# AI Show LIVE | Announcing a new Azure AI Translator API & Seth live coding

## Метаданные

- **Канал:** Seth Juarez
- **YouTube:** https://www.youtube.com/watch?v=NfWPijb6Cp8
- **Дата:** 21.10.2025
- **Длительность:** 1:05:17
- **Просмотры:** 11

## Описание

Tune in to The AI Show Live on Monday, October 20, from 8:30–9:30 AM PT to explore the new Azure AI Translator API (Public Preview) with Krishna Doss Mohan and see Seth Juarez share his latest work.

## Содержание

### [8:30](https://www.youtube.com/watch?v=NfWPijb6Cp8&t=510s) 9:30 AM PT to explore the new Azure AI Translator API (Public Preview) with Krishna Doss Mohan and see Seth Juarez share his latest work.

But look at me. Old school uh pietorrch. 2our show. Two hour show. Um, yeah. I used to write like a whole bunch of Look at this. Um, yeah, this is old school. Yeah, this is 2020. 2022. No, 2021. Look at this. I'm like literally showing neural networks. This is man back in the day. All right. So, this is what we got today. Transfer API with Christian agent framework. I was going to side of the eye. Look at the chat. Uh Janiscu number seven. Hello, sir. Hello. Vive. I have never I was talking to um it was in Portugal last week. I was talking to someone um brain fog uh about uh having someone on teach me how to vibe code. I don't know how to do it correctly. I don't trust it. And maybe if I knew how it worked, I would trust it better. You know what I'm saying? All right. Uh so let me uh let me take this down and I think Krishna is here. Um let me bring him on real quick. Krishna, are you ready to Oh, his screen's dark. Maybe he's not ready. Are you ready, Krishna? He's not ready. Um Oh, there he is. I see him. Thumbs up if you're ready to go. go, Krishna. Yes. Hey, Krishna. How you doing, buddy? — Hi. Doing good. — When was the last time you were on the show? I was trying to I was It was like a long time ago, right? Yes, it was on 2021 2022 time frame that I was in your second AI show. — Oh my gosh, I did not I'm like trying to look through the archive. I have AIA show episode 11, but I don't have before that. So, it's pretty cool. That's cool. So, uh we're talking about uh translation APIs today. Is that right? — Yes. — All right. So, let me do this. How about I queue it up and then you okay take questions afterwards? — Yes. Sounds good. — All right, let's do it. Here we go. AI show, not episode two, episode 8,000 million. Here we go. You're not going to want to miss this episode of the AI show where we talk all about the Azure AI translator API public preview, all the goodness with Krishna. Make sure you tune in. Hello and welcome to this episode of the AI show where we're talking all about the new Azure AI translator API in public preview with my friend Krishna. Krishna, it's been a while, bud. How you been? — Yeah, doing good. It's been more many years that we met. — I know. And we we've done an episode on the AI show many years ago. Turns out people still need translation. So tell us about uh what's new, what's going on. Yes, you are right. So it's very important to have translation with to break the language barrier across um the globe — and we have several interesting stuff uh coming up with the new API and um what we realized over the several years is um the enterprise and organization uh has a very diverse use case needs like um conversation translation, webpage translation, document translation, localization etc. Each scenarios has a different business criticalities, functional and non-functional requirements. However, on the technology front, there's an advancement with the generative AI LLMs which produce more natural and fluent translations and diverse translation based on the specific instructions what they use. But often it requires a lot of more time to generate the translation. However, the uh classic neural MD models produce more deterministic and but simple translation at much lower latency and which is better suited for the real-time translation scenarios. So what we are bringing these two technologies together in a single API that gives a customer to choose multiple AI models at the request level to address the diverse needs. Not only that, this also enables us to innovate further and introduce new functionalities like tone translation, gender translation as well as do a advancement in customization by bringing in adaptive custom translation by which user need not fine-tune a model but at the same time they can just bring that translation memory and use that as a reference data and produce customized translation. — So hold on. So this so I came at this thinking this was like text translation but is this also voice translation as well? — Uh be it text document or voice all this uses the translation technology behind the scene. — I see. So, so effectively what's coming in is voice or text and documents and then out is coming the translated stuff. What's the can you tell me the modalities that you can do? — Uh so the API produces a modality of text only. — Okay. — So a solution what is built on top of this API is the one which takes the voice as an input and the speech translation integrates this text translation API and produces it. And we also have a document translation uh which takes document as an input and uses this technology behind the scene and produces a translated document. — So I'm looking at this tone translation. So you're say tell me what this means is it like hearing the tone of the voice and then what does that actually mean? Oh, so this is um how so if you're writing a letter, so you can write a letter in a formal tone when you want to address to uh officially in your workspace work scenario but at the same time you want to communicate the same message informally to a friend or in a common setting then use informal tone. So you can communicate the same content in different tone uh in international languages uh based on the audience. — Oh I see. Because like for example I know that in uh in certain Asian languages there's a very formal tone and then there's like a medium tone right the term is called diglossia if you've ever heard of because I love linguistics and so you're saying that it can actually when the translation happens it can change the actual tone from high to low to business to whatever. — Absolutely. That's the magic this does. — That's really cool. Okay, that I was like having a little brain little I didn't understand. So, how many languages can you do this in? — Oh, we support 135 languages to translate across all of them and the new capability of Gen AI we support currently in 54 languages and our team is working on to add more and more. So you're talking one of the things that I heard from you is that you're there's still the neural machine translation which is the you know standard machine learning but you also mentioned that it's also doing uh large language model stuff too. Can you explain that? — Ah yes so the this is the new translator service. Okay. The translator service can do a translation either using a classic neural machine translation models or it can use a generative AI model. Currently in the preview we support GPT40 and GPT40 mini and we have a path to add more and to get the goodness of tone translation gender translation adaptive customization that's where you use the generative AI models I see. So okay so let's see if I understand this. So you can choose between which model that you actually use when you're doing the translation. Is that what I'm understanding or does it do something automatically? — Exactly. That's the one. So it will become more clear when I show it to you in action. — Well, let's do that. Let's take a look at what you've got. Uh so that maybe we can get a sense for how this works. What do you think? — Absolutely. — All right. Let's do it. — What you see here is um u the request body of the API. Okay. And here is the request. And here we are translating a text uh in from English language to a target language Spanish. And in this request you could see it's getting translated using a generative a model photo mini as well as here no uh model is specified by which we default take the classic NMT. — Okay. So let's pause right here. So this is effectively like a API caller. This Bruno app is just basically it calls APIs and this is calling the uh the translator API directly and this is the body of the request. Is that right? — Yes. — Okay. Cool. Yeah. I just want to make sure I knew what was saying. And then you're saying that when you do the translation you can do multiple targets. — Yes, Uh and you can specify for each target what model to use all in a single request. — I see. And then I guess on ostensibly on the right hand side is the uh translation. — translation response. Here is the response from the generative AI 40 mini. And uh here is the translation from the NMT classic NFTt model. And so how many tokens consumed for to do this translation? How many characters uh charged for NMT? It's all very transparently shown to the user. — Ah that's cool. And I'm looking at that cuz I know Spanish and the meaning is the same but they're slightly different but not it's almost like you know saying potato kind of thing which is kind of cool. — That's true. So that's why if you see in terms of translation accuracy for popular languages like Spanish, German, Italian so and so you'll find very minimal changes but the key thing is the GPT models produces a more fluent translation. Yes. — than the classic energy — more of a formal tone kind of thing. — Yeah. Like the word potencia muchosto is feels more um I don't know it has more of a good feel to it. The other one it's not that doesn't have a good feel to it but it feels more less vernacularish which is kind of cool. — Right. And then what I want to show you is um how to produce a gender translation. So — uh if you want to translate a simple word like a doctor, the doctor is a gender neutral term in English. But when you translate into Spanish, it has a very specific term based on the a male or female doctor. And by default uh when you produce with classic NMT it produces whichever is the popular and sometimes it produces a male uh — term here. But with the GPT you can specify whether you want a female translation or a male translation — and how can now I will show you a very more interesting example over there. So let's take um in this example uh I created a translation I want to meet Dr. Selena and what time the doctor comes to the hospital. So here the name Selena is associated with um female automatically by the uh mission translation system be it neural or NMT models it automatically does that — but at the same time if I replace Selena with uh Ramen which is an international name the system may not have the all intelligence and also sometimes the name is a genderneutral term. It means this ramen is a name can be used both for male and female in which case when I run a translation you could see that um the GPT uh and classic NMG produces slightly different translation. So here if you see the classic NMG took it as a U male doctor and so does the GPT models because GPT also understood it's a is a male — and uh when you direct the GPD hey I need Raman is a female and then it produces a right translation considering Raman is a female doctor. So how this helps in a real life scenario um the customer has different signals uh outside the translated text. So you can pass those signals as part of the translation request so that you get the right translation. In case of a conversation, if I'm conversing with the in a customer support scenario and when the c customer support knows he is conversing with the female client, then he can give that uh signal as part of the translation request and he can continue the conversation with the client in a right gender term. — That's smart. And can you scroll down? I want to see what the formal one does here at the bottom. — Yeah. is because in Spanish there isn't this notion of formal speak — uh and so it looks almost exactly the same but you have female but the fact you can put it in there is pretty cool what other what other things do you have here these is this is very surprising by the way — y and uh so to see the formal as you right I said uh that's where the German uh comes into the picture so the German language has the formal tone and informal tone so that's why if you see here it's my pleasure to have joining you as for tonight and you see the difference in translation with formal and informal tones. — Yeah. And I don't know German, but uh for those uh for the eagle-eyed German watchers, uh hopefully you're seeing that the tone tonality has been changed, which is cool. So, what else you got in terms of uh of cool things? — Huh? So, so those are the very short summary of um ability to translate um uh using multiple models and you can also use GPD photo as well as follow the same request. So you can produce multiple translations, you can produce translation based on the tone, gender and so on and so forth. — Oh, I see. So sorry. So effectively you can uh because the language at the top says the source language you can have multiple language targets with multiple with the different genders as well as the different tones if you want. — Exactly. — That's cool. So tell us about the road map. — Yeah that's cool. Uh so if you see the translation service currently we have the text translation u produces uh using both classic NMT and uh geni models and the document translation uh pieces in box and we should have this sometime closer to Ignite in your hands or let us by early uh next calendar year um that's where we have the document translation API which internally calls the LLM models based on your choice and produces the translation. — I see. So you're basically feeding it documents and then giving it the same tonality, gender, etc. And then it will translate those documents for you. — Exactly. And you would notice on the road map slide on the left side uh you see task specific small language model. — Okay. — So what does it means is um we are building a totally new machine translation model uh which takes the goodness of classic neural entity as well as the genai large language model and this task specific model would uh would be highly performant and with higher accuracy than both of them and that is in works and you would see us rolling out rolling it out in within next six months. — Okay. So I jumped the gun because there was I thought that this and this were the same, but this is different. This might be like tell what do you mean by like more task specific? — The task specific is a if you take a generative AI LLM models, it's not tuned for translation. It does so many other things. So it's overload. It's bloated. That's why it's a very large language. So what we do is we take those kind of large language model and tune it to do a specific task called translation. So thereby we can reduce the footprint of the model size as well as the time to perform the task. So that's where you get the performance as well as uh the accuracy of gen model both in one shot. — I see. So it's more specific to the task of translation a model a smaller language model to just do translation. — Exactly. Once we have that then the user will have a choice not just the classic NMT they can use gen model as well as the new model what we roll out both for text translation as well as document translation — and that's amazing. So here's a question where can people go to find out more? Oh, so we have a lot of links. So, ak. ms translator documentation that has a complete uh user documentation uh where you will see how to use this API and what are the limits and what is the pricing and etc. and aka. ms translated languages that will list you all the languages what is supported by classic NMT as well as the gen models. — Well, uh Krishna this has been amazing. Thank you so much. Translation is something that's near and dear to my heart. I don't remember if I told you, but that's literally what I studied in college, computational linguistics. It's one of my favorite things. So, thank you so much for being with us, my friend. — Thanks, sir. Nice to meet you again. — And thank you so much for watching. We're learning all about Azure AI transfer API, the public preview. My friend Krishna, thank you so much for watching and hopefully we'll see you next time. Take care. How about them apples? That was amazing. Uh Krishna, thank you so much for that. Uh by the way, um as I was uh because we recorded this uh I think not last week or a couple days ago. Was it last week or week before? I don't remember. Uh but as I was looking at that, I was actually in Portugal and I was like, "Wow, the speed at which these things work. You could probably make device applications to help you with do this kind of translation. " Have you seen stuff like that? — Yep. So, we do have the our API as part of our translator mobile app which you can use during travel and you can use voice as an input or text as an input and translate uh dynamically and have a real-time conversation with anyone in the in face to face. — That's so cool. Like I said, I love the the translation stuff. And apparently there's a new link that you wanted to tell us about. Tell us what this is. — So what I've done is um someone has asked in the chat also how you doing white coding. So basically this is a web app uh written with uh using white coding uh without using any physical manual touchup and I do have a blog on that and u and creating uh creating the web app then I thought okay it's good to share with uh in the Git app so that people can uh jump start with the uh API and try it out themselves without spending too much time and then once they are good yeah they can execute as part of their production workload. Yeah. And I like the thing that I love about this is I don't know like language barriers are a real thing. — Like I said, I was in Portugal last week and I speak Spanish and I speak French and so like I could sort of stumble my way with like Spanish or French words and people were like, "Okay. " But the ability to break down language barriers with this kind of thing is pretty amazing. — And so I I love this. So any tips for people that are wanting to use this? — Um the thing is um the API is very broad and has multiple capabilities. It depends on your scenario. You need to pick the right uh model and if you have a real-time translation and you need things fast then neural MT the classic NMT model does the job. But if you need something more variety, more var various translation where you need more fluent um then you can use um uh GPT models uh along with additional attributes and get the translation what you want and you if you have a post editing workflow then you can get the translation with both classic NMT and neural and then uh use your judgment to choose the right translation and do the right variance. with you. — And that's the cool thing, right? Like cuz like the thing that I think I'm even recognizing even more now that I' I've talked with you is that this could be used for multiple applications. Like for example, the live translation one shot, you know, use the standard model, get something out, but then like after the show is done, then you can use the uh batch kind of you could batch it in a little bit way and then change the translation depending on the your target language etc. and have people review and do even more stuff. And so there's a way like both modalities are supported. Is that right? — Exactly. — Well, Krishna, you're the best, dude. Thanks for being with us, bud. — Thank you. Thanks, Ed. And uh we'll see you next time. Uh thanks for being with us, bud. — Byebye. — See you later. How about that? Uh Christian's awesome. Like I said, he was like I think I'm just remembering that he was like on our one of our very very first shows a long time ago. So uh we're going to give him a round of applause here. Um thank you to Krishna on that. Um uh because like I said, this is super important stuff. uh when it comes to um like translating like it's so accessible. Now I'm turning because I want to share my screen here because we need to get into uh some live coding. I don't know. Vi I'm not a vibe coder. Ma I was talking to Mattie Montakeia. Uh she we're on the we are on the same teamish. Uh and she was like hey you actually need to learn how to vibe code correctly. And she gave me some pointers on who I could have on. And so maybe we'll have them on next week. be like, "Hey, set me up on the vibe codings. " Um uh on the vibe codings. Uh okay. So, let me go and let me go back to my screen here because uh I want to we wanted to take a look at um Man, it's so empty in here. I got to put some music on or something. Uh okay. So, we I wanted to look at the um agent framework. So, let's bring this over here. And I haven't done a C# thing in a while. So let's do a C# thing. So I am going to go straight um Oh jeez, that was dumb. Somebody turned me off for a second. Wow. You know, if I bet there's some people in the world that be like, you know, we can turn you off just like that. By the way, like I was saying, I was vibe coding uh for a second. Um I was like, "Oh my gosh, am I doing it? Am I vibe coding? I didn't know you could do this. " Uh so this is cool. I'm just going to say um net new sln minus n a gf pro project. Look at them apples. And so now what I can do is I can control shiftp and I can be like uh okay. So, we'll skip this and I'll just say thanks. You can stop. You can stop. Thank you, sir. Love it. Look at that. It I think I half vibe coded something. By the way, I forgot to mention I wanted to make sure I could see some people. Uh yes, I know. I uh No sound. Yeah, I know. I fixed it. That was dumb. I made a silly. Um uh let's see. See, we got some people. Hello from um uh hello my friend from Bahrain is the is that's the flag. We did some our uh Rian our producer did some research. We have a crack team of professionals here on the AI show. Um Bahrain uh thank you. Um uh yes. Uh by the way, if I'm looking at the side of the eyes, I call that vibe coding as yolo coding or yo-yo coding. I like it. Yes. Uh we lost audio with Seth. Yes. Somebody there's a little knob. I guess I'm the little knob. There's a little knob that turns me off or on. And it was I was off. Uh Pablo, can we listen to Seth? No. Yes. Sometimes people can listen to me. Other times people don't want to. Uh, okay. Oh, it's cut. Cutter. Oh, man. Cutter. Uh, thank you. Um, by the way, I um the Arabic language fascinates me. I had a guy I lived in Spain for a couple of years and u I had a guy from Sagal talk to me about um uh teach me Arabic. Uh so looks like our looks like the cutter flag and the Bahini flag are similar. Sorry. We were close. We were close. Apologies my friend. Um, okay. So, we vibe coded that. Um, we're starting to vibe code. So, let's do this. Control shift P and we are going to Can is there like a open solution? Yes. Yeah. Sorry, that was loud. Oh my gosh, I'm so sorry. Open solucion. Um, oh, it's here. Look at that. So, I just found this out that you can do this in Visual Studio Code, which has been delightful. New project. We are going to do a Holy cow, there's so much stuff. I'm just going to do a console app. Um, my first agent. Not agent. Yeah, this is great. I love it. Yes, I love it. So, now that I have this thingy, oh wow, has grown up. Look at this. I can just hit play. What? Guat delightful. Uh, hello world indeed. Okay, so let's see what we need to do for agent framework here to get this to work. Uh, so netep quick start basic agent. I don't want API keys though. Is there a way to not do API keys? Um, bearer token. We'll do that. Yeah, cuz I already have um this is cool. So, let's do this one. So, we'll we'll uh wait, is there more examples? Let's see. Go agent. Let's go to uh getting started with agents. Hold on. But first, I got to add the package, right? Let me go back. I got to hurry up, too. I'm going to run out of time. I want something to work. So, net add package. Okay. So, I'll control I'll copy that. And then I'm going to add is there a way to like add a dependency here? Add nougat package. This I don't even know which one we want. Is it Microsofts. ai? latest. Oh, that's not what I want. Remove nougat package. I might have to go to ls cd my first agent and then let's go to um let's go to uh this thing and let's just cop let's do it let's do the right thing we are going to there are no stable versions Uh oh, I see. That's why I was sort of bloop. Okay. Uh so I want to do pre-release. Man, I am loving this. net stuff. I uh by the way, um I my first ever thing was net. like I was a VB6 developer in the early as and then uh as soon as C# before I even finalized um I moved to C early 2003 and so that's and then I went and did research in I was a computational linguist I can turn my tongue into a computer um and that's where I switched to Python. Uh, so let's go over here and now that we've done this, let's go to uh basicagent. net. Let's go to getting started with agents. Uh, let's do this one. Uh, let's see what we got. Let's just copy it. YOLO. YOLO for life. Uh, oh shoot. I got to go over here. Uh here arr. Uh why is there like some packages I need? I think I need some packages. Let's see what we got here. Um maybe it's in here. I wonder if I just copy this and then I go to uh this. Let's see here. I wonder if there's like a net restore format format. How do you format this? uh format format and then let me go like this and let me double check to make sure that I'm not doing anything target framework net 9 uh implicit using root name space I'm just going to copy this what what's project reference include Oh, I see what's happening here. So, they are looks like they're using the Microsoft agents AI like they're actually like. Yeah. So, that's why they can't get this, right? So, I have to control zed this uh because I need to download the preview and then I think I need to get those other two. Hopefully, those packages are stable. Uh yeah, they must be because um those um yeah, I think I need all of these. So, I'm going to put those in here. And then uh like so. And then let's see what the error is when we try to uh uhoh. Oh my gosh. include a lower bound. Uh okay. So let's go to here and let me say uh net restore and then re-release. Okay. net restore. Well, now this is getting frustrating. Um, no depth. So, what was the error? Unable to find packages. Okay, so I'm going to do this. I'm going to net add. So, let me take this off cuz I think I need to pre-release this thing. Uh net add. We'll do this. But we need this thing right here. Open AI. Let's do this. Okay. Uh net restore. Okay, there's some I wonder if there's an identity one. Uh, let me do this. Let me just pre-release all of them. Uh, so I'm going to take this one out and then um let's do this cuz I don't Do warnings bother you folks? They bother me. Um, okay. And then this one I'll do the same thing. Uh oh, here it is. Uh add the version number unless you are using central package management. Uh thank you Christian. By the way, this dude is the man. Oh, my earpiece just fell out. Um I think we might have met in person. Um but I don't remember. Thank you, Christian. Uh, so I am doing that now, but I'm doing it in like a silly way. Uh, okay. Okay, now we're back to this goodness. What is this? So, I can't do this. So let's see here. Dot. Looks like there is something here that is not um that's unhappy. I think this is a copy of the other thing. Um CLI credential works for me here because I have to do a key. Uh let's find a let's find an endpoint here. AI. ure measure. com. Do I have any models in here? Let's see. Uh, Modellos models. Modell. Yes. Here, I have a I have one of these handy dandy things. Um, I don't think I need all this. Okay, we'll go full GPT4 mini. Okay, so this is the problem. Uh, it's not this is a real thing. Um, so let's do this. Um, bar client equals Yes, thank you. Go. Uh, cool. Uh, and then we need to uh do AI agent dot how do I make a new one? Uh, so uh var agent equals new AI agent client go to definition. Is there a way to go to definition? There is. Oh, it's an abstract class. How do I make one? Oh my gosh, this is frustrating. Let's go to the sample here. Maybe I should have gone to just the simple sample. Simple sample. Um, basic. net. Yeah. So, I'll let me try this. Um, uh, let's I should have Mark Wallace come on here one day. So, this is worse. No, let's go back to this. Let's see what I do with a client cuz I want to get something working here before I have like like um is there such a thing as an AI agent builder? Uh okay. I will So, I will do that for now. Um, and let's see what we can do with the client. Client dot, but already did that, right? I don't understand. Uh, why can I not just do the agent here? Um, let me go back. I think there's something um in this agent's AI thing that I do not have in mine, which is as stinky as stinky. Uh because yeah, see I don't think I have this in here. Let me check my project here. Excuse me. Uh I do not have it. Okay. Uhnet add package. No, it looks like I do have that. So, let me go control Z dead. We're going to go back to the beginning there. So, this is the thing that does not have it. Interesting. Um, I don't know. Uh, let's see. Let's see what the chat has to say. Uh, while I'm looking around. Oh, surely. We can hear you fine now, unfortunately. Yes. Yes. I know. Sorry. Uh, I'm trying to look at this. I think I might have like a I might not have the right build on this. Um, uh, not just the Microsoft. extensions. Yeah. So, I uh I think I went through and fixed all of these things. Uh, so they all have the right deal here. So, let's see. Let's see if we can is there a way to like to look at the um let me look at the um Microsoft. agents. Is there a way to like view what's in there somehow? Uh Microsoft. agents go to definition. No. Um, yeah, something I might not have the right thing here. And it's what I'm looking at. It's it looks like it's implying that I need a build of this thing. And the build of that thing is int. Source I think this is the thing I need and I just don't have them. AI agent extensions as builder. Um, so AI agent builder is the thing. And so let's see if I can get a new uh bar builder equals new AI agent builder inner agent. Is there like a number two inner agent factory? So I think Oh no, it doesn't like that either. Um, what can I dot into this? I think this does not create an agent. It creates a client. So, I got to find out how to build an agent. Hi, agent. Gosh. No. No. So, I don't know how people can actually do this without because obby this isn't an agent, right? No. This won't let me do it. Oh my gosh. Why? Uh, let's go back and let's see if I can Oh, Steven Tob, Mark Wallace. He's the man. Maybe it's create AI agent here. Let's um I'll put this over here to the side so I can look at it. So, open AI client, but I don't want to do like Oh, okay. It's get response client get create AI agent. I don't have that. Is this uh unhappy? Oh, here. Uh-oh. Time for the walk-off music here. Oh man. Yeah, it looks like um What a bummer. Uh, let me see. getting started with agents. Yeah, I think these samples assume that you have the correct and latest Microsoft AI agents open AI thing and I just don't what a bummer. I don't have those things. Um, I'll ask maybe. Is there anything I need to be doing here? No. Implicit using. Let me see here. Yeah, this is all enabled. exc. It's the same thing. Well, my friends, I don't know why it's not letting Well, no. I know exactly why it's not. It's because the Microsoft Agents AI OpenAI Wait. Wait a minute. Do I Did I do this one correctly? Oh, I think I did. Yeah, I did. It's Oh, wait a minute. Oh. Oh my gosh. I was wrong. Let's play it. Here we go. Error. Uh, what's the error? Uh, client model not found. Oh, let me take this out. Running it one more time. Oh, yes. Yes. So, it was me. I was the dumb here. Uh, I just needed this. I needed this in here. Um, cool. So, I got my first thing uh working. Uh, I just want to say thank you so much for spending some time with us. I know uh you have a lot of places that you could go to, but you have chosen to spend some time with me. Thank you so much for watching. Next time on the AI show, we have uh let's take a look. Uh AI agent for beginners with Corey. Uh he's awesome. A good dude. Uh we'll get that one in the can and uh going. Um and then hopefully uh looks like I have some TBDs and so maybe I'll continue to work on this and not do a silly. By the way, I haven't used net in a while. So, uh, that was amazing. Thank you so much for spending some time with us. I know you have a lot of places to be and we appreciate you being here with us on this the AI show live. Thank you so much for watching and hopefully we'll see you next time. Take care. Heat.

---
*Источник: https://ekstraktznaniy.ru/video/45953*