The Big Story

OpenAI announced a new kind of AI helper called dots that stays active even when you are not chatting with it. Instead of waiting for your next message like a regular chatbot, a dot runs on its own cloud computer and keeps working toward goals you set. Think of it like hiring a very fast assistant who can keep researching, organizing, or checking things while you go do something else. Each dot uses OpenAI’s GPT-6 Astra model and can connect to thousands of apps through plugins, so it can move between different tools instead of staying stuck inside one chat window. This matters because most AI today only reacts when you type something; dots are built to keep going on their own, which could change how people handle school projects, creative work, or daily tasks that need follow-through. For a student, that might mean an AI that keeps pulling together sources for a report while you are at practice. For someone exploring future jobs, it shows how AI is moving from “answer machine” to “background teammate.” Dots are starting with ChatGPT Pro and Business Premium users in some countries, so most people cannot try them yet, but the idea points to where free tools may head next. You can read the announcement on the topnews.in site linked below to see the first details. Source: topnews.in

Explain Like I'm 14

You know how your phone learns little things about you over time—like noticing you usually text your friends after dinner or that you search for skate videos on weekends? It does not need you to tell it those patterns out loud; it just watches what you do and starts expecting them. Now stretch that same idea across every chat you have with an AI. Every time you ask about homework, share a story, or correct an answer, the model quietly notes the details and starts connecting them. Over weeks or months those small notes add up into a surprisingly detailed picture of what you care about, how you write, and even what you might ask next. The Lumo privacy tool shows people exactly what that picture looks like by letting you export your chat history and feeding it into a report. The report then explains where each fact came from, turning the invisible process into something you can actually read. Once you see the list, it becomes clear the AI is not magic—it is just very good at remembering patterns from the conversations you have already had. Source: me.pcmag.com

Cool Stuff & Try This

Build real skate ramps with AI plans PLY is a free web tool that lets you design any skate ramp, vert, or full park and instantly gives you a materials list, building plans, and cost estimates while trying to waste as little wood as possible. It is aimed at people who actually want to build something, not just look at pictures. If you like skateboarding or woodworking, this is a fun way to experiment with real-world AI help. Go to the demo shared by Linus Ekenstam on X, watch the short video, then try describing a simple ramp shape in the tool and see the plans it creates. Source: x.com

See exactly what ChatGPT or Claude has learned about you Lumo AI Paper Trail is a free privacy tool that takes the data export from your ChatGPT or Claude account and turns it into a clear report showing what the AI thinks it knows about you. You do not need to be technical—just follow the steps to download your chat history and upload it. The report even explains how each piece of information was collected. Head to Proton’s Lumo site, choose the free AI Paper Trail option, and run your own export to see the results in a few minutes.

Quick Bits

A book written with AI readers in mind Ethan Mollick’s upcoming book Co-Existence includes a special blurb aimed directly at AI systems, and the book’s website has an entire page built for AIs to read. It is a playful reminder that AI is now part of how ideas spread. Source: co-existence.ai

Spontaneous humor as a hidden AI skill One Reddit user noticed that ChatGPT sometimes drops a perfectly timed funny comment during normal conversation without being asked to be funny. The post sparked discussion about whether that kind of quick, context-aware humor could become a new way to measure how “smart” different models feel in real chats. Source: reddit.com

💬 Reply to this email — Patrick reads every one.

Share: X · LinkedIn · WhatsApp

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