web application.

I I would say main focus was and to develop um uh mostly web applications and what I'm proud uh when we started how do you say it's not like a dashboard it's when an analytical tool on top of other term system so when term system produce all knowledge collects all data together and we had few solutions like that we link over API API application program interface if Somebody doesn't know but anyway so we link one software with another software built dashboard and making intellectual analytics on top of that and one of them was about retail stock analysis.

Another one was about um health issues and another one was about tracking devices and basically to to give a hint where everything like a like an item is located which floor which level and so on.

So this is pretty much common in many of solutions but again knowledge on top not knowledge it's a it's not just simple data analysis it's not it's more about giving control giving it like dashboard or power to connect or turn it on to switch it off and this is what I'm proud and I have not just as I said like not just one few of them similar kind of structure

yeah we we did something too Alexa I was just telling you about the project before we joined where we actually took we're in construction space and we took uh a lot of data from a device which was out in the field and we put it onto a dashboard and we had you know good visualization tools.

You could see uh you know minute by minute all of the changes for the health of the system but also you know for operationally like what is it doing and what are the results of um the work that it's doing. So like you it's using um you know backing databases it's using the dashboard visualization.

I mean I think when people worldwide web is about websites you know and you go to a simple website and you know you see the pictures and the marketing information uh without realizing just how much there is

that's right

happening with the within the data that's moving around in the background the APIs the applications um program interface

when I was studying being in mccquori university over here we had one lecturer was a co of one big company I I I don't want to call the name but anyway so that company and that uh guy it was teaching operation management in MBA program and one uh operation management more how to bring efficiency within the company and he was talking about Kaizen

like a Japanese philosophy or lean manufacturer and so on and he actually took us like his class to his company where he was running and he showed how he changed the flow uh without lin and v with a lin and it impressed myself during the time. It showed me how to adapt dashboard, how to deop control, how to build um because that is very complex and it's very sensitive for health because you know like you to in order to go to production site you need to do like spend 15 minutes to cleaning yourself before coming.

It's a very serious production but but anyway so everywhere they had dashboard controller like screen big screens display and it's all just about software how to control all this stuff this is what impressed myself before running kiteun before starting it and that's where where I'm sort of like looking towards to as example to improve operation efficiency of any business

let's dig into that a little bit this kaizen principle Um, from what I understand, it's like a a board where you you move things from the to-do to the in progress to the complete and it's put up there for everyone to see. And it's all about making a flow.

So, not to have too many things in uh in the pipeline as or in each column, I guess, is a better way to say it. You don't have too many things in each column. That gives you focus on the po important thing until that's resolved and then it gets moved over. Am I getting that right?

Yeah, that's completely right and completely correct. Yes.

Yes. So that Toyota manufacturer production uh when he introducing his Toyota it's more like to build a conveyor you do nothing till you get a job. So you and the less waste like waiting basically to eliminate waste of waiting period of time.

So basically when you do nothing that is a big challenge for organization the whole philosophy just about eliminating waste. So basically whatever it's waste like and build efficiency and and in in in world where we are people adopted is a more agile approach.

It's it's exactly the same logic but it's just put they put in agile approach and agile and I would say right now the more and more we moving for forward even agile sort of it's kind of like probably lose interest and basically people losing interest in agile but they still it's a demand when you do the job when you nec like when necessary like it's a SAS platform SAS uh any software working like that like by demand they charging calculating how much you need to pay for the service,

right?

Yeah, it's everywhere.

But my point it's still that particular dashboard whatever in that product in operation that was impressed me and one of my driver why it's very important and very necessary um to understand it's very complex subject because you need to understand how to template any like when I you or myself as it um consultant as IT person as a providing IT service come to whatever business there I have know sometimes I have zero clue about their business but I need to understand how they run why they put let's say this hammer on this place not on that place and why because I cannot even answer why but but when you s when when I sort of collect this information understanding this is really interesting how

to pack software just for that business not bringing my ideas or whatever somebody sells software towards to that business and say you need to run this way that's that's that's a I would say dilemma or challenge between business side uh the business side and IT side

how to bring all link between the

I think it's an interesting and we talked a little bit about this coming in too about about the the problem where is the problem I mean I meet many many business people uh all the time they don't usually open up and tell you oh look I've got this problem over here. They're usually telling you how great their business is.

Um, and it's not until you sort of get um inside the business somehow you get a way in that you can see the problems and then if you see the problems then you can design a solution to it. Um, yeah, I think often that access is one of the biggest problems, you know, for businesses like yours and ours is getting that access to to to people to open up and show you where there's opportunities to solve problems. I think there's the sense of vulnerability that have that they have um and a sense that they don't want anybody poking around in their business, right?

Yes.

So, there's a lot of trust you have to that you have to win.

Of course. Of course, of course.

And uh uh our or my industry depends on the trust only.

It's like yeah website is is one source of like trust. But people not coming because they saw nice picture somewhere. They coming because they know they by referral by you know like by communication by like like building relationship and it's very important.

And after relationship is built of course

like a project comes comes later but but not not before because it's industry specific understanding it's a lot of stuff you know like yeah it doesn't work like like I met you today and tomorrow we do business in our industry not yet but when we building u solutions online yes it may happen there but it's a bit different industry it's more product based not consultancy sort of so service based

yeah a bit different

so let's let's talk a little about the theme of of the the of the show so AI now what's interesting is we've been talking for 15 20 minutes and we haven't talked about AI yet we've actually talked about

all of the stuff that to

me AI cannot affect and that is the relationships we have between people right and uh or what is an organization a company or a company is an organization is a group of people now that's still true today. It may not be true tomorrow, but at the moment it is. So, are you using AI to help you with that side of the business? The the relationship, the the bit that we were just talking about. Uh yes. Uh in every step I'm using AI a lot and I'm not I didn't start even this year. I started I think at 2024 or even earlier.

Uh so um where but of course technology progress since then and more um incorporation into business development process or building custom software or content generation it it started um growing inside every particular point of the business uh differently.

So what what I mean for example let's say start with a simple conversation of course my English is a second language and sometimes I'm making mistakes I need to check it somewhere somehow you know like like yeah so this is where I mean I like like if I how do you say if I say I'm perfect in English I'm still not in perfect perfect in English so it still requires some development in in writing style and so this is helps the first point it's simple but it helps the second uh point I at the beginning we spend so much so many hours to develop software businesses software requirements because when people comes they say I have an idea I mean I'm generalizing here but some people have idea what

they really want but many people have I got an idea I need an app with no details with nothing so basically I I I address many many times when I develop my idea the idea I I I wrote the business requirements for themsel and then come into development basically from just one sentence I had to do everything.

Yes.

And it took a lot of time in the past. Right now it's efficiency is much faster. Uh less people required to develop same and faster and more advanced documentation.

So that's where I would say AI helped us um to structure to how to say to to to deliver something quicker because before I spent on quotations hours you know not just one hour hours before coming to write how much software should cost but right now I don't spend those I mean I spend still those hours but much more advanced documentation will be produced within like same time frame right So, you're using the LLM tools, you know, that's I guess the

the big jump. Um, and I I I've gone all in on those as well and found that I can not only do better work, I can do faster work. Um, but you brought up something else which um I want to explore.

So, you said something like

from one sentence you can develop all the requirements.

So, there must be a few things going on there. Must you must make some assumptions. You must build on your experience and you must use the information that the LLM creates for you, right?

Mhm.

Correct.

You must have gaps though and I guess you must go back to the client and say, "Hey, look, I need clarification on these things, right? "

Correct. Yes, that's that's right.

Uh uh I uh in the past uh as I before AI started progressing uh I had so many different type of clients completely from different industry and because it was my skills I sort of could play by myself one or another type of industry or business and because of course it's my personal experience helps me to to do those assumptions and um how to pass it to AI yes it's probably maybe to build skill set or something like that like but again it's too broad or too generic you still need to have some senses you know in a personal senses way not AI senses you know because AI can capture but not everything and in this sense what you ask how to address those assumptions how to address those point uh I

had my measure always how quickly I can predict and how many changes I have to make if I didn't guess this right and and yes of course it requires talking and sometimes it requires conversation but but still AI does not feel that particular model yet even if if it generates based on a skill set complete document based on assumption it may be right but I don't know maybe if in the future it will be okay but at this stage it's still not very perfect yet

right so you have to have some experience so

yes

I think that's where um I I guess people who are a little older have a slight advantage over those that are younger.

Uh because if you've always had AI there, you don't know what's right and what's wrong. Right. There's a

Yes.

I heard I was at a lecture um about a month maybe two months back professor of one of the senior universities over here the top universities was saying he's an AI specialist saying he doesn't believe that a you know um AGI you know artificial general intelligence is a thing. He thinks that we don't have the right tools for that.

I mean the AI that we have is great does a lot of things but AGI it's not there and the reason it's not there is because there's no understanding of the real world there's no physics there's no you know I can't throw a ball with my hand that's going to travel three or four miles I can write that down and the LLM can get it all wrong but it's not going to work in in the real world in physics now this is in construction space

so it's very you know very physics it's very uh reliant on physics I mean, you can't build a building unless you have the fundamentals of of physics. Um, and there's not a lot of that sort of embedded in all the information that the models are built on. Not yet.

But I guess in your world, in the IT world, there's a lot of information that these models are trained on. So, it's probably got a good chance of getting it right, but still not 100%.

Uh, it's coming to that point. It's coming.

But the challenge after using let's say like some services to generate a code uh I we found and I I see progression is not making it better it's making more overwhelmed with no reason you know it's basically like like let's say in a simple way to say if we need to generate a simple like procedure or function or something before it it could generate like simple piece of code not not complete solution but simple and it was more likely right but right now because it tries to build broader more advanced and assuming that I need everything but I don't need everything and this is where sort of control with the linguistic um it's a bit hard to say something how much you need in a language where you

express what you has to be done and assumption is much more bigger it's a And sometimes it reminds me myself like how I was guessing from one sentence too much business requirements. It's a sort of same same is happening right now. And this is where control has to be put in a place somehow. How much is enough simple or more advanced? And and maybe I mean it's they they haven't done those kind of control yet. But I I assume it will be fine. It's just

just a matter of time when they will do it. Yeah.

Yeah. One thing as we're talking I I wanted to change a little bit. You have you have at least two languages Russian and uh English.

Um do you think that's an advantage when you are trying to you know um articulate uh you know building requirements or in any other area

I uh definitely yes I while I was in Russia I didn't understand this advantage.

Uh even I learn like language in Russia but but still I couldn't get an advantage I could get an understanding where like knowing another language not just Russian any other kind of language is advantage of understanding mental philosophical side of one or another language it's not just how you structure your sent for for example in let's say let's take it back to Russian and English in Russia it's a language which itself is a little bit philosophical because we can put suffixes, we can add prefixes, suffixes so and we can construct any word in many different ways or shape or whatever can create on the spot.

In English you cannot do like that but because of that Russian is built on a lot of assumptions assumption it's a metaphorical.

If you get famous books like dusty or some some others you will see the linguistic is very comparison you know like it's a metaphorical speaking in English not much metaphorical talks but in English it taught me um more precise procedure focus basically I do this whatever I do and it's more easy I'm using with a if we talk about our subject different type of languages for different type of reasons because if I write a if I write in English it's too simple I cannot express some shapes or grades or feelings in it

if I write in Russian I can make it better

because of language yeahing because yeah when you write prompts in Russian it gives you more different basically pretty much more

faster towards to your goal what you want

if you front up through an LLM let's say ChatGPT or something and you start writing in Russian. I'm assuming you got the keyboard. Will it answer you in Russian

uh as well as it would answer you in English or, you know, can you mix and match? How do you

how do you handle these different languages now? Because I've only got the one, so I have no idea.

That's But it may be worthwhile to start learning another one.

Maybe. Yes.

Yeah.

No, I I really found knowing uh different words like different how to structure sentences whatever one or another it it opens your mind for other ideas and uh in in a sense of uh uh if you I mean if somebody will check my um profile on YouTube Kunian channel you will see I'm doing content generation like songs or like business videos or all AI generated and in by doing that uh coming back to a language by doing because sometimes I'm writing poems I can generate I can reuse somebody else's poems but still doesn't matter from one language to another I still need to adapt it right to to sound it right in one and depending how I write poems sometimes I'm using like let's say Russian poem first

before English or sometimes English is better than Russian

yeah wow and and it's a yes language is very important. The more languages you know I mean I I found the more languages if I learn much better I would say it you will be more how to say more towards to achieve your goal you can achieve faster.

So he the thing the question that I have for you though is mo I would have thought most of the data that the LLMs are trained on is in in the English language you know

American Canada North America you know Australia Europe you know there's a lot of English language the model does the model work better in English than it does in Russian or does it not care

I think I mean I don't know details I didn't look into source code of them how they produce it but But but but my assumption it's a model is the same uh capacity of um how do you say um capacity to generate results I would say it's exactly same but uh like let's say if I write in Russian it may answer in English depending how I put

sentence I can mix two different languages so basically if I say I write write me in English, it will write in English even when I put a Russian whatever poem there. But but on the other hand um it this is what is really hard to explain.

I I mean when when you when people start doing it they get the skills why why I'm what is the difference between one and another language but because this is little sort of fine tuning I would say like it's all kind of like let's say professionalism in one language or another language or in a business industry because fine-tuning it comes with experiments and and with my experiments specifically as I said like for content generation I found it's very important depending what kind of like feelings I want to put in let's say music

like to make it more sensitive to make it more vibe not just I I cannot say generate um like in English it's too simple generate me sensitive so in Russian I can

say many different shapes how to address this particular sensitivity you know like if I want to generate pro more more colorful I would say more wider not just cinematic because in English I Oh, that's too simple.

No, that's interesting. I mean,

I appreciate English.

We've had we've had lots of guests, but we've never had never gone deep into using LLMs across languages. So, I'm going to take away that it doesn't seem to matter. Um, and in fact, maybe you can do, as you said, even more because you have this multi- language rather than I was thinking that maybe the models don't work as well uh with other languages, but that doesn't seem to be the case. Yes, it's not the case.

It uh it's in the music for example what I found mixing languages it sounds very nice.

I mean I haven't published those particular songs but mixing like Russian with I know I have one Russian and in English sounds very nice and uh and when you bringing like Asian languages like Indians or whatever Sri Lankas or any other non maybe not many people speaking even I don't speak but still it brings that particular sensitivity of that language how they seen in certain certain words and the whole composition becomes more richer I would say more more advanced than just simple you know

that's why I see beneficial of using this AI in the past uh in the past it was really not possible because I didn't know other languages I couldn't speak any whatever outside of Russian English I

couldn't talk properly in another lang but right now having this capacity I can advance my skills even

so

to next level

so here here I am in in the US and I deal with people in uh other countries but it's always English English speakaking. Um one assumption that I wasn't even I even thought about it. I mean if somebody in another country in any language jumps on to ChatGPT they can effectively use it. If I'm in Mongolia or you know speaking Mongolia and if I have access to that I can use it. That sounds that sounds like what we're saying. It doesn't really matter what language I'm using.

That's right. And this is and this is big advantage of this all AIS AI systems.

However, the main problem of AI in this particular sense of different languages you will not be able to chat right now like on our podcast what we are doing right now. It it would be really really hard if I would start translating every sentence. It wouldn't sound natural. That is disadvantage. So I have to speak this language, English, in order to communicate with you. If I don't speak, okay, maybe simple language, yes. But you wouldn't talk, you know, to me.

Now, we're all there's a very famous book called the um Hitchhiker's Guide to the Galaxy where they have this thing called the Babel Fish that you stick in your ear and it trans it translates all languages no matter what perfectly.

But I think you alluded to the fact that there's just

there's structural things um and nuances and the way the language is has a foundation which which doesn't really allow you to do that easily. So

that's right.

Yeah, that was interesting. All right. So where else in in AI? I mean the obvious one we're both using and that's you know to build software. I mean has that affected your your business? Uh yes in a good way not in a bad way. So um

I attended a while ago. I think it was 2025 or yeah I think it was early 2025 somewhere.

Um I attended in video workshop in Sydney and uh I'm using that as a quotation because on that workshop they use uh to use they they compulsory they push every developer within their organization to use AI AI tools and uh like a shell to basically to generate a code and without those skills pretty much developers wouldn't be able to sustain within my organization. I came to the same approach after I heard it from bigger company. I I I approach exactly same model to start adapting for developers to understand developers must understand AI or how to generate because it helps to progress.

However, it's not how do you say in the junior when junior developer comes with zero knowledge, it's a little bit hard.

They use AI but they have no zero clue why and and this is AI becomes like it should be like a teacher basically treated as a teacher to teach themsel to program properly before coming to work within organization because before teachers uh sometimes didn't know some new subject but AI knows the new subject so basically if you talk to AI by yourself it will be like a second teacher teaching yourself to a level and then you start working within organization and for me for my business it's very important that developers need to know AI AI does not allow I mean you can generate full complete solution but the problem is a complete it does not structure proper architecture basically uh how to make

sustainable growing up or building like longlasting software solution it's not addressing that properly if developer has no skills, it it may be problem, you know, like a temporary solution is okay, but longlasting it's not okay.

Yeah. So, the way it was explained to me by one of my guys was if you have a problem with the code, you can use AI to fix it, but it's like putting a band-aid on it and then you have that problem comes back again, you got to put another band-aid on it and you end up just patching the problem because you never really had it set up structurally correct in the first place. So, I mean, I have my guys, too.

I say, "Look, and and I, honestly, they had to convince me um and they did um to use the tools, and I know I pay for the tools, and they're so much more productive. Um, but I say to them, you have to still know every single line of code. You have to know it all. " And do you say the same thing?

Um, not really.

Uh yes, knowing the code is very important but not really to know the every single line because the challenge is here

uh you need to use a shell in order to understand what change was adopted or to use the language to address specific change but knowing the full every line it's not necessary because it's a lot of stuff like a templated in software like like authorization process whatever swagger or some some other technical terms, but it's still already uh there. You don't need to think too much. It's more just how to set tuning, right?

But but knowing the change is very important because if you don't if you if they don't know the change, it may cause issues later on and you will have no clue when the trouble will come.

Yeah. Right.

Because it's changing very easily a lot.

I I see it in a graphics uh even it's not related to source code but when you generate like a picture or like a video because in my case I'm generating longlasting videos like the longest about 6 minutes or and to keep consistency on one object basically like visuals or something it's really really hard because it breaks so much if you don't control it right and and same in the code visual is more easy to sort of point but

to see but in code it's not really easy but but logic is the same it's the same it may

may keep or help it may break

yeah yeah I think that's a good segue the way you've taken it down I think we've all had that problem maybe with ChatGPT generate me an image you

know of a you know a teddy bear and a rabbit on a slip you know on a seesaw uh and it does something really weird like it's it's a pig and and a rabbit or something and you say no no I said you know a rabbit and a kitten or whatever and it just can't it doesn't seem to get it right no matter how many times you try

it got some of it a lot of it right but something fundamental that really stood out that made it wrong

I came to a point when I would say AI systems or something it's not like for plane or you know like just for me to generate pictures just for because I want to like like a toy basically to play with a toy. It's not like that.

AI I would say in the long run will be used more professionals. Professional means when you understand how to write prompts or you train yourself how to write prompts, you will be pretty much faster than anybody else. But to teach how to write prompt to generate a perfect result from one sentence, it's a like it's a art I would say.

Well, I would say I was in Dubai this time last year. Um and they said that um they wanted to have a uh an initiative or a university that was going to have 1 million prompt engineers um over the next few years. They were going to de you know that was a thing and that's how committed they were um to that.

So you know you'd have on your resume you're a prompt engineer not a developer with a language set. So have you tried you know writing code yourself with these tools? Oh

yeah I um my my background as you know like it's always was in software. I was I have degrees in software development. Of course I know many different languages software I mean languages and and of course I have to experiment by myself otherwise I I cannot advise and and in this sense it's very important for me to do that. I mean I'm not writing all my solutions.

I have developers to write but still I'm experimenting in order to understand what what is there why why is they talking I without that I I wouldn't have an experience and I and AI is very related towards to IT engineers because um it's not for let's say for art as well but more closer who people who create doing some something creative it's very good tool to to be used to like and because of this I need to know the tool otherwise how can I talk to my developers

right right I so I I um fired up the um claude you know anthropics version to write some code I had this idea for an app in 5 hours I had the app fully fully working fully debugged on my iPhone and iPhones a little bit more tricky

I don't know any of the languages that you know I think it's swift that that the iPhones run on and I just was able to get it to run. And I wasn't really interested in the code. I've never really been that keen on learning code, but I'm very interested in the process like being able to take those requirements and and run a a process to build to create a result. That was very interesting to me. And I was able to skip that, you know, very deep dive you have to do into the language and learn the language. I was able to skip over that.

Um, and it made me realize that you can be quite comfortable working across different code bases, you know, iOS, that's right,

Java and, you know, Android or whatever. Um if you can I guess position yourself more like what you suggested where you're you're you're really handling the requirements and ensuring that you're delivering a good quality product and understanding the the process and and dealing with the people rather than um you know going down to that syntax level which

yeah and and Peter to add to this point u I develop JSON viewer JSON viewer. Kun. Com.

Au to you it's a portal like it's there is no reference on my website but I'm saying like this is a subdomain but anyway so the whole point in my industry was very important to sometimes to see JSON like JSON file basically the structure in it and a lot of online tools out there but always it was the problem because I didn't trust those website I didn't know where they save on information or not saving right now by using AI I by myself I generated this particular page or portal or like one page and I I I built all this donation kind of like stuff like component like with a stripe I also built specific advanced tool correcting JSON file because sometimes there is a mistake you need to correct

and you don't know where it's the problem I adapted to all different experts and all this particular like like export JSON file into database in code whatever whatever we use it basically I build it for myself. I didn't build it for somebody else. I built just for myself. And this is where advantage come of AI. You can develop a tool for yourself. Doesn't matter whatever. Maybe it's not a perfect, it doesn't matter. If you want to bring it to production, you can ask developers to do. But understanding how to build the tool for yourself. What I I always wanted, but I never wanted to invest in it by by pain, you know. But right with AI capacity, it's it became possible.

Becomes possible.

Yeah.

Look, Alex, uh we're going to have you back, but we're running out of time. There are so many more questions I've got for you, and we're going to go deep because we have so many things that we have in common. Um but we're going to bring it to to a head. Do you have any final comments before I ask you the final question?

Uh I would probably say uh just not to you for everybody who is listening this I would say find a way how to adapt AI find a way how to find tangible for your business it's not me telling you how to adapt it's you need to understand where like to be creative to to replace some stuff to improve some stuff to reduce time like spent on doing something find a way anybody they have to find a way how to adapt. It's not a replacement jobs or something. It's about good analytical tools which analyze one or another industry and give you good suggestions and up to you to use as much as you can but my recommendation would to find a way how to adopt. AI

I think that's a very good way to to leave it.

So for one final question, Alex is AI hype or help?

Uh probably after my pitch today, you probably can hear it's a help. Uh but hype I would say it's a built around hype is more uh it depends how people apply. You say if I say like I use AI to generate perfect picture, no problems. You can use it as a hype. But AI can generate a perfect PhD picture and can speed up your business and it becomes as a help. So basically convert hype into help and you will be fine.

Yes. Well, thank you Alex. It's been a real pleasure having you on. As I said, we'll have you back. There's a lot more I want to dig into. So thank you for coming on the show.

Thank you Pete for calling me.

Okay.

Thank you. Byebye. Well, I hope you enjoyed that as much as I did. Now, Alexe was an expert in many many fields that are very similar to ours and he started his career in software development. So, he uses LLMs like ChatGPT and Claude and others to help him with requirements development. Now, traditionally that would take many many hours to write up these requirements. Now, with LLMs, we can do it much more quickly. Of course, we have to make lots of assumptions and there's lots of things that you know that can certainly go wrong if you just trust the LLMs.

And we both agreed that you know experience um adapted and helping us to much more quickly do a really good job but not uh not to allow it to completely um run without oversight and that's where sub experience really really comes into the four. We also touched on the importance of relationships in business and now corporations are groups of people and and at least at the moment that is what a company is. It's it's people working together. I don't believe that's going to change anytime soon and neither does Alexe. And importantly, we both believe in the importance of human relationships.

And then the third aspect we touched on was how the tools for developing software have really revolutionized the whole software industry. How it's made code creation much quicker and how we've both adopted it um very fully in our companies to develop uh better code to develop quicker code to develop features more cheaply um but not to let it run without understanding the structure of the code because it becomes like a band-aid on top of a band-aid well I hope you found that podcast interesting if you want to know more about kite union please click the links in the description if you want to know more about Skillion please do the I hope that helps and thanks for watching.