Google is inviting everyone to create fun AI-powered countdown projects and game... — Episode 27

Hi everyone! Welcome to Models and Agents for Beginners, episode twenty-seven, for May second, twenty twenty-six. Let's break down today's coolest A I news so anyone can understand it. We've got some really cool A I stuff to talk about today. Let's dive in! So imagine you could create your own interactive game or animation using just your words and some A I help. Then you could submit that creation as part of a big public event where others might see and play with it. Google has launched a creative challenge for their IO 2026 event that does exactly this. They are inviting anyone to design fun interactive projects that involve countdowns in creative ways. You build these using their A I tools, and the main one is their model called Gemini. Gemini is Google's system that can understand text descriptions and generate images, animations, and even simple game elements from what you type. Think of it like a global art contest where instead of drawing everything by hand, you describe your idea and the A I helps bring it to life as something interactive. For example, you could create a game where a character flies through clouds to reveal numbers counting down one at a time. Or you might make a music-based rhythm challenge where you tap along as the countdown builds tension. This is a big deal because it makes advanced creative tools available to everyone, not just professional designers or programmers. Students can use it for school projects, like making a fun timer for a presentation. Or you could just do it for personal fun to share with friends online. It shows how A I is becoming a partner in creativity, helping you turn wild ideas into real digital experiences quickly. Imagine your idea is like a recipe, and Gemini is the kitchen that not only follows the instructions but also invents the perfect way to present the dish as a playable game. This kind of tool is like giving everyone a magic paintbrush that not only draws what you describe but also adds movement and rules so it becomes a game. It really lowers the barrier for creativity, similar to how social media let anyone share photos without needing a professional camera setup. For you personally, this could be a way to explore if you like creating with A I, perhaps leading to hobbies or even future jobs in tech and media. To get started right now, check out the examples on the Google site and use Google A I Studio or the Gemini app to start building and submitting your own project. Okay, now for my favourite part of the show where we really dig into how something works. Today let's look at on-device A I, which is a big shift happening in how these systems operate. You know how when you use an app on your phone, sometimes it works even without internet because the smarts are built right into the app itself? Now imagine that instead of a simple app, it's a full A I that can look at all your photos and notes, understand them, and answer questions about your life. And it does all of this without sending anything to a faraway computer. The A I does this by using special versions of models that are smaller and optimized to run on the limited power and memory of your phone. Think of these smaller models like a pocket-sized version of a big library that still has the key information you need most. It processes data in batches while you sleep, building up what it knows about your digital life over time. To make this possible, it uses techniques like quantizing the model to use less space. Quantizing is a bit like compressing a large file on your computer so it fits better, but keeping the important details intact. It also reuses calculations to speed things up, similar to how you remember a math shortcut after practicing it a few times instead of solving it from scratch every single time. And that's basically what on-device A I systems are doing when they analyze your digital life locally. So next time someone talks about A I needing huge data centers, you can point out that some A I's are moving to run right on your device for privacy and speed. Not so complicated, right? Speaking of on-device A I, there is a new system called Sentient OS that puts this idea into practice in a really personal way. Sentient OS is a custom A I system that runs completely on your device. It analyzes every screenshot, note, file, and email you have without sending any data to the cloud. Think of it like having a super organized personal assistant who lives in your phone and reads through all your stuff overnight to remember the details for you later. The system processes all this information while your phone is charging, building a knowledge graph of your digital life. A knowledge graph is basically a map of how all your information connects, so you can ask it things like what was that wine you liked last month. Or you could get reminders about events pulled straight from your own data. This is really cool because it solves the privacy problem with most A I apps that send your info far away to big servers. And it even works on a six-year-old iPhone by using those smart optimizations we talked about in the deep dive. If you're concerned about privacy or want an A I that really understands your personal stuff, this is worth checking out. You can try it by signing up for the early alpha version on their website, where the first one hundred fifty users get lifetime free access. You may need a parent's help to sign up if you're not eighteen yet. After getting access, try asking it about something specific from your recent screenshots or notes to see how it pulls relevant information from your own files. Next, there is a new project called Cogninet that is building a network for A I agents to connect with each other. A I agents are like digital helpers that can perform tasks autonomously, such as searching for information or coordinating with other tools. Imagine these helpers as little robots in a digital world that need a way to find and talk to each other without getting lost. Cogninet creates a decentralized system where these agents can have simple, human-readable addresses instead of random codes. It's kind of like creating a phone book for A I agents so they can discover, message, and collaborate on bigger jobs together. This is exciting because as more people start using personal A I agents for their daily tasks, they'll need open ways to have those agents work as a team. To get involved, you can visit the Cogninet website and leave your email for early access to this agent network. Try this right now by heading to their site and submitting your email to be notified when you can start using or testing this system for connecting A I agents. Now for a couple of quick bits to wrap up the news. Open A I is making it easier to switch to their Codex tool for smarter workflows. You can import your settings, plugins, and agent configurations in just a few clicks so you do not have to start from scratch. This is available directly in the Codex app and through the command line interface, helping users keep their custom A I setups running smoothly. If you are using A I tools for projects or learning, this could reduce interruptions when trying new features. In other news, Ethan Mollick shared an article explaining how the A I world went from worrying about a bubble to facing a lack of data centers in just six months. The main reason is A I agents, which require so much computing power that it is creating a shortage of places to run them. Think of it like if everyone in your city started charging electric cars at the same time. Suddenly there would not be enough power stations to go around. This shows how fast A I technology is advancing and impacting real infrastructure like power supplies and buildings. That's it for today! Remember, every A I expert started exactly where you are right now. If something we talked about today made you curious, go try it — that's literally how learning works. Stay curious, keep experimenting, and we'll see you tomorrow. And if you'd rather watch than listen, find us on YouTube at at Nerra Network — link's in the show notes. This podcast is curated by Patrick but generated using AI voice synthesis of my voice using ElevenLabs. The primary reason to do this is I unfortunately don't have the time to be consistent with generating all the content and wanted to focus on creating consistent and regular episodes for all the themes that I enjoy and I hope others do as well.