#221 Mastering Prompt Engineering And Building AI Products In Public: An Insightful Conversation with Ali Abassi

0:00:02 - Mehmet
Hello and welcome back to a new episode of the CTO show with Mehmet. Today, I'm very pleased to have with me joining from Vancouver, Canada, Ali Ali. Thank you very much for being on the show. The way I like to do it is I keep it to my guests to introduce themselves, so the floor is yours. 

0:00:17 - Ali
Wonderful. Well, thank you so much, mehmet. Thank you for inviting me to be on your show. I've had the opportunity to listen to a few of the prior episodes, so I'm very excited to be here and chat with you. So I'm from Vancouver, Canada. I'm a marketer by trade. I've been in marketing for, I would say, 10 years. At this point, when chat upt came out, I became obsessed with AI and, like many people, have just tested and tested and tested to a point where I came to a discovery on how to, how to leverage it in a meaningful way, and I'm building a product and products around that. 

0:00:57 - Mehmet
And yeah, that's the brief description. Yeah, yeah, that's great, but it's an interesting move from marketing to kind of something related to tech. So let me ask you what excited you about ChatGPT the most? 

0:01:15 - Ali
Yeah, I would say the thing that excited me the most about ChatGPT was the unlimited opportunities and the potential. Right Like, like you mentioned, it's very unique for somebody who's in the trade of marketing to start going into the AI space and start to build product, and I think with a tool like with ChatGPT, a marketer can now code significantly easier, allowing them to 10, 20, 30, 100x their potential by automating and by creating products that just amplify their work. So, yeah, I would say the exciting thing about it was just the amplification opportunities that an individual can have. 

0:01:59 - Mehmet
Yeah, yeah, of course, like this is. I've seen this, you know, happened a lot in the past. I would say now it's been like almost nine, 10 months since ChatGPT is out of the wild and a lot of tools out there. So, ali, I understood that you specify in what is called prompt engineering right. So myself I've been using it. I cannot say day one, maybe it was like four days, until I forget that it was out of the wild. I knew about OpenAI. I always wanted to understand their API. I didn't have time to explore, but I was seeing people creating app space on the API, but then they get it out to the wild. And here we go and we started to hear about prompt engineering. So I want you to explain maybe we didn't do much here about you know the art of prompt engineering and what are like some of the advanced techniques. 

0:03:00 - Ali
Yeah, great question. So let's start with the art. Like, what is prompt engineering in the first place? It's really just talking to a computer, it's talking to a machine and with tools like ChatGPT, you can now do that in natural language. In the past you had to do it through code, but the learning curve for writing code is quite high. Not everybody can write code. So having a system that allows you to just simply talk to it and ask for what you want is ultimately prompt engineering. The challenge is there is getting consistent results. The variables of natural language what one person says and what another person says with a different, with a few different words can really change the outcome. So, yeah, prompt engineering is a natural language that one uses to speak with the computer. 

When it comes to advanced techniques and the things that I've discovered, what I found was that, as a marketer using ChatGPT for some of the most common use cases write content, write social content, write website content, create marketing briefs sorry, creative briefs or website copy. 

What I found was that most people are accepting the first draft. When they talk to ChatGPT, you say, hey, please write an email, and then you get frustrated and you're like, wow, it sounds like ChatGPT wrote it. It's so obvious. But in any situation, if you had an assistant, if you were working with a small marketing team, you never accept your team's first draft. There's revisions that are required, there's evaluation and improvement. So what I found out was, if I can ask ChatGPT to evaluate its own work, to give advice on how to improve itself and then give it a loop so it continues to improve itself over and over and over again, your output is significantly better, and that was what I put together when I put my prompts together, which allows the average user to get, honestly, really good results from the platform instead of accepting draft one. 

0:05:03 - Mehmet
Yeah, that's 100%. It's something also I explored myself as well, and for me, it was not by chance that I discovered this, but because, when coming from a technical background, I like to understand how things work. I understood how ChatGPT or other language models they generate their content, so I figured out that actually they need to iterate on multiple contents themselves and this is why I started to ask ChatGPT okay, can you do it, but in this tone, can you do it in that tone? 

And then I bring all of them together and say, okay, tell me which one is the best. And then, yeah, like really amazing. So Ali decided to build a business out of that right. 

0:05:59 - Ali
Yeah, absolutely so. What I decided to do was once I figured out my prompt framework, which is a bit of a unique framework. It's written in JSON format. The prompts that I have all start with a task, a role, success factors, a criteria that that has to utilize, and then steps. So there's rules based on that. 

But ultimately what ends up happening is a user with any level of experience Can pick a prompt for example, right me a marketing plan. They put in my prompt. Gpt asks them a series of five questions. It does the work, it evaluates the work and then gives itself advice and then it asks the user you want me to revise the work based on my feedback, do you have any questions? Or so on? So, like once I figured that process out in that framework, you know what I did is I built a community called a effort work. There's about 2700 prompts on it and there's about 140 different Job titles categorized within departments marketing, legal, medical and so on. 

I wasn't ready to launch it, but I spoke with an individual on tiktok just in the comment section and he found it and he ended up making a video about it before I was ready to put it public. And I woke up one morning and there was 400 people who signed up Wow, and I was like, oh, that's interesting. And then at this point it's about 40 days in and there's 12,000 people who have signed up. There's no marketing, it's just this one tiktok in a lot of word of mouth. So it shows me that the need and the desire for Knowledge workers to really leverage this tool and maximize their results from it and improve the way they work is quite high. I think the interest there is there. 

0:07:42 - Mehmet
Yeah, I've been, you know I'm, you know I checked your website and because you know this is something I want you to talk about. So people think Usually that, okay, you know, we know it's good, you know charge EPT or any other, maybe the other ones also that are catching up now like bar and Lord, but you know in general so people think that the only way I can use it is to you know marketing content. 

But when I checked your website, you know I've seen plenty of use cases, right? So sure can can you, you know, elaborate more about you know what else other than just marketing content that you know people can leverage this technology for. 

0:08:24 - Ali
Yeah, absolutely. I think what we're seeing is this tool, specifically GPT, for when we compare all the options, is the Most powerful one at the moment. There, it's the one that provides the best results. What we're seeing is we're seeing an expert co-pilot for any role and from different use cases, the ones that I found to be very effective legal right, like not only writing contracts but Figuring out the steel man argument in a case, or the strawman argument. 

0:08:57 - Mehmet
The next one is medical. 

0:08:58 - Ali
There was just a report that came out that chat GPT versus doctors. Chat GPT got a right about 76% of the time. 76% of the time I've used prompts to help diagnose situations for my son, who's one years old, after seeing, like before seeing his doctor, and I go to the doctor's office and I know exactly what I'm talking about. I have the exact say, the exact outcome that the doctor's coming out with and he's quite impressed. He's like how, how did you know that? Right, like so. I think the use cases is anything that requires expertise or information. Can now this can be a co-pilot and help guide you to the right answer. 

0:09:41 - Mehmet
That's fine, yeah yeah, not content, but Like, If you want to convince me, like, imagine now I'm a business owner, right, and let's say I'm kind of an you know, I'm technology little bit for sure yeah. Yeah, you need, you need to you know I'm not doing this question on purpose, by the way yeah, and you want, like, in two, three sentences, like you want, to convince me why I should consider Leveraging AI into my day-to-day business. So what you can tell me? 

0:10:18 - Ali
Yeah, so say, you're a business owner who Is anti technology, but you think in terms of investments in and out inputs and outputs, I would say that the that leveraging AI is the greatest you the greatest leverage opportunity in the workplace right now, for the simple reason is you can hire A non-experience individual and empower them to think and to perform like a much higher experience individual. 

For example, you can hire a virtual assistant in the Philippines and ask them to make decisions based on a framework that you provide. You could ask them to leverage chat GBT to build SOPs To complete tasks, quite advanced tasks. So I would just say you get significantly more out of your, out of your current team. They can all perform better, which allows you to just be more competitive in the marketplace. But then the second thing that I would say is the investments happening outside of this one conversation amongst a business owner. That's anti tech and reality is every company Literally every company at this moment is trying to figure out how to use AI and if you ignore it, you're going to be left well behind. Google announced that they're releasing duet AI, which is going to be an assistant for every employee chat GBT enterprise. So the technology is happening so fast and, whether you want to use it or not, how essential it becomes for your business. 

0:12:05 - Mehmet
I think the market will dictate that 100%, and I asked you this question as I was telling you. So I will explain why I asked this way. Because Some people they think it's a hype. Some people they think, yeah, like, we still need. You know, that's usually. 

This is what happens when you know a new technology comes up, right. So back in the days when something new, they would say, okay, let's see a couple of guys that the early adopters, let's go with it and then we will think about it. But I'm telling people, guys, this is a moment different than any other technology or new technology moment, and it's like a one that happens every 10 years maybe. So the previous one, where, in my opinion, it was the smartphone, you know, yeah, so, and you know we had like a couple of things before that. And this is where you cannot ignore this technology. 

And you know, like I remember people who ignored the mobile game. At that time they lost, right, but here you know, they had like at least two, three years time frame before they they figured it out. But here I think you have weeks, right. So things are coming very, very fast now. And I know you talked about GPT, for, but have you tried yourself, ali, to evaluate the other ones also as well, like Bart or I know that from API perspective, but it's cool, they didn't open it yet, but I mean from prompt perspective and you know people sometimes they say that the Anthropa, the cloud cloud one, is like better sometimes. So what you can tell us about? 

0:13:41 - Ali
you know this topic, yeah, okay so I've tested I wouldn't say all of them, because I haven't tested the smaller, lesser known ones, but all the popular ones, being barred Claude, open AI. I've tested them all and I would say that there's no such thing as a perfect prompt, right, the what is a user's goal and what can be achieved. After testing all of them, bing is by far the worst, and it's the worst because it uses GPT-4, but it's very, very, very safe. It's a safe to use tool, and what I mean by that is you could just go back and forth with it and it ends the conversation. 

It says I do not have the opportunity to answer this and it just it's like if you're working and you need a real answer, it just it doesn't get you to the final, to the final stage. It's annoying to work with, unfortunately. Bard is a fantastic tool, but it's not as good a GPT-4. So it'll do a lot of different things. So, once again, there's no perfect prompt. So it depends on what the user's goal is For Bard. I think many people will get what they need. And then I think you know the cloud is a fantastic one. 

The context windows is 100,000 characters, which allows you to put in long PDFs, long podcast transcripts and ask questions with it. I think it's incredibly useful, but others will catch up to that. And then, when it comes to testing all the different open AI ones, gpt-4 just blows it out of the water. The prompts that I have on AI for work only work with GPT-4. So it's just the reasoning capabilities and the fact that with GPT-4, when you put in such a large mega prompt, it doesn't forget it after three conversations, after three back and forths, which is incredibly impressive. 

0:15:38 - Mehmet
Yeah, unfortunately, where I live, I know like people they told me you can put a VPN and you can try a cloud. But I don't want to do it this way. I want to put my hands on it and I need access to. I need maybe in the future, access to the API. But I have a friend who tried it and it said good, bing, yeah, I agree with you, it's like very limited, but it's good when it comes sometimes to like more recent information that it's not in the knowledge base of chat. Gpt right, totally, totally, but to your point as well, and I'm waiting for the do it to come out and see what enhancement they have been doing. 

Google, bart, it's fantastic, but for me sometimes it's the same thing that you talked about Bing, so it stops the conversation. And the other day, for example, I was saying you know like, I need a summary about this topic and this is you know some information. And it refuses because you know like, oh, I don't know anything about this topic. Yeah, it's all that it needs to go and search, and you know like. So I didn't understand exactly my problem and you know, a couple of minutes back I repeated the same prompt. I didn't change anything. 

Maybe a couple of you know commas and you know dots here and there and here you go, but I never faced this. I never faced this issue with with chat, gpt, honestly. And the reason I'm asking you this because people in the past few weeks they started to say that the performance of GPT for like it's degraded a little bit. I didn't see that, but you know from someone who's been like, working extensively and building the prompts, what have you seen? Do we really saw some degradation in the model and really GPT 3.5 became better? What's your opinion on this? 

0:17:45 - Ali
Yeah, great question. So I would say there's, there's a trick here. So the trick that I found to be working the best because, like, when I thought about creating the platform that I created, which is diverse in the sense of so many different roles, so many different levels of expertise will be using this I had to figure out how to get the most consistent results from GPT for any different skill level. I tested natural language prompts but the problem was that the it never gave me a consistent result. So like if I, if I wrote a paragraph and you tested it, and then I tested it, and then my wife tested it and her friends says everybody would have something very different come out of it. But what I found is that if you can write your prompts in a structured language like Jason, that have parameters, like rule steps and provided a great deal of guidance, the results are always very, very consistent, which is odd because this is a natural language platform that should understand natural language. 

But I found that there was drift specifically with GPT, for and you can even see it when you're, when you're on chat GPT they say the model number on the URL and I found that as I refresh it, the model number keeps changing. It's using an older model or a different type of month and I think they're likely doing that just from a capacity standpoint, to make sure the platform stays live. But what I what? I keep testing the prompts written in JSON format. The results continue to be consistent. There was no drift, it just worked out better. So yeah, I would say drift exists. They do switch up the models on you, but if you really want the best results, I would say structure your prompts in JSON format. 

0:19:33 - Mehmet
Yeah, that's. That's good point. Just I can add, like my own views now I think the score 3.5 enhanced a lot 3.5. I can see you know the latest one. It's almost as good as GPT, for it doesn't have the what I call it, the reasoning part, like the ability to to iterate. You know couple of prompts still it like this, but it's much better than you know when they were at the beginning. You know couple of months back. 

The other thing, and here where what you do, ali, and you know what I try sometimes to do is you know the prompt itself. So for people who doesn't know when Ali is mentioning you know structured format, this is what the language models they like, actually, because this is how the best algorithm will work From my experience, because I know how I need to give that. So I didn't face this problem honestly much. What helped me is when I checked the API documentation of open AI. So they added something called you know a role of the AI. 

What do you want it to be? Do you want it to be a, an agent, you want it to be an assistant or do you want it to be a user? Right and based on that, when I understood the logic behind that, of course, like not everyone have that, I mean that not everyone is a, is a have, is a tech savvy, I would say, and they would understand the API's. But I try to take this and convert it to you know, to the chat box that we see, and I found that, if I structured the way you mentioned, honestly I didn't see any drifters like on the opposite. And guys, you know, we would say fact and Ali, I want you to explain this also as well to the, to the fall, and this is because you have mastered, I would say now, have you seen it like also beside drifting, but does it get better when it starts to know more about you and what you are trying to achieve? Have you seen this? 

0:21:39 - Ali
100%, absolutely Okay. So if you're in marketing, if you were to ask chat to PT to provide you a script for Facebook ad for your business, the first response to be very generic. But if you were to say, hey, please provide me a script for this Facebook ad, here's my customer persona, here are the pain points that they're currently dealing with, here's an analysis of their customer reviews what they love, what they don't love. And these are three examples of really good advertisements that I really like, the. The output on the second one is going to be so much better and so much more personalized than that first one. So, yes, more context and more higher quality context will definitely get a better result. 

0:22:24 - Mehmet
Yeah, 100%, and I think they added recently, I think it's only for the plus user, so sorry if you are watching or listening to this and you are not a plus. So they have added something which they call you know you can give some information which charge you PT, always remember wherever you are, you know. And then it's kind of a customization, it's called custom instructions actually, so I can tell them, like this is my name, I hold the city or show podcast, this is my consultancy, and whenever I ask a question, it tries to go and relate to that. So this helps also. Now let's jump on something which I it's my best Episode ever and I was doing solo that time AI agents wonderful, okay, yeah, very. 

Yeah, so Because when I stood first time I think it was like two or three months is back. I cannot remember when I when I saw the first video about auto GPT and then later on, baby agi, baby agi, yeah, and you know, all of a sudden they start to pop everywhere and I had a guest and we talked about how powerful they are. But from your experience, ali and I know that you have explored them what you can tell us, what's happening in that space. 

0:23:41 - Ali
So by AI for work, business is really just a community. It's, it's building a distribution. We have an email list and so on. It's not the core function of my focus right now. My core function, what I'm building, is around AI agents for the workplace, so you already know what agents are. I think within two years, every company is going to hire AI employees, digital employees that do critical tasks. 

What I've seen today is there are there are many tools that are trying to be as useful as possible, but they just don't meet the mark. They're not useful. They glitch out. The promise is there, but the reality is that they're just not very useful. They don't get the job done. The best use cases that I've come across are things like research, to help me do research or to gather information. But I think what's what's starting to trickle out to all these new inventions happening literally in the last seven days, eight days is giving us this Opportunity to really build a useful product that can autonomously figure out next steps, so that one can actually build a digital employee. Yeah, so that's kind of a summary of where, where I'm at with the, with AI agents, because I've tested many of the different platforms and I don't find them to be great today. 

0:25:06 - Mehmet
They lack. I would say, actually, you know, when people they start to talk about hallucination of Charge, ept, so so it's like hallucination Multiply times five or times six, it's not hallucination, but but they can go off off Topic. I would say I tried them couple of time. To be fair, I was impressed because you know, at the end of the day, you know the, the logic behind them is very strong, both from a technical perspective and they are actually trying from Some of the guys who participate in these open source projects, when you know they are into psychology and they are into like such things. But as you said, ali, I agree with you what I figured out, that even on the search research part they can drift very quickly. I mean, yeah, for example, I say, okay, go search a niche in this specific and go and Find ten and then put them at the end in a world document or like a text document, whatever it is, because they are like all comment the line. 

You know software, yeah, you know, it's very promising, by the way but I think they. 

Yeah, I think, for For this one we need to wait, but I honestly I'm not expecting to wait much. I'm believing people are seeing years. I believe month is. I think we are month away. Yeah, and I'm, I want to see you know once Google also released, you know their APIs, and when you know like cloud becomes more like globally Available, I want to see, and then we have llama from from Facebook also as well. So all these things happening very, very fast. So this leads to my next question, ali, like what are we going to be doing you know next, how our work Job, what will be our job descriptions actually in the workplace? 

0:27:03 - Ali
Yeah, I think I thought I thought about that an awful lot and the honest answer is I don't know. I really don't know, because I think many Prominent CEOs will come out and they'll say, hey, ai is not going to take your job including the CEO of GitHub, right, he's like programming is not going to be taken over by by AI and he has better visibility than I do In that. But based on what I'm seeing and when I think about it from a philosophical standpoint, I think what what we're going to see in the next two years is the adoption of AI agents in such a significant way, capturing a market share in it, like rapidly. There's a tool called air air dot AI. It does sales calls. It has an agent that Calls, calls your customer, has a conversation with them, books a meeting. That is one of the biggest information and knowledge rules today. 

But if you can have an agent do that reliably, I Can't imagine why you would pay somebody 60, 70, 80 thousand dollars a year to do that job when you could deploy software to do it at $400 a month. The challenge right now with that tool is that it's not perfect. There's a small gap between the conversation, when it starts and ends and the the positive or more, but they're going to solve that problem pretty quickly. And if you can deploy a thousand of these calling agents, why would you? Why would your team Grow rapidly anymore? Right like you would? You would just do so much. You would be able to achieve so much more with such a smaller team. So I think AI agents is gonna Be detrimental to the workplace and I think these promises that we're seeing where AI will not take your job Is because they see that there's no deployable tool today that isn't completely reliable. But, like you said, it'll take months for that to be solved. Yeah, I don't know what the future looks like there. 

0:29:09 - Mehmet
Yeah, no one can predict the future hundred percent. But you know two things quickly from my side. On this one, yeah, I remember, you know, still openly, I think it. They were like maybe two years old, it was 2016 or 2015 where I shared an article. I wrote an article, yeah, and then I asked this question in the future would be would be seeing a machine selling to a machine, right, yeah? 

0:29:37 - Ali
and. 

0:29:38 - Mehmet
You know, at that time, you know, because I was seeing the trend, that automation, and you know, of course we were talking, we were thinking, when we thought about automation, we're thinking about robots, right, not like as a software, but a real robot, yeah, and you know, it came to my mind Okay, so if, if, if a robot can talk well, let's say now it says you know, it's an AI powered voice Software, right, so if it can talk, if it can do calls, if it can send emails, right, fine, because we are assuming the hypothesis here is like, on the other side we have a human. What if on the other side it was an AI? Also as well? Send them like, imagine, like an AI calling an AI say, hey, I have the solution that you should, you should see so. And there is not mentioning this, ali, because you know, I believe personally that the type of work we will be doing is different. So, like, if you are today a sales guy, let's say, or a marketing guy, you will be an kind of a chef of orchestra. I would say, right, so, so you are the one just giving the instructions. 

And there's like this you know, you know autonomous bots, let's call them that. They are doing the job for you and it will not replace your job. And I think when they are saying you will not replace, they are not giving the hundred percent accurate, in my opinion phrase you will. You will not lose your job, but you will not be doing the same job that you are doing today, because your job would be to you know, I would say, operate AI. And if you don't know how to operate AI, this is what will happen. So this is where you will lose your job. So it's a philosophical, I agree with you. But anyway, ali, like I like to follow People who does entrepreneurial work and you know, and for you I think, you are building this in public, right? So tell me, because the movement of building public is, you know, it's been quite some time over there. So tell me, what are the advantages and the challenges of building something in public? 

0:31:59 - Ali
Yeah, that's such a great question because you're absolutely right. We're seeing this major trend of building in public and just transparent founders. The biggest challenge is being vulnerable, right Looking stupid amongst the more established players in the area. Sharing trade secrets that you discover publicly is definitely a major risk. For example, for me, just building in public, I'm finding out things. I'm finding out clever ways to acquire customers at a much lower rate and my competitors are all following me. They're all seeing what I'm doing. 

The major benefit is connecting. When it comes to social media, it's after testing out many different ways to post. The things that seem to resonate the most with an audience is a story of triumph, like being able to accomplish something and they see themselves in you. When you're building in public and you're able to connect with so many people on a deeper level that when you do meet with people through it and I've met with so many people at this point of all different places in their career. But the intent of the conversation is how can I help you? That's the biggest benefit because you're publicly asking for support and people are connecting with that. 

0:33:34 - Mehmet
Yeah, and did you use some other than social media? Because I've seen people and I tried it myself for a couple of months, just for the sake of trying? Have you used something like Product Hunt or these platforms? 

0:33:47 - Ali
Yeah, I've definitely used Product Hunt, but my audience is really small, which is a big reason why I decided to do the whole building public and start to look for new connections. My audience has been really small. Building this platform that I'm building now, building it in public, has allowed me to amplify that quite rapidly. But yeah, I've used Product Hunt before. I didn't get good traction on it because I just wasn't able to drive an audience externally that I didn't have. 

0:34:16 - Mehmet
Yeah, and what are your plans, Ali? What's next? 

0:34:21 - Ali
Yeah, so right now I'm working on an MVP for an AI agent. 

Everything that you said was 100% accurate, because I think one part that you may have missed right, because you stated that I think the future of work is somebody sitting in an operational role working with autonomous agents, and I think, to a degree, that is going to be a stage, but there will be a time where these will develop to a point where it will be hands-off, where an agent will buy and sell to another agent, and the reason I say that is because, based on the work that I'm doing and what I'm seeing is the framework of an agent that makes a lot of sense. 

Right, you decide the task, you decide the steps to achieve it, you figure out the right tools to connect with and you continuously do it. The problem that we're missing right now and I think this is the secret sauce amongst the many, many AI agent companies that are going to come out is going to be the decision-making framework. Right, when you hire an employee, you're hiring them based on their ability to make good judgment, and making good judgment can be mathematical. You can make a rubric for decision-making, and one thing with AI agents is you can have 100 AI agents as a knowledge base to make the decision together and have multiple iterations before the final decision is made. I would bet that within a two-year period of time, we will be able to have AI agents that can make better decisions than the operator and that over a period of time, we'll be able to trust those decisions much better. 

0:35:58 - Mehmet
So what are we doing? Nali, Tell me. 

0:36:01 - Ali
I have no idea, but I think the most common thing that others say is likely some sort of UBI universal basic income where people don't have the amount of hours that they work, that the amount of work is reduced. But I think it's hard to tell, because the real question is when will we first get the taste of this? If autonomous driving comes out in the next 24 months and there's a significant change in truck drivers, that could be the first major hit. We haven't seen such large displacement yet, but we're going to start to see that rapidly over the next two to 10 years. I don't know what the future is going to look like in that scenario and I don't even know how we're going to react to the first iteration of it. 

It was leaked that Shopify is going to be laying off about 1,600 employees and customer success. So the question is like why? What made them to have this decision? Well, they just released an AI chatbot. This AI chatbot helps the user figure out the best ways to use the platform, really just taking over the customer's success role. There's more engagement that users use them everyday versus a success manager that they'll call once in a blue moon. I think this is one of the first times, we're seeing a new AI tool being released to the public and then seeing layoffs being so correlated to it. It's a tough situation. 

0:37:37 - Mehmet
It is. I'm on the optimistic side of this, ali, I can say, because I believe the society will evolve to a place where it's acceptable that OK, because you have to come up with the outcomes. So a human have to come up with the outcomes. You have to put the plan Now, regardless who does this. It's another story. I'm seeing it as another liberation, by the way, for humanity, because if you remember, 150 years back, people were forced to go and work for someone else in their farms and they were getting peanuts, and then industrial revolution. All this, but again we did the same thing. But now we are on a place where everyone has this technology. It's very democratized, I would say. It's in the hand of everyone. So it's up to the people. If you want to achieve financial freedom, let's say you need to learn about this technology. Or you would say but yeah, I'm expecting the changes to be faster than other people are expecting. So people are talking about, usually all my guests and all the experts in the field. They are saying the optimist takes once. I mean in the sense of when this will happen. They are saying two years, the more. They are saying that it's going to take time, regulations will come up and blah, blah, blah. They are saying five years. I'm seeing it shorter than this Because the amount of innovation that is coming out of that and the amount of things yesterday and, of course, when this will be released it will be a couple of weeks after, but just yesterday, although I have technology background, I never wrote code. 

I mean, back in the college I wrote code, but I mean I came up with a full-fledged idea of an app and I was able to give that to something that used GPT and with the interface, with the sign-in mechanism, with the payment gateway, and it's an alpha version. Of course, I'm thinking should I release it or no? But yeah, but if you think about how things go fast, like nine months only ago it was just a chat box, right, and here we go after nine months now, or maybe less. Yeah, we have a huge, completely different landscape for this. So we have to wait and see. So, ali, as I come back to the end, where people can find more about you, yeah, for sure. 

0:40:14 - Ali
So they can find me on LinkedIn, just Ali Abasi. Or they could go on the website aiforworkco. I think they'll find. They'll see how I structure prompts. They can learn from it. But I'm sorry. There's just one thing I have to tell you based on what you just mentioned. Ok, so you mentioned, I'm assuming, you use the platform cursor to complete the application. Did you use cursor? 

0:40:37 - Mehmet
No, no, no, OK. So you're saying, ok, I tried a couple of them, so I think one was called actually all of them. They are good, honestly speaking. So the ones I tried, and just I tried, I was trying them. By the time when this comes up it will be two weeks back, so I'm not sure there will be more tools by the time. So there is one called type block Type block and there is another one called Framer can do this also, by the way, the design tool and there is something called Pico. So Pico is what I use Pico. 

0:41:17 - Ali
Nice. 

0:41:18 - Mehmet
So still, there are like two, three tools I want to evaluate. It was late at night and I stopped after some time because I figured out that I've been sitting for two hours evaluating these tools and you are a marketer yourself and I learned this. It's very easy to be hooked up because when you give free, they give you 100 credits or, for example, 10 fries, so you say OK and no credit card required. So of course, yeah, why not? Let's try it. So yeah, you wanted to say something, ali. 

0:41:47 - Ali
I think, with the advancement of these tools and they're advancing so fast, as you mentioned I think the real threat here is going to be large corporations that think that they have software modes today. And the reason I say that is right now, yes, you could build an alpha version of an application, but I think two years from now if you said, hey, build me a CRM that's just like HubSpot, you can build a CRM just like HubSpot. So the cost of implementing expensive software starts to go down significantly for every user because now you could just build your own version of it. Build me my own version of Salesforce, build me my own version of Worker right, and I think that's the real disruption that we're going to start to see in the high tech sector. 

0:42:36 - Mehmet
I'm very transparent, guy. Ali, this is now one of my pillars, that sometimes when I talk to someone I say, hey, don't get me wrong, but what if you can build your own CRM inventory management system and you can customize it the way you want? Why? Because, simply, you can do that with AI and it's kind of a productized service that I'm trying to build myself. Yeah, and it resonated with a lot of people, so I have the percent I agree with you. So I got to make sure that the link is there. And is there anything else, ali? You wished I had asked you. 

0:43:16 - Ali
Ok great question. But, anything you want to add. Yeah, no, your question, your line of questioning, was fantastic. You asked everything from AI agents to the future of work to prompt drifting. I think that was very conclusive. Thank you so much for allowing me to be here and to speak to you. 

0:43:52 - Mehmet
Yeah, so I keep this window. I ask this question not to put someone in a hard situation. I ask this question just in case I missed something during my preparation, because I put in my mind this is a sequence I will follow for the questions and maybe some stage I will forget. So, yeah, thank you very much, Ali, and thank you for staying late this time. It's really late for you and this for the audience. 

Guys, I am liking the reviews that you are writing. Keep them coming and tell me what you want to hear about more. Like you, recently, you said you want to hear more about AI, because I was thinking that I covered a lot of AI. So keep your feedbacks coming. What are the topics you want me to cover? I'll try to find like a team guest. Like I came across with Ali, I was following him and then we get in touch and here you go. He was a guest today and if you want to be a guest as well, don't hesitate to reach out to me. We'll make the arrangement, no problem at all and, as I say always, thank you very much for tuning in. We'll see you again in the next episode. Bye, bye. 

Transcribed by https://podium.page