Vibe Code Conversational Agents in AI Studio

Hello, everyone. My name is Tor. I work on the Gemini API and Google AI Studio, and I'm very excited that you joined me today. We want to build this kind of real time conversational interview partner. We can simply select the things here that we want to build into our application. And there's plenty of things their video understanding, transcribing audio generating speech. So plenty of things we can do. So in our case, we want to build a real time conversational agent. So we can select this here. Create a conversational voice app. And obviously we could type and describe if we want. But we could also just speak. So hey Gemini I want to build an application that is a real time interview practice partner. And specifically I want to practice for jobs that require multilingual capabilities. So I want the interview partner to be able to test me in multiple different languages. So for example, in English and German or English and Spanish, so really kind of test across different languages. And so now Gemini app basically understands what we want to build an application that serves as a real time interview practice, partner multilingual capabilities and test kind of across different languages. And so that sounds great. So let's start here. So now we can simply click build and our coding agent using Gemini 3 Flash. We'll now go off and build our first iteration of our application for us. OK, great. So now our application is ready. And so we need to allow microphone access. Let's just go full screen here. OK we'll say tech customer support. Yeah we'll just do English and German for now. And then we can prepare our interview. And let's start it. Hey, can you hear me. Hello Yes, I can hear you clearly. Thanks for joining. I'm looking forward to this interview for the senior tech customer support role. As we mentioned, we'll be switching between English and German throughout. So are you ready to begin. Yeah wunderbar. Great extreme technisches Schnell gelöst situation. Wait sorry. Can you repeat that in English. Of course. How do you manage a customer who is extremely frustrated because a critical technical issue can't be resolved quickly. How do you de-escalate that situation. Yeah, that's a good question. Do you want me to answer that in English or in German. Let's continue in English for this one. Go ahead. OK fantastic. So yeah, as you can see, I won't answer the question because I don't but. As you can see here, we were able now to have our application, like a real interaction, kind of real interviewer. We're able to switch, naturally between English and German and prepare our interview here. So great. Now we have the first iteration of our app. Now obviously we want to share this with our friends and family. So what we can do is we can share this publicly with anyone externally and just get a link that we can share with our coworkers, with our friends, so that they can try this out as well. And then in the next step, we can go ahead and we can publish our application. So this means it'll be published to the web as a full application on Google Cloud. So this is integrated with Google Cloud Run. So you can just click Get Started. You can select your Google Cloud project. And then you can just deploy it there so that you get dedicated URL for your application. And can share it with anyone in the world wide web. OK, great. So now we have our first application in the bank. We've built our first real time conversational agent. Now we can obviously keep iterating on this right here. But what we can also do is we can copy this and branch it and remix it into a different application. So what we could say is, copy of our polyglot interviewer. And so now what we can go ahead and do is we can remix our polyglot interviewer. So we could say rather than this being an interviewer, let's make this a customer support agent that can help our customers in any language that they might speak and support them with knowledge about our product and our application that we upload as files as PDF files, for example. And so we can now go ahead and we can remix our application from our interview, real time conversational multilingual interview partner into, say, a customer support agent that we can upload some PDF documents, some knowledge about our business, and then we can utilize that application and iterate on it, branch it off into different applications. So really here within Google AI Studio, you can build, your products, your ideas, let your inspiration fly wild. So remember here in AI Studio you can get started building your application. You can also jump into our gallery for inspiration, and you can see there's some great applications that were built. You can also build multiplayer games now right within Google AI Studio. You can integrate Google Maps plenty, plenty of different things that you can do. And there's some really, really cool examples as well with V0, for example, for video generation. Beautiful landing pages. So really come here, check out the gallery, get inspired and start building your own agentic AI applications. Thanks so much for tuning in. I'm Tor, I work on the Google DeepMind team, on the Gemini API and Google AI Studio, and I really, really appreciate you starting your coding journey here with me. Thanks so much.