With the emergence of generative AI, we can now ask AI to do anything for us. But the way we ask and the way we instruct it is fast becoming an essential skill. With so many different tools available in the market, using the correct prompt efficiently becomes very important since asking AI costs money.

In this week's talk, Amit and Rinat talk about Prompt Engineering, what it is, why it is relevant, how people are using it and a lot more!

Hi, everyone. Welcome again to Tech Talk podcast where Amit and I talk about various technology related topics. And today's topic is a very interesting one, and a very upcoming and very, very, very interesting topic with a lot of questions that we've got from our viewers and listeners. It's prompt engineering. This is a very new topic, and especially it came about with the advent of chat GPT and other AI tools that are becoming very popular in the market today.

Thanks for the introduction. I think people are really excited with the current AI tools that are available in the market, especially with what you can do with them. Chat GPT has been the most popular one. But before that, there was Dali mid journey, and many of the tools that actually became very, very popular and they actually gave a whole new world. I mean, they gave a whole new perspective. To creators and influencers in terms of what they can achieve with an AI tool. So initially, people thought that okay, with the AI tool, it would be very hard to replace the creative people. But it seems like with a two in the first thing that it's replacing is the creative people. But it's not replacing as in replacing, but it is maybe helping the creatives to think more creatively. So, prompt engineering is just a way to write something into the tool, the AI tool and get an output out of it. So, suppose you say give me a photo of potatoes. Then it will regenerate a photo of potatoes it will generate a photo and then you say generate a photo a photo or with generate a photo containing four potatoes so it can it can create a photo with potatoes and tomatoes and some salad, maybe human etc. So, two different problems, two different images. And that's what prompt engineering is all about. So based on the input, how can we engineer the output?

Yeah, absolutely. I'm probably doing it's such a interesting because you've mentioned the part that I actually wanted to say is that you know a lot of people you know, since you know the advent of automation and AI, everyone is worried that would AI or automation take away people's jobs now? Absolutely not, as we've been saying it in many of our episodes, but this is actually a very perfect example how AI could create new jobs now, because not only is it augmenting the human performance, but it is prompt engineering. Is you know, it couldn't be classed as a job, which is the professional AI whisperer, if you'd like to call it that is by understanding how artificial intelligence work in the back end and then leveraging that understanding in the way you talk to it? So not everyone can talk to an AI in a way that gives you the exactly the right output that you're looking for. You can say something to the same AI interface and it would come back with a different answer than what I would say. And now the engineering part is knowing how the AI works, what kind of information or training that data it has been trained on, what to expect from it, and how to manipulate these knowledges in a way that it would come up with exactly the information or the image or the whatever the output is that you're looking for. You can even also think about engineering, some sort of creative sort of open ended, sort of you know, decision making part for the AI so you keep leave some things open as you said, you know, in your example like a potato. Now, you could prompt you know, prompt engineering. Let's go back to prompt engineering the town you know, how you were prompting you know how you're engineering the prompt you give to the AI, that's where the term prompt engineering comes from. Now, if you wanted to design a prompt to get a picture of a potato, you could actually specify the type of potato. You could say, give me a picture of a Maris. Piper potato rather than a, for example, another type of potato, he could say, generate the picture with two lights on the two sides. So, it would generate that kind of photo and if you had some more expertise in photography, you could specify the focal length, you could specify the aperture in your prompt, so the AI would take those prompts, take those directions into account while creating that image. Now, it's not just about photography, knowledge, but it's also about the subject knowledge you know off which your head getting the output so if we go back to potato if you know the types of potato there are the usability of potato there are and you know how it behaves in certain scenarios like in a warm condition, you know, you might want to see the skin of the potato turn different if it's if it's above a certain temperature etc. And then there is the engineering of knowing how that output is going to be used. So, so the customer you know what your customer wants, or actual subject matter expertise, the knowledge about potato and the medium. The knowledge about the medium that the photo is being generated, if it's going to be an actual real-life photo taken, you know as if it's taken by a camera, or it could be drawn by as a cartoon. So that medium so you need all three types of these knowledge to engineer the exact right kind of prompt which will give you the perfect approach that you're looking for. And the same goes with text-based AI tools as well like chat GPT, etc. So that's very interesting on how, you know, you could actually think about strategically so you can leverage all of this knowledge to engineer your product and that's all product engineering is about.

I think Rinat mentioned a lot of a lot about the process of generating a product, but I think so. So, let's, let's classify, I mean, the first initial AI based tool that became open to the public was a text-based text to text, you input a text and you get output as a text. So that was GPT one GPT two GPT Three, chat GPT is basically text to text again, but instead of inputting a text you input statement or a question and the chat GPT would come up with a response. And then you ask another question and will it will take into account the response it has given previously so suppose you say, Okay, can you describe Can you describe a aero plane will describe an aero plane? Can you tell me about its features so it will describe the features of the aero plane that is it has just described so sometimes, so that's why it's it remembers what you have asked previously? So that's one thing and then so that's text to text. And then text to image is what Rinat mentioned was Dali E, mid journey, stability, division, and many other tools. And that's basically you give an input as a text and the output is an image, but the image could be in the form of a painting it could be the form of a photo, it could be the form of special effects, like highly specialized, special effects, image, graphical image, etc. So, it all depends on the output that you want. Now the engineering part, so as we are repeating ourselves here, but the engineering part is basically describing the input and the input will come from experience as well as practice. So, two things. When we talk about experience, the experience is coming from your own experience as a creative person. So suppose you have looked into in photography you know how to describe objects, you know, how to take shots, you know, like okay if I want the image of Amit Sarkar, from the from below, then you know what kind of shot he will get, or you say I want a drone shot of Amit Sarkar on Planet mars, so something like that. So, you can describe based on your experience, because you know how shorts work, because you have been a photographer, but that will not come to someone who has never done photography. So that person will have to then train themselves and that's where the engineering part comes. So, it has to be served through some kind of training or practice. So of course, there are no trainers right now, who are maybe we can become trainers. What sorry, not, but I think the aspect that what we are focused on is practice. So, you try to give it an input as Rinat mentioned trying to figure out what output is coming and you take notes and every time you give a input, how the output changes, and based on that you then see okay, how it's generating and then you can become specialists. So, suppose we say text to image you can become specialists when designing logos, designing book covers, designing album arts, album covers, designing thumbnails, etc. So, you can you can speak an expertise, and you can choose that topic, and then keep practicing in those areas. And then wherever someone comes up with something, so you want to prompt engineer you advertise yourself as that specializing in generating brand images or logos, etc., or book covers or album arts, and that's how you promote yourself. Suppose you are a text to text. So, you will say that, okay, I have a summary and I want to create a marketing ad out of it or a story out of it. This is the setup. Can you generate content, and you will use the tool instead of coming in by yourself. Now, the difference between the previous thing, the era before prompt engineering and right now is that previously, only the creatives could participate in such kind of activities like designing a brand, designing a logo, etc. I'm not a creative person, I can't design a logo. But if I can describe what a logo should look like, I don't have to be a painter or artists or anyone. I can now just type something and the AI will take care of it and that's the beauty of the era of product engineering.

Yeah, absolutely. And, you know, just going a little bit further on the example you gave in terms of photography, someone who would need to have photography understanding to a degree, but then I would also argue that it's not just photography. You could also tell them to paint a picture in the style of Van Gogh or

exactly that is an example. Yeah,

Yeah, yeah. So, in many ways, prompt engineer is someone who is Jack of a lot of trades. Rather than a master of one. Yes. So, they could, you know, they could potentially be tasked with creating a creating an image in a cartoon type setting. Right. So for that, you need to know that what should be the paint you know, the paintbrush size and colors and that kind of things, you know, for someone who sort of draws cartoon, that knowledge as well as another task, maybe just the next task could be about you know, actual oil painting or it could be about actual photography. So, you're not, it's not about knowing or learning about all these different kinds of media medium knowledge, like you know, the medium of art. And it's also not about knowing all different kinds of products of which the picture you're trying to generate. But it's, it's actually putting in some critical thinking, understand what the user might be enticed with the most and then sort of tailor your prompt in that way. So, you get the desired output. So, it is a very sort of crucial and well, you know, it's quite like a I don't know what you can call it thoughtful sort of job to do. And it could be a very like, you might think that it's, it's, you know, you it's not something that you do you know, if this was your professional, you don't do it like every five minutes, you just keep prompting the AI. But you know, you think about it, you know, with some time you research and then you come up with the perfect prompt that would potentially generate I think that's how you're going to be measured in whether you're a good prompt engineer or a beginner, is that a beginner might need to prompt a few times to get the desired output but the inexperienced or a very expert one would maybe need a lot less prompt and this might actually be a deciding factor because at the moment Chat GPT, Midjourney ,dally E these are free tools for everyone to use. But if you if you I mean, as of today, Chat GPT is at capacity so you can't actually use it you can only use it if you're lucky to be honest. So, you know, prompting and for the AI to generate output takes a lot of processing power. And you know, the people who can get the desired output with less prompts are going to be financially beneficial to the AI company or to the company that they're working for, because then otherwise going to be charged for each of the prompts. So yeah, there is definitely a financial involvement as well and the more you can save money to a company to get the desired output, you're going to be valued that way. So absolutely prompt engineering. I think it could become a profession. Maybe not the way we're seeing it now. It will evolve as the AI tool evolves. I think, you know, I mean we were talking about it earlier before the topic and you might you might actually have an opposing view on that. What's your view whether Trump's engineering is a legitimate profession, could it be or

Yeah, so I think I like to go back to what you said in terms of trading and experience. I think that when I was talking about the photography example, I meant a person who's a photographer who knows the skill set of a photographer, and a person who doesn't know photography and then trying to act like a photographer and trying to use that skill set to generate an image. So, the difference between the two ways one is experience and one is no experience. And I was trying to highlight that aspect. Now of course, the person who doesn't have any experience in photography, they don't need to learn photography. They can talk about paintings and different styles on say, click a photo photograph in terms of in the style of this person, and they can imitate that. But the whole idea was that there is a creative person using an AI tool and a non-creative person using an AI tool, and there has to be a differentiation. So, a creative person, as you said, would take less prompts to generate desired output. And a non-creative person without practice would take for larger prompts. But if they have practice over a certain area genre or an area where they would like to specialize, then maybe yes, they can become specialists. When it comes to the opposing view that I had was actually I was listening to Sam Altman, who's the co-founder for open AI. And he was saying that prompt engineering may not be the hype that we are trying to give it in the current. I mean times, because he's saying that you, prompt engineering is not about coming up with that magic word. That will just change the output. It's about a mixture of words that you write based on your experience. So good prompt engineer is someone who's an expert. And a bad prompt engineer is someone who's a novice, and it's the same with any skill, acquiring any skills in life. You could be a very good batsman, or you could be a very average batsman, and the only difference is someone who's got a lot of experience gone a lot of training, and someone who hasn't practiced enough hasn't got a formal training, and that's why they lack so they are both using the same tool which is the bat and the ball, but they're getting desired different outputs. And that will happen with prompt engineering as well. So, you will get varied outputs based on your experience. With how you describe something. So, it's not going to be that, okay, that one magic word, and it will give you that amazing output. It won't be that it has to come through practice and through your own vision and articulation. So as a result, I think you mentioned it correctly. That it would require a lot of research into what the what we actually want or the companies actually want. And based on that research, then you would generate the output currently it's free, but when it gets becomes paid, which it will be, then there is a cost associated with every prompt that you are trying to get an output for. And that is going to be quite expensive if you're not very careful. And one of the good examples, which I've published in my newsletter as well is the AI cover for Cosmo. So that was actually generated using dolly and they hired a influencer on YouTube. And she basically used dolly to generate the cover for Cosmo Cosmopolitan magazine. And it was the first AI generated cover and it was really amazing because she describes the process of how she generated that image. Maybe I'll share that video as part of the description of this video on the podcast. And you guys can have a look. But I think it just shows that okay, what kind of output she was getting and how and were the creatives at cosmopolitan happy with the output or not. And finally, what was output and finally what was put on the cover and what was the prompt for getting that output? It's kind of interesting because it shows the whole journey.

Yes, yes, absolutely. Actually, I mean, I don't fully oppose with what you said in terms of the you know, the requirement for to become a prompt engineer. And yeah, absolutely. You need practice, practice, practice, you know, without the experience and the expertise on the actual subject matter or the medium that you're going to portray. You're not going to go very far. But what I want to sort of add on to it if you'd like rather than opposing is that you can do a lot of practice, you know, by yourself and, you know, give a lot of problems and not really understand where the actual differentiating factor is. So, you need some kind of direction. In terms of actual you know, the knowledge that we've actually been learning you know, people are saying that with the advent of AI, you know, you don't need any you don't need to go to school anymore, because all of the information is right there. But actually, if you don't have that kind of knowledge of, of, of a specialized subject, you can't even know how to talk to AI to generate something that is valuable for a business or a customer, etc., etc. So, I think those creative jobs, those creative sort of expertise are very much needed if even more so now in a greater nuance, a greater detail. So, you can actually know how to manipulate it enough. I mean, you need to know, to a degree, I mean, you know, maybe you're a carpenter and you know, you know, how to you know, build the basic furniture, etc., etc. But now, I mean, there are actually patented designs of how to join two pieces of wood and you know, they've I mean, I've actually seen someone paint on something last year, a newer design of joining two pieces of wood without any glue or without any other nails or anything and but it will just get stuck, you know? We just with the design, so someone has to be that level of expert in carpentry to be able to manipulate exactly what they want out of an AI tool. You know, an AI tool might be just geared for you know, carpenter related knowledge and it will output new designs but if you don't know all the existing designs which are paid and then not participate in the traditional etc., etc. How is you prompted, so it would give you a design for a new furniture based on the material you have etcetera, etcetera. So, nowadays it's become even ever more important to become a next level of expert on those disciplines. It could be carpentry, it could be medicine, it could be you know, becoming a lawyer. You need to be able to manipulate AI to a degree that you get the desired outcome. So for that you need to be even more I mean before, you could have probably gotten away with mediocre Korea, but you know, you had food on the table and paid all the bills, but now you want to be you know, a lot more of a subject matter expert on each of these disciplines. For you to get the proper value from Ai. So yeah, prompt engineering is really good on how to communicate with the AI but then you might want to have an actual BA or business analyst or a subject matter expert who will give you that field of expertise which will enable you to give the right prompt and this could actually become a team task rather than you know, obviously that you could I mean, even if it became become costly, it may not be too costly, so it could potentially easily give 510 20 prompts every hour and you'll still be okay but ultimately everything that we do in business or in life is actually time constrained. So, you want to get your output faster because otherwise your competitor will get it before you and then reach your customers before you so there's always gonna be that constraint that you want to get the right optimized result as quick as possible. So, you can do something with that output. And for that, you need to know how to talk to the AI.

Very, very, very important. Yeah. How to Talk to an AI because I think you're right i If, if, as a creative person, you know what to look for. What you don't know you don't know. So, like I have a one-year old son. He knows that okay. If he sees a step, he has to climb. If he sees the corner of a bed, he knows he will fall down. So, he knows this much. But suppose he hasn't seen the edge of a bad or he doesn't know what falling looks like he doesn't know. So, there are things we know and there are things we don't know. And things we don't know we don't know. So, no matter how expert you think you are, there will always be things that you don't know. And that's where the expertise comes in. Because these experts, they would have vast amounts of knowledge about every single aspect of their area of expertise. And that's where it will make a differentiation. Because think of it like this. These AI tools are given to people who are experts, the experts will become even more experts because earlier they had to do the research work, they had to do a lot of like guesswork, etc. But now they know through their years of experience like what they need, so they'll just write three or four prompts, and they'll get an a rough idea. And they'll write the exact prompt which will give them the exact output so with three or four prompts, they know what they've got some ideas and they will with the final product, they will get the result that they want. Someone who's not an expert, they will firstly try to figure out what they what they want, because they don't know what they want. So, they try to figure out what they want. So, when I was given access to Dolly, I mean not given when I got access to Dolly. I tried to write something, but I didn't know what to write. Because I don't have a use case right. So, I wrote Okay, Scooby Doo we Scooby Doo, launching on a rocket ship. And that's the prompt. And then I tried to modify that prompt okay Scooby Doo launching in SpaceX rocket looking like this ultra-real this that and I was just trying to see what the images come up look like. So that basically, but if if suppose you give it to an artist, the artist would say draw the image, draw the cartoon character Scooby Doo, in a very like a stylish way or something etc., etc. But the idea is that they would know what to say, rather than be just coming up randomly with some idea and then trying to iterate on that rather than coming up with something concrete. And that's where the difference lies. And I mean, I think you've mentioned a very good point, we might have people who are experts working with engineers, who will do the hard work of typing in the prompt. But let's see, let's see how this picks up because property engineering, whatever we might say, it's doing one thing and that is putting the power of AI into people who are not very creative. So, this level of the field because now you have the creative aspect of, I mean, have a tool in the hands of user’s generic users who have never done any creative work. And imagine if you give that power to them. It's like giving a smartphone to person in a village. They don't know what to do with it. But then they eventually figure out and they then become very productive. So eventually, people will become very productive, but initially they will be very, like, not that great, but it basically means that they will there is now a level playing field. Of course, the experts will always be experts. So even with any AI tool, the expert will always come up with a better result and a non-expert will come up with a less better result. But now everyone can get results. And that is what prompt engineering is. I mean not promising anything but these AI tools, bringing.

That's exactly I mean that's exactly what I was gonna say that once again, you know, technology is actually eliminating the most time-consuming parts. If you think about if we go back to, you know, the image generating AI is again, for example, you know, back in, like 100 200 years ago we still had images being generated by painters and you know, the creative people, but then say for example, the A King is ordering a like an artist to you know, draw this this this scenario, or whatever. And then they are taking two, three months or so, and then bring it back and then the king might say, oh, change it in this way or that way, and then they take another two months and that we've come a lot further, you know, every I say a decade, maybe we've kind of gone to the next step. Now the artists could do it in hours or minutes in digitally, etc., any changes, etc. But now, then that wasn't fast enough. Now with AI. It doesn't have to be a king or a business owner or someone with a big project or funding etc. Anyone can ask a painter and artist which is itself the AI tool so they can try 20 different things within 20 minutes, and then they can go away with the one that they like. So ultimately, the part that is being you no more efficient is that to have the boring process of painting it out or drawing it out. But the actual creativity still is necessary. It's just taken a different form, which is, in a way how to tell the AI your creative idea, but you still need the creative idea for a product to be successful in the current world. So, the people who were creative, they are still valuable and they're just having to sort of reform their expression of their creativity in a way that the tool understands, which in a sense is prompt engineering. And another part I wanted to add is that we talked about something the AI tool to give a completely finished product as an output. We may not actually want that we or we might actually want just making parts of it because I've actually seen a video where someone is creating a donut website, you know for their ecommerce business, etc. And they don't they want the banner they want you know the various design backgrounds etc. So, there is a few images that are needed. So you know, even just for the background or the main landing page, you know, of the website they have instructed to create a few donut based background which was created and they have to then make the decision on which one are they going to take back into Photoshop or Illustrator and then cut out the parts that are not needed an ad from which they might actually combine four different outputs of the AI you know of the AI tool. And the four different output could be the four different corners of a picture one corner could have done it and one the other corner could have you know a donut shop in the background which is completely different to prompts and the actual the person who puts them all together is that your creative person who is who is the only person who can do it. So yeah, there is absolutely more and more actually opportunity to have newer jobs and newer skills which are actually interesting and not having to do the boring part of actually drawing it out. Because the artists who do a phenomenal, timeless painting, they had that vision in their mind before they started painting and then they made it into a reality. But what is what is powerful and what is valuable about that painting is that vision because anyone could have used paints and colors and all painting etc. and created anything. But it's the vision that the artists had in their mind through their own creativity is what made it made an in genuine sort of in appealing product. And if that value remains the same, even with AI because you know, it's the prompter who has the vision and then the sort of orders AI to make it into a reality which is actually the boring part. So absolutely prompt engineering is, I think, a very interesting role. I don't know if it would become a profession, but it would certainly be a very interesting role to go forward.

Definitely and I really liked the way that you described. The part about the artist and the manual effort of drawing something. It's basically the prompt engineers are basically these AI tools are basically automating the whole process of painting. It's like, dishwashers have automated the process of washing dishes. So similarly, AI tools are now automating the process of painting. So, painters no longer need to paint they just need to come up with the idea and the tool will do the rest. And the same goes with 3d sculpture. So now there are tools in the market that can draw 3d models. And then you can use those 3d and you can sculpt those 3d models. So, like you have a like a big giant cube, and then you sculpt around it like you would sculpt a piece of wood. And then you have a sculpture, and I saw a video about this happening. And maybe I'll post that as well. But I think I finally get the vision by just by talking in this podcast. And I think it's this vision of automating the process of creativity itself. Like focus on the creation of the idea focus on the idea itself. Rather than the boring part of creating. Creation is more important than the act of creating. So sorry, not the creation, the idea of the creation is more important than the act of creating it. So, if you have the idea, then you can create whatever you want. And this is basically what we do in automation. So any boring repetitive work we automate. So, here's the boring every bit of work can automate it by the tool. And now you can just focus on being creative. So yeah, it's a beautiful thought. And I think it makes a lot of sense in a world where a lot of people are afraid and I think people should not a lot of jobs lost when the industrial revolution happened. But there were a lot more jobs created. So, it's not like once the industrial revolution happened. The economy went down and people lost jobs, people lost income, etc. No, no the the industrial revolution would happen if we had we added a lot more jobs the economy grew and do the same thing with AI is that AI will take away the mediocre jobs, the boring ones, the ones that should not be done by humans. And it will be it will make new jobs which are more exciting for humans which they want to do, which they are good at, because they have to use their cognitive skills. So yeah, I think it's a very exciting time to be alive in.

Yes, yes, absolutely. I couldn't agree with you more. And I know in our example too focused on generating incidents, but the same thing goes on text base-based AI tools as well. Yeah, I mean, you need to know the subject matter very well to prompt the AI tool to, to sort of output a paragraph or an essay or, or whatever you need to be you know, understand lawyer speak, to be able to tell it that, you know, write me a contract for this. If you just say write me a contract or a tenancy agreement, then it will write you a very generic one, which you can probably just Google and get a template. But if you want us if you are a lawyer, you will know write me a contract of a tenancy agreement of this country in this setting with force measure clause with this other clause and mentioned this other legislation into it, then you actually come up with a actually bespoke tailor made contract, which is applicable to your current scenario. So yeah, I mean, the knowledge that you needed is still very much needed. And you can now actually do more with it because now you can try different things you know, you can you know, Nuance your prompts. Based on the output, you get to even improve more and more. And I've heard people talking that now if students can just, you know, get the chat GPT or any other text-based tool to write essays and paragraph and assignments from them. How would they understand or learn critical thinking? But if you really think about it, to give a prompt, I mean, teachers are the assessors of those assignments has to be more. More sort of careful on how to mark them because everyone, all the students can go to an AI tool and get output but some students will still prompt better than the other students and get a better approach. So, they might have to redo their assignment. to tailor make this these kinds of scenarios and the student who is prompting, based on you know, prompting for and getting a good output. They're actually thinking, how is this going to be used by a potential customer if this was in a real world? How is the business going to get, you know, actual, you know, maybe financial benefit out of it? So, they're actually doing a lot more critical thinking in prompting, well, then just, you know, blindly writing whatever comes to mind in an essay. So, there is a lot more critical thinking and contextualizing that goes on in prompting and getting a desired output. So I don't think it's gonna sort of you know, break the you know, the education system rather teachers or the assessors or the planners has to be more you know, has to demand more in terms of from the essay and has what they would want to see that the essay is consisting of details, which only a good prompter can generate. So, the assessing criteria, and the questioning criteria could be tailor made to make room for these new technologies.

Definitely. And I think that's a very good example about essay school essays because a lot of students have already started using that we've already read it in the news. So, if you look at the text to text generation tool, it can be used for a lot of other scenarios. And there is examples like you can generate code from it, you can generate a lot, a lot of other things. And in each scenario, the whole idea is that I mean, as Rinat mentioned, the contextualization the thinking that went behind the prompt, because as I mentioned, you get access to a tool, but if you don't know what you have to ask, or it has to, if you can't describe what you want, then the tool will not do it for you. Right. So, it's the same way I got access to a tool, but I didn't know what to do. So, I just typed in the words. But if a person who knows exactly what they want, types exactly what they're thinking, then they will get their desired output. And that's the difference. And that's what the dating part is about. So, everyone has access to the prompt. It's just how you engineer it to get the desired result. So, well. I think we've talked a lot about this topic. So, thank you so much for listening. Rinat, Do you want to add anything else in the end?

No, no, I was just gonna say that the same thing, Yeah. Our audience Yes. do reach out with your thoughts. I think we we've kind of discussed and I very thorough, thoroughly enjoyed this conversation with you Amit. And hopefully our audience did too. Please reach out with any comments, feedback you guys have or if you'd like to join us as a guest. We're more than welcome. So, yeah, hopefully we'll see you guys again next week in next week's episode. Until then, thank you very much and bye

Take care everyone bye

Take care