The Big Story

OpenAI held its yearly DevDay event and showed off a new product called Dots. Dots is an AI agent that looks like a cute, colorful blob with eyes and can do multi-step tasks like building simple virtual worlds inside the chat. Think of it like having a tiny digital teammate that remembers what you asked earlier and keeps working on the project even if you step away for a snack. Unlike regular chatbots that answer one question and stop, Dots can chain actions together, like first sketching a scene, then adding characters, then changing the lighting when you say so. This matters because most of us have used ChatGPT or similar tools for homework or creative ideas, and agents like this could soon handle longer creative projects without needing constant reminders. For students, it could mean turning a quick idea into a full interactive story or game prototype in one session instead of many back-and-forth messages. For anyone curious about future jobs, seeing how these agents are built now gives a real peek into what “AI teammate” roles might look like in design, game dev, or content creation. Right now Dots is only available to paid ChatGPT Pro, Business, or Enterprise users, so if you have access through school or a family plan you can try the agent features by starting a new chat and asking it to build something step by step, like “create a simple forest scene and then add a moving character.” If you don’t have the paid plan yet, watch the free DevDay recap videos on YouTube to see exactly how it works. Source: theverge.com

Explain Like I'm 14

You know how when you’re playing a video game, your character can remember where you left off and keep exploring the map even while you’re in another room grabbing a drink? Now picture that same idea but with an AI instead of a game character. The AI keeps a running list of everything you’ve said so far, breaks your big request into smaller steps, and checks off each one without forgetting the earlier ones. It can also hand parts of the job to other helper AIs that specialize in different skills, like one that draws pictures and another that writes text. When the helpers finish their pieces, the main AI puts everything together and shows you the result. That whole process is what people mean when they talk about AI agents working on longer tasks instead of just answering one question at a time. The surprising part is that the AI isn’t actually “thinking” the way we do; it’s following patterns it learned from millions of examples of people solving problems step by step. Once you see it that way, the magic feels more like a really well-organized to-do list that never gets bored or distracted.

Cool Stuff & Try This

Google’s camera that reads the fine print for you Google added a feature called Guided Vision inside the Gemini Live app on Android phones. Point your camera at anything with tiny text—like a medicine label, a menu in another language, or a complicated instruction sheet—and the AI describes it out loud in real time. It’s useful when you’re in a rush or the print is too small to read comfortably. Anyone with an Android phone can try it today by opening Gemini Live, tapping the camera icon, and pointing at text you want explained. Ask it to “read this out loud and tell me what it means in simple words” for an instant helpful voice description. You may need a parent’s help if the app asks for an account sign-up. Source: theverge.com

AI that makes school slides actually look good People who care about their presentations are now using AI to create interesting layouts, add funny visual jokes, and build clear diagrams instead of sticking with boring default templates. The AI suggests fresh designs based on what you want to say, so your slides feel more like something you’d actually enjoy watching. If you have a project coming up, open any AI chat tool you already use and type “turn these bullet points into a fun 5-slide presentation with simple diagrams.” Then copy the ideas into PowerPoint or Google Slides and tweak them. It takes the same amount of time as the old way but the results look much more polished. Source: x.com

Quick Bits

Low-poly car commercial made with AI video tools One creator used simple 3D shapes and AI video tools to make a full car commercial for almost nothing compared with filming it in real life. The trick is mixing a few high-quality pieces with lots of cheap low-detail ones so the final video still looks polished. Source: x.com

First AI model to pass a video Turing test A new model called Griffin fooled almost half the people who watched its videos into thinking they were talking to a real human. The conversations felt natural and flowed both ways without awkward pauses. Source: x.com

💬 Reply to this email — Patrick reads every one.

Share: X · LinkedIn · WhatsApp

Forwarded this email? Subscribe here — it's free.