UX for AI

EP. 96 - Stop Ideating, Start Creating: A Hands-On AI Workshop

• Bonanza Studios

Share episode

Send us a text

 This session dives deep into using generative AI tools for video, design, and app development—without code. Learn how mastering prompts can unlock professional-grade outputs, from animated marketing videos to functional web apps. It’s a hands-on walkthrough of creativity and productivity in the AI era. 

Interested in joining the podcast? DM Behrad on LinkedIn:
https://www.linkedin.com/in/behradmirafshar/

This podcast is made by Bonanza Studios, Germany’s Premier Digital Design Studio:
https://www.bonanza-studios.com/

1
00:00:00,000 --> 00:00:16,000
Welcome to UX for AI. Hello everyone. Asando, how's it going? You alright? Yes, I'm right. I hope you too. Yeah, not bad. Okay. So you hear me well and clear, I hope. Yes.

2
00:00:16,000 --> 00:00:35,000
Alright, let's just wait for a couple of more minutes and we can kick off, depending on how many people join. Sawati, I hope I pronounced your name correctly. Yeah, that's right. Hello, how are you doing? Not bad. Thank you very much for asking. Are you good too? Yes, I'm good too. Thank you.

3
00:00:35,000 --> 00:00:56,000
Alright, today I'll, let me see what I can show. I wanted to show you a couple of cool apps to bump into. That's the first thing that bunch of white marketing stuff I'm doing. I think that would be interesting for you in case you doing certain things. Yeah.

4
00:00:56,000 --> 00:01:18,000
Yeah. Alright. I think it's a good time to start. So, I know Sandor, Julia, Andrea, I hope you guys are okay. Is this your first time? How many second timers? Sandor is second. Sawati, have you been with me before? I think your name is Femina? Yeah. Did you ask me? No, I clearly remember you, Sawati.

5
00:01:18,000 --> 00:01:33,000
That's what I wanted. Andrea, Julia, pleasure as well. Yeah, first time for me. Okay. We can't hear you. At least I can't hear you. Yeah, I should be able to, my microphone is down. Julia, how did you hear about this event? I think it was through LinkedIn.

6
00:01:33,000 --> 00:01:58,000
And I've always followed Bonanza for a while. Yeah, really pleased to be here and listen and provide feedback if needed. And what do you do? I'm a senior UX researcher here in Berlin. I work for Doodle scheduling. Great. Great to hear from you. Thanks for being here. So, I'll speak to the introduction. Who am I?

7
00:01:58,000 --> 00:02:14,000
Maybe just my LinkedIn in case you haven't added me on LinkedIn, it would be nice to get stay in touch. This isn't merely a work session, so I'm just going to create a bunch of stuff together. Working through a bunch of things I bump into. That's it pretty much.

8
00:02:14,000 --> 00:02:33,000
In case you would like to add me on LinkedIn, I run a product studio. I do a bunch of stuff all the time. That's what we do. That's what we get paid for. So we're doing new things. I'm going to walk you through, I don't know if you're following my LinkedIn, but most of my posts recently are GenAI.

9
00:02:33,000 --> 00:02:59,000
This is GenAI. Actually, this is well produced. I'm a video producer. I produce it. I'm GenAI. The entire video is GenAI. Can you believe this? There is no shock. It's all GenAI. I can show you what I've done. To me, that's groundbreaking. I would have dreamed of having something like this two years ago.

10
00:02:59,000 --> 00:03:13,000
Now it wasn't easy to make, but now I know how I make it. Next time it will be very fast. Also the video here is GenAI. The rest is in Figma, but the video is GenAI. I can walk you through that as well.

11
00:03:13,000 --> 00:03:32,000
I'm just in the beginning of my reaction video. This is GenAI too. This is more complicated than this. Both are complicated. Also GenAI. That one too. I can walk you through some of the stuff. I think it would be very useful for whatever you are doing right now.

12
00:03:32,000 --> 00:03:48,000
Whatever that you're creating. I think the chargivity for image generation has been imperative to stay the least. I've used quite a lot of, I've used mid-journey, used different tools, but I don't know what is it about it that is so far gone.

13
00:03:48,000 --> 00:04:04,000
It gets what you're trying to do and do it. I've walked you through a couple of things like this, but I don't know if you have heard about ICON, that made a buzz. Arguably the most groundbreaking app I've bumped into in the past week.

14
00:04:04,000 --> 00:04:20,000
It's basically, if it replaced your entire performance marketing department the way it claims. I haven't used this disclaimer, we don't have these kinds of use cases. But what it does is research your competitors, the trends around your product.

15
00:04:20,000 --> 00:04:35,000
It does that. Again, I haven't used it. It's the claim that they make in this video. I'll give you the look later. But does the research and competitive analysis, research ideas, create many ads at once.

16
00:04:35,000 --> 00:04:50,000
I blocked my YouTube when I worked. That's why I cannot show it to you guys. It's happening what it does for you. Create many ads from them. See which ads are performing the best.

17
00:04:50,000 --> 00:05:11,000
You can create more ads out of the winning ads. It does it with $39 per month. It might be like you made $100. I don't know about the pricing and the scope of your business, but that's what it does.

18
00:05:11,000 --> 00:05:27,000
If it does what it does. That's basically what a lot of marketing agents do. And then what it does also, it's not only image, it creates video. Video products on it. These are the different ads that they produce.

19
00:05:27,000 --> 00:05:37,000
You don't even need to tell them what your business is about. You just scrape your website and learn about your website on the go.

20
00:05:37,000 --> 00:05:55,000
So that's very fascinating. In a department, in an organization that their job is just to maintain a competitive workflow. Some level of creativity could be potentially be run partially by. Yeah, fascinating. What do you guys think about this?

21
00:05:55,000 --> 00:06:01,000
It's the first time for me that I hear about it. Interesting.

22
00:06:01,000 --> 00:06:09,000
And you have marketing department that do the right, that they do this kind of stuff.

23
00:06:09,000 --> 00:06:25,000
Yes, we have a performance marketing department with a content writer, a person that does SEO and I think also graphic designer. So that would be super interesting I think for them to try it out.

24
00:06:25,000 --> 00:06:42,000
Yeah, I think so. It's crazy. I just, the stuff that some, some time ago was, you needed so many people now it can be done with a bunch of integration and AI on top of it.

25
00:06:42,000 --> 00:07:04,000
I bumped into another thing as well. And then, so this platform are not unique, has been making a lot of bugs in the world. I can pull up one example of Flora, which I find really interesting. Just a second bear with me. This is going to be people of an AI, very good in my opinion, in terms of this one. So see what it does.

26
00:07:04,000 --> 00:07:17,000
Another platform. Do you see what happened? That would have been, that's why you hire a graphic design agency for. That's just like half a million dollars per year, graphic design agency work.

27
00:07:17,000 --> 00:07:30,000
So, then, so what happens here is, so Flora is actually a really cool app. I used it in the beginning, I should have not used it the first, and the beta version of it because we were useless, but this is getting interesting.

28
00:07:30,000 --> 00:07:46,000
So you have this person and you have this person. First of all, with one prompt, you said like, make the hair gray. So you made it, you see what's happening. And then with another prompt, you say like, make the, make the clothes from red to blue.

29
00:07:46,000 --> 00:08:01,000
Okay, and then now you combining these two. So, the output will be. So you start from here and you end up here. And then now you give the character back, easy.

30
00:08:01,000 --> 00:08:17,000
See where it takes it. Now, it's, there are, so this, now they are using a lot of different AR generation, video generation models. It wasn't possible in the beta version I said, so that's why I found it useless, but they're using clean 2.0, which is good.

31
00:08:17,000 --> 00:08:33,000
I haven't used it, my cleaning is good, I don't need it. Interesting. So now you said like, okay, create four different ads for me, from my social media. Posing for camera, walking and posing.

32
00:08:33,000 --> 00:08:48,000
So there you go. That's right there. What you hired marketing creative agency for. That's done. So, a lot of agencies, they need to question what they are doing and how they should survive further.

33
00:08:48,000 --> 00:08:57,000
I mean, potentially they can offer this to their clients, but if the client is smart enough, they can do it themselves.

34
00:08:57,000 --> 00:09:08,000
But this is very powerful stuff. Really powerful stuff. Especially that design, it doesn't require creativity, like, you know, just like you're attacking and removing certain object.

35
00:09:08,000 --> 00:09:21,000
I was. I felt like it's game over for creativity when cleaning AI managed to make this video for me.

36
00:09:21,000 --> 00:09:32,000
And I show you my process, what happened. I was, I just texted my wife, I was like, this is crazy. This is scary level creativity.

37
00:09:32,000 --> 00:09:40,000
Yeah. And I'm walking through this process. How about that? Would you like to see the process of me making this?

38
00:09:40,000 --> 00:09:42,000
Now we're interesting, right?

39
00:09:42,000 --> 00:09:57,000
Is my audio better? Yes. Not this better. Okay. So let's see what I wanted to do. Working with the video AI LLMs is much better, is different than working with chat to BT.

40
00:09:57,000 --> 00:10:12,000
And what can walk you through a bunch of this stuff? Okay. Okay. So this was me trying to see how can I make, cause we have this weekly session. I need to make it exciting for people to join. So how can I do something different than before?

41
00:10:12,000 --> 00:10:30,000
Then before AI, we would create one template and each weekend would change the date and maybe some image changes, like, you know, and then done. Right? But now with AI, I can potentially do any, any changes here.

42
00:10:30,000 --> 00:10:46,000
For example, this was I for this. So this was a process for this poster. I AI the guests, you know, face my, my face. Then I asked chat GBT to create a graphic for me.

43
00:10:46,000 --> 00:11:02,000
And then I paste our images here. So everything here is AI generated. I just like did some masking. That's what I did for this week. I had this. So the video is also generally AI. Right. But I wasn't entirely happy about it.

44
00:11:02,000 --> 00:11:25,000
I was like, okay, that's, that's easy. Right. So what can I do more? Let me see which platform I use for this one. It was a Suzanne. It's a Cling. So both of this, I been having great success creating this stuff. Cling has been all the bars in the past weeks, but these bugs are very, you know, it's model by model.

45
00:11:25,000 --> 00:11:46,000
As soon as the new model comes up, people say, this is the latest and greatest and this is it, but this comes and goes. Sorry. I haven't had lunch. So this is what I've done. Right. So read the prompt. I think there is prompting. These things is the key to get it right.

46
00:11:46,000 --> 00:12:08,000
I think I have, I can show you Cling assets. So I wanted to, this was my first prompt. Like let me show you refresh your memory. But what we arrived at, I created these and then I just move it to Figma, Figma prototyping, a smart animate.

47
00:12:08,000 --> 00:12:21,000
I cannot go through it this session because it's a bit nerdy, but it's simple. You can just basically, you can figure it out. That's not the point of today, but just I put it here, animated it, and I got this.

48
00:12:21,000 --> 00:12:38,000
But as you can see, the rest is the same and I wanted more, right? So I had this basically. So then I went to Cling AI and I wanted to basically have some, you know, futuristic out of this planet plans coming in.

49
00:12:38,000 --> 00:12:58,000
And I gave this prompt first time, it gave me this. This is the second attempt. So it's still not there. Like, this is, this is what I wanted. Well, I didn't want it like exactly like this, but when, when you create a video needs to basically start to end of it, needs to follow a logic and make sense, right?

50
00:12:58,000 --> 00:13:13,000
And when you look at that video, it's all over the place, right? What does this mean? This is the third one. This is really like, Oh my God, I'm not getting it. And like the whole process took me at least one or two hours and I usually work at night.

51
00:13:13,000 --> 00:13:30,000
So when around 1am, you hit something like this and output like this, it is as if the entire world is crumbling on you, right? So I look at it like, Oh my God, this is terrible. I'm not getting it right. It's just now two person, what's happening?

52
00:13:30,000 --> 00:13:48,000
Then in the fourth attempt, it got it right. So what happens between the first attempt to this attempt to be able to produce an output that I can use it in my marketing. When they say prompt engineering is going to be the future, they say it very right.

53
00:13:48,000 --> 00:14:03,000
It's so, so crucial how you get your prompt written and what you ask the AI to do. If you without the prompt, the right prompt, and I can do a surgery on each of these attempts so you can understand what's going on.

54
00:14:03,000 --> 00:14:24,000
But basically what I've done, I let this one to run to each time. I wanted to basically walk, fix the prompt, bug fixing the prompt. I let this thing goes and I read every word, every sentence, every movement I asked and just iterate, iterate, iterate, literally.

55
00:14:24,000 --> 00:14:39,000
I don't know how you guys create these outputs, but you probably either write it yourself or you ask another AI to create a prompt for you then you feed the other AI to create that thing for you. In either way, there's going to be a few things happening.

56
00:14:39,000 --> 00:15:00,000
If you write it yourself, there is a lot of ideas that you have, you think you wrote it down in the prompt, but actually stays in your brain, you didn't elaborate on it enough. That's why when you input these things, you get an output that's far from what you have in your brain.

57
00:15:00,000 --> 00:15:15,000
So a lot of translation of what goes on in your brain needs to happen and you need to be very patient with it. Certain things that you ask the AI to do is not physically possible, or the AI is still not there to make it happen.

58
00:15:15,000 --> 00:15:32,000
So you need to identify those things as well and remove it. The other factor is very important to create prompts is that do not treat AI as a stupid, do not macro manage it. Just give it a direction and let the AI figure it out.

59
00:15:32,000 --> 00:15:50,000
Because when you go to an artist, let's say a painter, you ask it to paint you, you like their work, you like their style, and then you're not going to ask them what to do every seconds, they are doing it.

60
00:15:50,000 --> 00:16:08,000
Look at AI, especially in the creative world, as another artist, respect it as an artist. So when you give it a prompt, try not to macro manage it. Especially if you want to create a scene like that one.

61
00:16:08,000 --> 00:16:23,000
You can macro manage it if you're creating a short story and there is a way of, and I can show you what I've done for the Ghibli style. Hang in there, that's actually would be interesting. That would be a good homework for you. You can create a Ghibli style animation for yourself.

62
00:16:23,000 --> 00:16:39,000
But if you want to basically start to finish and you don't care so much about what goes inside each frame, try to give the AI a very exciting, simple and physically possible direction and let it does its thing.

63
00:16:39,000 --> 00:16:47,000
That's my advice. That's like six, seven hours of trying to get this thing video right advice.

64
00:16:47,000 --> 00:16:59,000
So, and what I've done this time that I didn't do before, I would get disappointed that I would say that, okay, this doesn't work. But this time I really read every, every line of it.

65
00:16:59,000 --> 00:17:17,000
So let me just like do a quick dissect surgery of my prompt. Then I can show you how I made this Ghibli style. So I created this thing entirely with AI. So any questions so far? So let's go through each prompt, each line of the prompt.

66
00:17:17,000 --> 00:17:39,000
Standing prominently on stage is a man initially glazing through thoughtfully towards the distant horizon. So far is correct, evoking a sense of mystery and contemplation. With a gentle yet deliberate movement, he turns his neck towards the camera, so far so good, making direct, confident eye contact, capturing attention elegantly.

67
00:17:39,000 --> 00:17:58,000
That's right. As his gaze settles into a powerful yet inviting stare, he raises his hand gracefully. The good exactly what I'm asking to, executing a gesture that suggests an invitation as though welcoming the audience into a new mysterious world.

68
00:17:58,000 --> 00:18:18,000
So far so good. In synchronization with this elegant gesture, vibrant futuristic alien-like plants begin to grow and bloom rapidly around him. Their textures and forms appear detailed, realistic yet distinctly otherworldly and mysterious.

69
00:18:18,000 --> 00:18:35,000
These plants progressively surround him and create a lush, enchanting and thriving garden reminiscent of an alien ecosystem. Simultaneously the camera begins a smooth pullback motion gradually widening the perspective.

70
00:18:35,000 --> 00:18:51,000
The alien accented lightning subtly emerges gently illuminating the plants and casting dramatic highlight. Oh my God, I lost count! Accented throughout the dark green spring. Did you see what happened here? Especially this part here.

71
00:18:51,000 --> 00:19:05,000
This was the problem. Like, I did not get this part right. First of all, I didn't time in the prompt properly when the plants need to come in. Too much detail. Yeah, exactly. Too much detail.

72
00:19:05,000 --> 00:19:17,000
First of all, my timing wasn't right. Like when you're creating a video frame, like when you're in the video editor, if you have different components, you need to basically tell them where it needs to appear.

73
00:19:17,000 --> 00:19:31,000
Like I thought when I say in synchronization, it means that from the beginning, but that wasn't the case. I thought that this is going to happen right after the guy gazes at the camera.

74
00:19:31,000 --> 00:19:44,000
First problem. Too many movements are masking. First camera, the plants are coming up. This happens, this happens, this happens, then the neon accents and lighting subtly emerges gently illuminating.

75
00:19:44,000 --> 00:19:56,000
Too many actions. It's not really possible. It's physically not possible at all these things happening. It could happen if I frame it better, like say, okay, this first needs to happen, then this needs to first happen.

76
00:19:56,000 --> 00:20:05,000
But the way I prompted it, it really brings everything together in one boilerplate, so to speak. That was a big problem.

77
00:20:05,000 --> 00:20:17,000
So let's look at the prompt of the last one. First of all, what you see, it's very important. This is the prompt here, and this is the prompt here, a lot less. Right?

78
00:20:17,000 --> 00:20:26,000
The animation opens on a dimly lit stage with deep dark green background. Interesting as well. I don't know if you see this. Let me see.

79
00:20:26,000 --> 00:20:37,000
So I like this image. I like the flower and the style of it. I attached it on cling, on cling. I mean, FYI, if you want to do certain things.

80
00:20:37,000 --> 00:20:45,000
So for this one, I gave my face here and this. So when you go to creative studio in video generation.

81
00:20:45,000 --> 00:20:57,000
Here you can add elements, exactly the same thing that Florida's that I show you guys. You can say, OK, I want to mix and this image and this image. And it does it really well. This is fame.

82
00:20:57,000 --> 00:21:05,000
I hope I pronounce your name correctly. Fame or Pharma. I think it's Pharma. Anyways, this is fame.

83
00:21:05,000 --> 00:21:17,000
This is basically. So there is also two things. So you have. You have agent for exploration. You have agent for brainstorming.

84
00:21:17,000 --> 00:21:26,000
You have agent for creation of the prompt. You have like creation of the text of the prompt, and you have agent for creating the video.

85
00:21:26,000 --> 00:21:35,000
Right. You need to look at the AI as different agent that does something for you. For example, you might ask chat GBT.

86
00:21:35,000 --> 00:21:46,000
You might send it as AI, as exploratory agent, do some search for you online. I give you some insights back.

87
00:21:46,000 --> 00:22:01,000
Then you can ask the chat GBT again to synthesize this. Now you the chat GBT verse and sort of synthesizer agent understand each function that throughout your exposure to LLM.

88
00:22:01,000 --> 00:22:07,000
You are asking to wear the dress of different agents. Right.

89
00:22:07,000 --> 00:22:20,000
So when you get to the video creation here, there is no back and forth with this AI. This AI does what it needs to do. It creates video for you.

90
00:22:20,000 --> 00:22:31,000
All your exploration needs to happen with another agent. Right. Important to understand. When you're here, there is no there is the only thing that would go away is your credits and create each of these credits.

91
00:22:31,000 --> 00:22:44,000
It's usually consumed between 40, 50. Exactly. It's just one way. Right. So you really be careful. This is a very especially especially agent to create a video and images.

92
00:22:44,000 --> 00:22:55,000
Your prompt needs to be really, really, you know, spot on. Right. So your exploration and brainstorming agent is typically in the chat GBT graph or anything between. Right.

93
00:22:55,000 --> 00:23:17,000
So, back to this. So let me tell you what I've done. You saw that prompt. I made the prompt a lot simpler, allowed the agent to basically connect the dots, the remaining dots by itself, but the framework that it tried to apply its own thing needs to be clear, the framework.

94
00:23:17,000 --> 00:23:38,000
The animation opens on a dimly lit stage with a deep dark green background. The man from the provider portrait, the one that I damage or attach stands at the center. Okay. Initially glazing quietly towards a distant immediate immediately at the start, like so I'm basically giving it.

95
00:23:38,000 --> 00:23:49,000
Okay. Very you should place this at the start future sick alien life plans. So now I corrected my prompt, it needs to happen immediately at the start.

96
00:23:49,000 --> 00:24:08,000
Like when you go to a, like it basically a throw so fascinating. Like when you go to a video editing tool like this, like, especially when you work with video a AI video maker agents. This is your frame you tell it where you put different pieces are.

97
00:24:08,000 --> 00:24:21,000
Right, same thing here. You basically trans your is still like interface is still like, you know, instead of dealing with this interface, you're translating the action you want to take into a prompt into a text.

98
00:24:21,000 --> 00:24:42,000
So you need to really be clear about immediately at the start futuristic alien like plans, inspired by the provided visual gently begin growing next to him. Now you've got it right. Something softly, softly on his left side, remaining, remaining secondary to the main scene.

99
00:24:42,000 --> 00:24:56,000
I think that you see how it clear it is immediately at the start, softly at the left in the background so I'm giving it also a direction in terms of set space, the man needs to be in the front.

100
00:24:56,000 --> 00:25:19,000
The plants need to be immediately at the back or secondary needs to be at the back. Oh, okay. So now it's making it okay now it's happening right after a short pause, the man slowly and deliberately turn his head towards the camera, which is doing it with so glazing to the horizon, and okay to his art toward the camera.

101
00:25:19,000 --> 00:25:27,000
Okay, got it. Competently and elegantly making a direct eye contact. This didn't get right, but I don't, I don't mind it.

102
00:25:27,000 --> 00:25:37,000
A soft spotlight gradually appears highlighting him clearly that's that's done really beautifully and then the placement of the plant. I love it so much.

103
00:25:37,000 --> 00:25:57,000
The re important the scene remains simple and focused, the man is in the primary visual, the man is the primary visual elements, also very important, while the growth of the plants softly complement the mysterious and powerful atmosphere.

104
00:25:57,000 --> 00:26:13,000
As you can see I introduced two objects, I clearly defined where they are placed, I clearly defined which one gets priority, I clearly defined the one that doesn't get the priority, how it relates to the other one.

105
00:26:13,000 --> 00:26:27,000
And now I got it. Fascinating. The way you create your prompts, I can talk about this over and over again. Look, I made this one, let me see where it is. Come find it. Is it here? Might be here. It should be here.

106
00:26:27,000 --> 00:26:52,000
So, I, I made this design and the finished work is here. This one is a bit like fairly complex. This is a little thick minor too but you see the movement of the plane, right. I learned the last part of the prompt that the movement is needs to be simple and focused, because I was exploring here and you can see if you do not emphasize on this, you get this range.

107
00:26:52,000 --> 00:27:17,000
Let me show you another one, or you get this range. What can I do with this? Especially for creating videos, for websites. This is too much of movement, I cannot do anything with this. You cannot loop it. You have to really cut it, and then for a good loop for videos, animation, marketing stuff, you need at least five seconds or ten seconds. This is terrible. I cannot do anything with this. Right.

108
00:27:17,000 --> 00:27:37,000
So, you need to make sure that the, the movements remain contained. I think I made this one with this one. Also, a realistic luxury passenger jet centered exactly mid-frame. So, I had to, I had to put this one here because I didn't want it to just give me a big range.

109
00:27:37,000 --> 00:27:57,000
Cruising perpetual forward motion through golden hour sky. Environment, volumetric clouds drifting gently across the background. So, I use my AI prompt agent, I think it was ChadGBT, to give him the prompts, then I edited it.

110
00:27:57,000 --> 00:28:09,000
Soft warm sunlight glazing of the fossil edge. I don't know that word. Camera, slow, really important. So, when you create videos, you have to define the camera movement. You cannot skip it.

111
00:28:09,000 --> 00:28:30,000
Slow cinematic tracking shots that follows the plane from slight three-fourth frontal angle, keeping the aircraft perfectly in the middle of the frame at a consistent scale. Motion, continuous forward progression. So, you get the camera movement and you got the motion of the airplane.

112
00:28:30,000 --> 00:28:45,000
Procedural cloud movement to accentuate forward momentum. Then a style. So, camera motion style. Premium. High end photo realism. Rich gold and peach tones in the sky.

113
00:28:45,000 --> 00:29:06,000
If you don't know any of these terms, and believe me, I probably don't know 30% of these terms, all you can do is to take a photo of a thing you like, ask ChadGBT to describe this photo for you as if you wanted ChadGBT or other LLMs would create the exact same thing.

114
00:29:06,000 --> 00:29:23,000
It sees it for you, identifies it for you, it tells you. Then you can look at the prompt and say, look, that material suggestion, I don't like it, and it needs to be more elegant. You can go basically word by word corrected it and go back and forth with it, right?

115
00:29:23,000 --> 00:29:38,000
And with that, that then gives me something I can work with. I can now go with the video editor, edit the movement a bit, move it back to the Figma, which I did here. So basically this is my Figma movement.

116
00:29:38,000 --> 00:29:51,000
It goes from A to B. Then I made the rest, which wasn't that hard compared to actually making the prompt was really harder for me, because that's the stuff I can make. That's not a big problem.

117
00:29:51,000 --> 00:29:55,000
Any questions regarding prompting? You can also unmute yourself and talk to me.

118
00:29:55,000 --> 00:29:56,000
It was nice, elegant.

119
00:29:56,000 --> 00:29:58,000
Are you making anything and stuff like yourself as well?

120
00:29:58,000 --> 00:29:59,000
Not yet.

121
00:29:59,000 --> 00:30:02,000
Alright, I think you can start.

122
00:30:02,000 --> 00:30:03,000
I think so. I think so.

123
00:30:03,000 --> 00:30:22,000
Alright. So now we sort of got the basic done of how we can basically prompt your video agent to make things right, how we can basically create movements, like 10 seconds frame. How can you make a short animation or short movie, right? So everything is in the prompting.

124
00:30:22,000 --> 00:30:37,000
Everything is in the prompting. You got your prompting right, camera, motion, style. You got to be able to place objects in the frame at which, where do they begin to appear, how they relate to each other. These are the things that you need to basically get it right.

125
00:30:37,000 --> 00:30:47,000
Then you have a story to tell. It's more than five seconds, 10 seconds, because these are the defaults. You cannot ask the creator. It's too much of a ask.

126
00:30:47,000 --> 00:31:01,000
So what you need to do is storyboarding. You create a storyboard. Typically a storyboard has six frames that tells a story and typically every storyboard width and height are the same, but.

127
00:31:01,000 --> 00:31:14,000
Right, so every story is a storyboard goes from beginning to the end. And then what you can do, create each image for each of these frames. Right. So I can show you.

128
00:31:14,000 --> 00:31:28,000
Yes, so I created this one, I chat GBT, I brought it here, make the man not, he's listening to music, and then just animate the rest. And then you go stitch them together, you have a short story.

129
00:31:28,000 --> 00:31:43,000
You have a short movie. You can, of course, you can basically, I wasn't really happy about this, I wanted to be having more movements. Man is thoughtfully listening to the podcast. He's not talking, looks around and he seems to be motivated.

130
00:31:43,000 --> 00:31:59,000
Right. It's still talking, but that's fine. I think that the movement and the way the level of emotion I saw was fascinating. So I did it. I went for it. But basically you create each of these images with chat GBT and you know how to make images with chat GBT.

131
00:31:59,000 --> 00:32:16,000
Right. And then you bring it here, any of his platform, says I'm video cling AI. And so basically I created this frame, then I asked it to animate. And it does a really good job, like passes through the office every's.

132
00:32:16,000 --> 00:32:34,000
And this is not even a bright from every is working. Who is every, like, it gets it. There is a vibrant vibe going on. Wow, I was even I even gave it a bad problem and got it right. So he walks into the meeting room and going towards the people, he seems confident, relaxed and smiling.

133
00:32:34,000 --> 00:32:51,000
A tip would be if you want to make a short story. Also at the end of the prompt, give it the scale. Don't let it guess. You just consume your credit. Look, I'm out of credit now. Right. I didn't give it a scale. It did create it for you.

134
00:32:51,000 --> 00:33:07,000
Just that there will be discontinuity. There will be like now everything is fine till here. But here it became this scale. And here it becomes absolutely like landscape. It doesn't really work. I mean, I got it right. But do not let it guess.

135
00:33:07,000 --> 00:33:28,000
Really define the objects. How they start, when they start, what their relationship to each other. Define the camera movement. Define the motion. Okay, if you have a man as a primary object, how he needs to move forward and backward, and define the style and also escape. Right. These are the things that you need to get right.

136
00:33:28,000 --> 00:33:44,000
Otherwise you get mess and then you think like these things are stupid. They are stupid because you didn't invest enough in your prompt. That's basically it is. When you invest, when you give it right, when you give it enough time and iterate on your prompts and be patient with it.

137
00:33:44,000 --> 00:33:59,000
I think you can create really cool stuff. Like I still cannot believe that this created. Where was it? I still cannot believe this to be honest. I've created videos like this in the studio for clients. I know how costly it is years back.

138
00:33:59,000 --> 00:34:13,000
Same thing with this. I've asked, created that image, then prompted, done. It's a very wonderful world of exploration. You don't have to rely on, I don't know what you guys are doing. Like what's your background? We got one, Julia is a designer.

139
00:34:13,000 --> 00:34:31,000
Andrew, what do you do? Destiny, fame, Sandor, if you guys are dependent on others to get some stuff to get certain stuff done for you, you can explore now. Oh yeah. So you understand fame. You understand where I'm coming from really clearly.

140
00:34:31,000 --> 00:34:46,000
Look, I run a product studio. I am at the mercy of developers and designers and creatives to create things for me. Sometimes they give me all sort of like reasons not possible. I'm a man that I don't, I like to achieve impossible.

141
00:34:46,000 --> 00:35:01,000
Before AI I couldn't do anything because I always like, okay, I don't know how to create videos. I don't know how to use Blender, but now I can just use Blender with MC, basically use MCP and work with Blender through AI.

142
00:35:01,000 --> 00:35:10,000
It's, it's, that's, it's, that's exciting. Especially if you are dependent on others' inputs to get your stuff done.

143
00:35:10,000 --> 00:35:23,000
Yeah, for me, for instance, it's prototype testing in general, because you know, to bridge the gap and make people understand what it could do for you.

144
00:35:23,000 --> 00:35:42,000
I've experimented with V0 and it's really good to remove the, the blocker, right? The initial blocker that you have the blank state, and also get everybody aligned on what to do next. So it's almost like encouraging brainstorming, which is super important.

145
00:35:42,000 --> 00:35:55,000
It is, it is especially for early stage prototype, it's for designers. You can even further, you can launch that app yourself, but you make a lot of mistakes and that's fine. You have to have time to make mistakes.

146
00:35:55,000 --> 00:36:11,000
But for designers, like give Replit a try. I hear really good things about it too. You can actually give Replit a try here too. I haven't worked with Replit, but just to showcase what's possible now, I am a bit more advanced than what you have been working on.

147
00:36:11,000 --> 00:36:20,000
But I'm about to release an app soon, God's willing, at some point, if I can get my by myself sometime.

148
00:36:20,000 --> 00:36:31,000
Okay. Let me bring it up. I did not ask a developer, a designer, everything that here you see I've done wipe coding. The app is ready.

149
00:36:31,000 --> 00:36:46,000
It's there. Certain things I wanted to do is like I have to build a big component for it. But that was me. And basically I just need to push it via Versa and deploy it, because it's the next JAS app. Yeah.

150
00:36:46,000 --> 00:37:02,000
Cursor, Winserv, I really want to give Cursor a try, but I invested heavily in Winserv. I don't want to lose that knowledge, but Cursor is very interesting. But yeah, it was crazy.

151
00:37:02,000 --> 00:37:16,000
It's basically, Julio, I think you were saying you just need two weeks of continuous work to get to this level that you can launch your app. Really, you just need to figure out certain things. And it's only learning by doing, right?

152
00:37:16,000 --> 00:37:26,000
Yeah, it's refining a lot of the time, refining the grant, giving more information, removing some information that is not relevant. Yeah.

153
00:37:26,000 --> 00:37:38,000
This is exactly. And then also you need to understand a bit of a code base and how you should launch certain application and what you need database, understanding of database, and next your auth layer.

154
00:37:38,000 --> 00:37:48,000
These are things that you need to understand like how you can like basically run a simple web application. You know, it's not like going to be an app that fit multiple applications. That's not the case.

155
00:37:48,000 --> 00:38:02,000
But a simple Next.js application you're two weeks away from understanding the core components of it and how you can build it. There is a lot of hacks that you can learn. I think next week I'm going to actually walk you through this.

156
00:38:02,000 --> 00:38:12,000
And hopefully by then I create that components that I need to create to get this thing launch ready. So that will be next week. But let me try Replete.

157
00:38:12,000 --> 00:38:34,000
I haven't tried Replete recently. I've used Replete before. Make me an automation for remote talents and help transcribe meetings. Note and pilot. Okay. Cute, cute. Okay. Let's see. Let's see if it gives me any free credit to play with. Do I have credit? Free apps. It's based on apps. Really? Okay.

158
00:38:34,000 --> 00:38:37,000
Should we start this rabbit hole? Should we make something?

159
00:38:37,000 --> 00:38:38,000
Go on make it.

160
00:38:38,000 --> 00:38:39,000
Yeah.

161
00:38:39,000 --> 00:38:40,000
Go on.

162
00:38:40,000 --> 00:38:41,000
Make it.

163
00:38:41,000 --> 00:38:43,000
Let's do this. Vibe code live.

164
00:38:43,000 --> 00:38:57,000
Okay. Let's do it. Huh? So let's see. I actually have a prompt for it. I wanted to build this. I forgot. I didn't have time. Okay. So build a, let's make it, make something fun.

165
00:38:57,000 --> 00:39:19,000
I want to, I have a great, useful chat LGBT agent, and that it analyzes complex situations from first principle thinking and offers and break it down on its core elements.

166
00:39:19,000 --> 00:39:40,000
And then I want you to give me, then I want to make that shown to users in some sort of in mind map. So there are two agents. Agent one, I can give you the prompt instructions when you need it.

167
00:39:40,000 --> 00:40:02,000
So when the agent says the user challenges user questions from first principle thinking, then user can see the results, ask it to reevaluate by typing and submitting or they say, I'm good with this, make mind map.

168
00:40:02,000 --> 00:40:18,000
If it happens, then the agent two receives the output of agent one and create the mind app type auto modern web app. Oh, that's great.

169
00:40:18,000 --> 00:40:34,000
I can improve prompt. Let me say let's see. Let's see what it does actually. It's my first time using Replet after two centuries. So I have no idea. The base dual agent system that analyzes this question using first principle thinking.

170
00:40:34,000 --> 00:40:50,000
The future is agent one user. Okay. Let's see. It's not really, let's do it. Let's launch an app together in about eight minutes. Fem, do you know, do you know which LLMs Replet or the likes of Replet they use in the background?

171
00:40:50,000 --> 00:40:53,000
I haven't used Replet for a very long time either.

172
00:40:53,000 --> 00:41:07,000
Okay. So you're on the same boat? Yeah, yeah. I used it a couple of times. This is learned, but I didn't like it. So I didn't touch it. I didn't touch any tools really being a developer, but then I looked at a few and I really liked cursor actually. So I've stuck with cursor.

173
00:41:07,000 --> 00:41:10,000
Okay. What's your favorite LLM right now to work with?

174
00:41:10,000 --> 00:41:11,000
Yes, Claude.

175
00:41:11,000 --> 00:41:12,000
Gemini.

176
00:41:12,000 --> 00:41:14,000
I am using it too.

177
00:41:14,000 --> 00:41:32,000
It hasn't been, it hasn't been that good forever, but recently it has been really good. I just mentioned in the chat, I've got a few prompts that I use for my process like kind of like what you've done here, but a few more steps are involved in the output and I use the output as a follow on prompts.

178
00:41:32,000 --> 00:41:37,000
Fem, one request. Do you want to take over the next Tuesday weekly jam session?

179
00:41:37,000 --> 00:41:38,000
Take over.

180
00:41:38,000 --> 00:41:40,000
Yeah. Yeah. It's actually, I bring a lot of guests.

181
00:41:40,000 --> 00:41:44,000
Yeah, let's talk. Yeah.

182
00:41:44,000 --> 00:41:45,000
Will you DM me on LinkedIn?

183
00:41:45,000 --> 00:41:47,000
Yeah, sure. I will do. Yeah.

184
00:41:47,000 --> 00:41:54,000
Cool. I love that because you had me on Gemini.

185
00:41:54,000 --> 00:42:09,000
Look, like something is getting a bit technical, I'm going to wind it down. Just to show you to fame. Sorry, I'm going to get back to that. Look at the credits, Gemini is 1X. I don't know if it's the same, which I think it's super, super powerful.

186
00:42:09,000 --> 00:42:17,000
I've been using it. It's uncertain other things. It's seven, five credits. My God.

187
00:42:17,000 --> 00:42:24,000
That's a lot. Right? Yeah. In cursor, I think it generally uses Claude 3.5.

188
00:42:24,000 --> 00:42:27,000
Claude is, I do not like Claude.

189
00:42:27,000 --> 00:42:42,000
It does okay on the code. If you're going to do an app and you're going to get it to generate an application like this, and you don't know code, then you're, and you're really going to just vibe. Claude does actually pretty well. And the way cursor works, it will fix errors and it will just keep going.

190
00:42:42,000 --> 00:42:46,000
It creates a messy file structure club 3.7.

191
00:42:46,000 --> 00:42:58,000
If you don't tell it, it's like what you said, right, with the aspect ratios, if you don't tell it what you want, it will go off and do what it thinks. It's the right thing. I like that too.

192
00:42:58,000 --> 00:43:08,000
Who is that for saving personal? Yes. Hey, that's pretty cool. Like, see what it's doing. Yeah, that's pretty cool. That's pretty cool.

193
00:43:08,000 --> 00:43:20,000
Create an account and then you can save your own, my maps. I really like this. Timeline, visual preview, two minutes, personal version. Ooh, that's pretty cool. That's pretty cool, huh? I'm going to make a video out of this. That's really cool.

194
00:43:20,000 --> 00:43:27,000
Pick approve and let's start. Let's go. Come on. We've got four minutes. Let's do it. Right. Let's do it. Let's do this.

195
00:43:27,000 --> 00:43:38,000
Here changes. I'm really fast. I'm looking at these things because I work with Bolt. Bolt again. Bolt, I worked with it a month ago. I had a lot of problems with it.

196
00:43:38,000 --> 00:43:47,000
And I'm looking at this thing, seeing what is offering. What I found really brilliant is the app plan. I think that's brilliant. I think that's brilliant.

197
00:43:47,000 --> 00:44:00,000
And I think fame would understand why I say it's brilliant. But to just elaborate on it, because if you don't set up your Postgres, which often you don't do when you're white coding, you have no awareness of it.

198
00:44:00,000 --> 00:44:19,000
I think, what they need to do is to check how savvy you are with the code. And if you don't, they need to make all those decisions for you as well, in my opinion. Yeah, you can. I mean, because I'm using cursor, like, you know, MCP. Now cursor will go off and just do all the super base.

199
00:44:19,000 --> 00:44:32,000
So with cursor you use super base for the backend. And it does all the set up for you, which is pretty cool. Like before you had to go into super base, register an account, create a database and then go get your API keys.

200
00:44:32,000 --> 00:44:38,000
Now you can just do that. Very similar to what this is just done. You don't even need to touch it, right, which is really cool.

201
00:44:38,000 --> 00:44:45,000
Because if you're not a coder and you don't know how to do that, that's going to be a couple of hours going down a rabbit hole, trying to get things sorted out.

202
00:44:45,000 --> 00:44:55,000
At least a couple of days in my case. So let's see what it did. It's still building. So it's cool. It's still going. It's working.

203
00:44:55,000 --> 00:45:14,000
It's gotta tell you the output of it. But do you know what they do this like white coding platforms that get you to that result that fast? Because when I do it with Vint Cerf cursor, you really need to, it's not, it doesn't give you that one shot effect too fast.

204
00:45:14,000 --> 00:45:23,000
You really need to go a step by step. I don't know what they do in the background that keeps you to that level of finishness. So to speak.

205
00:45:23,000 --> 00:45:42,000
I guess it's level of complexity of a thing that you're trying to build. If I just said build me a to-do list application that I can run in my browser, stores data on my machine and that I can just use offline and use simple HTML, CSS and JavaScript.

206
00:45:42,000 --> 00:45:54,000
I can do that perfectly fine. That's still in the olden days. That will take some time, but it can generate that in 30 seconds and you're done. It's amazing.

207
00:45:54,000 --> 00:46:01,000
And then the more complex you go with your ask, you have to have a better prompt. You can still achieve a lot.

208
00:46:01,000 --> 00:46:13,000
Maybe I will show you the prompts next week to show you how detailed those prompts can be. Because unlike the video, you can have a lot of detail in there and it actually helps. I would love that. I'm excited already running.

209
00:46:13,000 --> 00:46:24,000
I think I'm going to improve a lot of my work based on the instructions. But like you see what it's doing guys, it's installing all the packages because you need to install lots of packages to run it.

210
00:46:24,000 --> 00:46:35,000
I think it's going to go with Next.js. It's basically creating all the stuff that you need, which I think is very impressive. This is really cool because this will take a developer.

211
00:46:35,000 --> 00:46:51,000
This will be that very first step in architecting an application is identifying the libraries that it's now installing and just using. It's just, I think if you're a mid-senior developer, you are in trouble. You are in big trouble.

212
00:46:51,000 --> 00:47:07,000
Or you're going to work way faster. Well, I already established that there is no future for marketing agencies. Developers are out of the game too. Still, I think that you need a lot of senior developers because you really cannot build enterprise great software with this crap. It's jokable. It's laughable.

213
00:47:07,000 --> 00:47:27,000
Yeah, this doesn't really, a lot of these tools are not really being used by enterprise agencies, agencies who deal with people like Coca-Cola and Nike. No, we don't. We're not even allowed to do so. Yeah, you're not allowed. So for them, they're really focusing on building their own that they can use.

214
00:47:27,000 --> 00:47:43,000
I mean, our team does ask the LLMs to do certain things in the code, in that specific file, but they do not use the right side panel. That's a noble. Yeah.

215
00:47:43,000 --> 00:47:59,000
But if you've one problem, it touches five different files, lock debugging, but it's the same for code as well. A lot of these tools are not, you're not going to see an enterprise big global brand using cursor. No, that will be suicide. There is a lot of dependencies.

216
00:47:59,000 --> 00:48:15,000
There's a lot of like, like the macro services structure that that's no. Yeah. No. All right, then I'm going to see it. So if you guys want to stick around, I cannot leave this to be honest.

217
00:48:15,000 --> 00:48:30,000
Let's see what it does. It's actually I wanted to build this and like, show it or put it on our website as tools. I have a really good AI agent that does first principle thinking for me, which helps me a lot. So I wanted to expose it.

218
00:48:30,000 --> 00:48:53,000
But just mind you guys, this app is very expensive. I cannot. I don't think it's a, it's a cost that I don't think I can get behind it. It's going to be costing me per user session between one to $2.

219
00:48:53,000 --> 00:49:07,000
Going back and forth with AI agent, creating the mind map and it's, it's no, no, I don't, I don't think I would do that. It's expensive application to give it for free.

220
00:49:07,000 --> 00:49:20,000
I wanted to build this because I think it's going to be really useful to go from, I don't know what the hell I should be doing with this thing in my life to here's a mind map, all clear, but there is, I don't know if you guys use notebook LM.

221
00:49:20,000 --> 00:49:38,000
Yes, it's beautiful. Have you seen the thing that you can do here is upload your CV and then come back 20 minutes later. You will love it. If you ever want someone to like pat you on the back and give you some major credit, and you want to feel good about yourself, upload your CV in here.

222
00:49:38,000 --> 00:49:57,000
Is it, it's brilliant. It will literally go through everything you see in your CV and highlight and like really give you so much credit and be like, yeah, he's like, he's done this, but he's not only done this. He's done it in such a way that he's next their revenue in just a year.

223
00:49:57,000 --> 00:50:16,000
Yeah. The problem is how much, how long it is right? Sometimes if we want to get, or like get the attention of the hiring manager and they have like a limited time span, that's going to be a lot. So you also need to give the right prompt, especially within the land.

224
00:50:16,000 --> 00:50:25,000
If they are already interested in you, I think that could be important. So now that they are interested in you, they can listen to the podcast.

225
00:50:25,000 --> 00:50:33,000
This is not to send, this is just for you to listen to. If you never heard it before or seen it, it's amazing.

226
00:50:33,000 --> 00:50:42,000
So let me just, um, um, CV sample.

227
00:50:42,000 --> 00:50:44,000
Okay.

228
00:50:44,000 --> 00:51:00,000
Anthony dive. Let me just do this. So what should I do here? Get the CV and you just upload it in here. Yep. Choose file. Maybe it needs to be a PDF format and not entirely sure.

229
00:51:00,000 --> 00:51:16,000
Yeah. I was trying to sign into mine and see if I have a, oh, it's here. That's done. Okay. So, um, right. Is there, so what should I say? You need to say anything. You, you just, you click on the right hand side, but in the third panel where it says generate the blue button.

230
00:51:16,000 --> 00:51:34,000
That's it. All right. Generating and one of my colleagues actually used it because she was keeping notes for a product marketing conference. And then basically she put the notes into notebook and it generated a podcast and she shared it with us.

231
00:51:34,000 --> 00:51:37,000
It's amazing. It's really, really good. Yeah.

232
00:51:37,000 --> 00:51:54,000
So guys, this is actually moving forward. It needs open air. Why am I not doing this? What happens? Sorry. I went to open air off. No, I'm just saying. Alright, end of the day guys. End of the day. Yeah. Okay.

233
00:51:54,000 --> 00:52:20,000
So it's actually doing, but it didn't do what I wanted to do. I asked it was like, I want to use two AI agents. So you're asking me for one API key. It's fine, but that's the stuff it bothers me a bit live. By VINCEF, I can specifically in my end files. I've got two API keys. Here's a thing. Use this one. So, okay. That's the one I'm looking at.

234
00:52:20,000 --> 00:52:42,000
That's bothers me a bit. I really like to have complete ownership on my builds. I'm just that kind of person. So for the sake of seeing it, I'm going to create it, but I'm not happy about it. And I don't want to be bothered by saying, no, you stupid. You used to have created two API keys for me.

235
00:52:42,000 --> 00:52:43,000
Not now.

236
00:52:43,000 --> 00:52:51,000
But did it, or maybe not. That's mine. Okay. Can I, can I pull it up? It's goes on the recording, fam.

237
00:52:51,000 --> 00:53:15,000
So you sent this over famous acts CV and wow, this is somebody who has really been around the block a few times on in tech. We're talking to somebody who was coding for hotmail back in the 90s and is now a fractional CTO. Right. So let's deep dive into this career and really see what it tells us about how the industry itself has evolved over these last two plus decades. Yeah.

238
00:53:15,000 --> 00:53:17,000
It's like having a front row seat to history.

239
00:53:17,000 --> 00:53:18,000
Yeah, exactly.

240
00:53:18,000 --> 00:53:22,000
You can see how things have changed. Okay. So why let's start at the very beginning. Okay.

241
00:53:22,000 --> 00:53:24,000
1999 first roll.

242
00:53:24,000 --> 00:53:32,000
AKQA some big names right out of the gate. Microsoft, Hotmail, MSN, Expedia, X-Box. Who's who of the early internet?

243
00:53:32,000 --> 00:53:46,000
Basically, it's actually great confidence booster and like pitching the resume. Oh, it's actually, it's good for a lot of people that are not shy introvert, et cetera, et cetera, et cetera, cannot talk.

244
00:53:46,000 --> 00:54:08,000
And that's the only one I mean, I, I've been thinking about, okay, I can actually create podcasts out of this. So I actually did do a bit of a hack. I did a podcast. I downloaded a, I probably shouldn't say this and probably shouldn't record it, but I downloaded somebody's book on coding and I chucked it into this, downloaded it and turned it into a podcast episode and uploaded it. It was quite cool.

245
00:54:08,000 --> 00:54:12,000
Wow. We did that. That's crazy. That's crazy.

246
00:54:12,000 --> 00:54:35,000
I see, I see you're sitting, come in, let's look at what's happening. So it's actually that thing I like because great to see you Swati. I hope to see you soon again. But the cool thing about this, I mean, is that it's, it's FDS. It's basically full automated box fixing. I don't have to do anything.

247
00:54:35,000 --> 00:54:39,000
I'm just waiting for the preview to go. I like that.

248
00:54:39,000 --> 00:54:42,000
It's getting there, but now it's going away to fix it again.

249
00:54:42,000 --> 00:54:50,000
Again, the fact that I don't have to do anything. Welcome to First Prince, I can help you break down questions into fundamental.

250
00:54:50,000 --> 00:54:51,000
Okay.

251
00:54:51,000 --> 00:54:57,000
I'm dealing with this situation. I want to expand my business.

252
00:54:57,000 --> 00:55:14,000
Okay. Let's give it a spin. I'm dealing with a very important decision that will be the next. Look, I've created an application. Are you going to see next week? I think see you tomorrow. Okay.

253
00:55:14,000 --> 00:55:27,000
That will affect the next five years of my career. Not sure to expand my business to the US or Dubai or Middle East.

254
00:55:27,000 --> 00:55:33,000
I need clarity. Okay. Let's see what it does. I cannot send it. Let me do a new session. This is not working.

255
00:55:33,000 --> 00:55:46,000
It doesn't seem like it's hooked up. Is it giving you any feedback? Send button is not working. Is it hooked up properly? Open AI, API, can you bug fix it?

256
00:55:46,000 --> 00:55:53,000
I mean, it's impressive that it got the layout decently right. I didn't have to instruct it anything.

257
00:55:53,000 --> 00:56:05,000
That's the problem with this white coding because it's so designed and purposely built towards giving you something that you can preview. It opts in for simplification for no reason.

258
00:56:05,000 --> 00:56:16,000
I quite like that, actually. It separated it out so that you could, doesn't have to get to the complexity of doing backend changes because when you make front end changes, you've got to change everything in the backend as well, potentially, so to say.

259
00:56:16,000 --> 00:56:22,000
So basically it didn't hook up the backend just to give you a front end preview?

260
00:56:22,000 --> 00:56:29,000
Okay, just give you front end in this in this example. I haven't seen that often, actually, and can I say it will just do the backend as well.

261
00:56:29,000 --> 00:56:35,000
And then when you want to make a change, if you change the front end, you remove a button and you change it. It changes the backend as well.

262
00:56:35,000 --> 00:56:41,000
So that's kind of a lot of heavy work going on. But I like this approach where it just thought about the front end.

263
00:56:41,000 --> 00:56:49,000
And then once you have the front end defined and because you may want to change that and make it look better things around, then you certainly just got to do the front end code.

264
00:56:49,000 --> 00:56:57,000
It's much less work, less tokens, what have you. And then once you're done and you approve the front end, you can then start working on the backend.

265
00:56:57,000 --> 00:57:08,000
I usually start from the other way. I go with database, get some of the backend done, then go front end. But you know, this way is also to fix the numeric. Interesting. That is interesting.

266
00:57:08,000 --> 00:57:18,000
I'm actually very, I was very impressed that in the beginning that gives me an app plan, which I think is a right way to go for these white coding platforms.

267
00:57:18,000 --> 00:57:28,000
Yeah, I don't think I still want to give it a try to build my application with. I didn't send a message. No, it's still buggy frame. It's buggy.

268
00:57:28,000 --> 00:57:29,000
Restarted.

269
00:57:29,000 --> 00:57:32,000
Guys, it's not a difficult application.

270
00:57:32,000 --> 00:57:33,000
The application.

271
00:57:33,000 --> 00:57:34,000
Okay.

272
00:57:34,000 --> 00:57:42,000
It's still working. I don't want to do something in the beginning. Review a message, take a screenshot. It's taking a screenshot too, which is really good.

273
00:57:42,000 --> 00:57:47,000
It hasn't added a database. So what did it do? Where is it storing this information then?

274
00:57:47,000 --> 00:57:55,000
I really don't want to believe this. I really don't want to believe this. Let's see what I don't think it can get the mind map right. Let's see.

275
00:57:55,000 --> 00:58:00,000
It's 13 connections, but it hasn't visualized it. Right. It's a button. No. Okay.

276
00:58:00,000 --> 00:58:03,000
But it had the mind map component.

277
00:58:03,000 --> 00:58:04,000
You did it before, didn't it?

278
00:58:04,000 --> 00:58:06,000
Yeah. I don't know why it's.

279
00:58:06,000 --> 00:58:17,000
That's cool. You've got the JSON. Now it needs to actually visualize it in a component. Yeah, I think I'm basically four or five prompts away from building this.

280
00:58:17,000 --> 00:58:23,000
It doesn't look like it's got a database, so it's not going to save this anyway. So you're going to have to add that database.

281
00:58:23,000 --> 00:58:33,000
And then also make sure if you do a new session, the old sessions are saved and you can go in between backwards and forwards, but that would be it, then it'll be done.

282
00:58:33,000 --> 00:58:39,000
All right. Sounds good. Thanks a lot, Fem. Let's touch base on LinkedIn and let's get that session locked in.

283
00:58:39,000 --> 00:58:42,000
Yeah, cool. I'll message you and let's work something out.

284
00:58:42,000 --> 00:58:44,000
Looking forward. Next Tuesday, same time.

285
00:58:44,000 --> 00:58:45,000
Yeah, awesome.

286
00:58:45,000 --> 00:58:47,000
Thank you a lot, Xander, too, as always.

287
00:58:47,000 --> 00:58:56,000
Just a question. You mentioned that there are some records about the presentations. Can we reach it? Is it in the Slack?

288
00:58:56,000 --> 00:59:01,000
As long as it's in Slack. I'm going to publish the same recording after it's uploaded on Slack.

289
00:59:01,000 --> 00:59:05,000
Yeah, we'll have it in about 10-15 minutes, God's willing.

290
00:59:05,000 --> 00:59:08,000
And which channel should I look for?

291
00:59:08,000 --> 00:59:14,000
Oh, yes. Yes. Good question. So then let me stop sharing.

292
00:59:14,000 --> 00:59:18,000
All right. So this here, feel free to join in.

293
00:59:18,000 --> 00:59:23,000
Both of you. It's a cozy bunch of people. Feel free to join in.

294
00:59:23,000 --> 00:59:27,000
And it's also recording for last week also and Xander, too.

295
00:59:27,000 --> 00:59:34,000
So I will publish this as soon as it's fully encoded and ready to be uploaded.

296
00:59:34,000 --> 00:59:41,000
Yes, there's the link. All right. I copy. Wonderful. Good. Thanks a lot. And we see each other next week.

297
00:59:41,000 --> 00:59:43,000
All right. Thanks. See you next time.

298
00:59:43,000 --> 00:59:49,000
Thank you for listening to UX4AI. Join us next week for more insightful conversations

299
00:59:49,000 --> 00:59:57,000
about the impact of artificial intelligence in development, design and user experience.