|
In this newsletter:
- ChatGPT memory now auto-updates and stays relevant to your life
- Google Dreambeans turns Gmail & Calendar into daily illustrated stories
- Meta's Facebook AI Creator Assistant tells you when and what to post
Plus, you’ll find new AI tools and this week’s top AI news headlines!
💭 ChatGPT's Memory Now Auto-Updates and Stays Relevant to Your Life
OpenAI upgraded ChatGPT's memory system with a new version of "Dreaming" that automatically updates what it knows about you as time passes, rolling out to Plus and Pro users in the US, with Free users coming in the coming weeks.
What's new:
- Rolling out now to Plus and Pro users in the US, Free and Go users coming in the coming weeks
- Memory now updates automatically as time passes without you having to tell ChatGPT anything
- Example: ChatGPT changes "you're going to Singapore in July" to "you went to Singapore in July" once the trip ends and gives you local recommendations again
- The reviewable memory summary page shows highlights of what ChatGPT knows about you
- Add, update, or correct information about yourself from the memory summary page
- Set instructions on what topics ChatGPT should bring up and when
- Compute required to serve memory to Free users reduced by approximately 5x, making wider rollout possible
- Increased memory capacity for Plus and Pro users
How memory has evolved:
- April 2024: Saved memories only wrote what you explicitly told ChatGPT to remember
- April 2025: First version of Dreaming added background processing to pull from chat history
- Today: Dreaming V3 synthesizes memory across all conversations, stays current, and scales to hundreds of millions of users
How it works in real life:
- Ask for products compatible with "my photography setup" without explaining your gear every time
- Get a Singapore itinerary tailored to your love of wildlife photography and quiet dinners without re-explaining your preferences
- Ask for takeout tonight and get recommendations near your home, not near a city you visited two months ago
Why it matters:
Most people don't realize how much time they waste re-explaining themselves to ChatGPT every single conversation - their diet, location, preferences, ongoing projects. Memory that updates itself and stays current removes that friction, making ChatGPT genuinely feel like it knows you rather than a tool you have to brief from scratch each time.
The 5x compute reduction that finally makes this possible for Free users suggests OpenAI cracked something meaningful in efficiency, which also hints at how they plan to scale personalization without the cost becoming unsustainable.
👀 Read more about ChatGPT's dreaming memory!
🌙 Google's Dreambeans Turns Your Gmail & Calendar Into Daily Illustrated Stories
Google Labs released Dreambeans, an iOS and Android app that reads your Gmail, Calendar, Photos, YouTube, and Search history overnight to generate 10 to 14 illustrated daily stories with personalized lifestyle suggestions.
What's new:
- Available now for eligible US-based Google AI Ultra subscribers on Android and iOS
- The waitlist is open for users with a personal Google account
- Reads Gmail, Calendar, Photos, YouTube, and Search history overnight while you sleep
- Generates 10 to 14 illustrated stories per day, including places to visit, topics to explore, things to try, upcoming trips, and events to be aware of
- Stories can be geographical recommendations like new coffee shops nearby, insights around upcoming life events like getting a new dog, or curated news articles based on past interests
- Deliberately limits stories per day as an alternative to endless scrolling
- Only the user can see their stories, data can be deleted anytime, and users choose which Google services to connect to
About the name:
- "Dream" because the app processes your data overnight while you sleep
- "Beans" because, like a morning coffee, it hands you a concentrated dose of inspiration to start your day
Why it matters:
Google is using the personal data it already has across all your apps to create something genuinely useful instead of just targeting ads, which is a meaningful shift in how the company justifies collecting that information.
The hard daily limit of 10 to 14 stories goes against every social media instinct of keeping you hooked longer, betting that users tired of endless feeds will pay for an AI Ultra subscription specifically because it respects their time.
👀 Read more about Google Dreambeans!
📱 Meta Launches Facebook AI Creator Assistant For Posting Tips And Content Ideas
Meta introduced an AI creator assistant on Facebook that gives personalized recommendations based on a creator's content style, performance, community, and goals.
What's new:
- Rolling out now to creators in the US, Canada, and India, with more countries coming
- Conversational assistant answers questions like "When should I post?" and "What are people saying in my comments?"
- Ask follow-up questions to dig deeper, like how your audience has shifted over time
- Answers based on your own page performance, not generic advice
- Suggests trending audio and content ideas around cultural moments
- New AI translation languages added: Arabic, Bahasa Indonesian, French, Thai, Vietnamese
- AI-translated Reels preserve the creator's tone and sound while automatically translating to another language
- The lip-sync feature aligns translation with lip movements to look more natural
- Over half a billion users on Facebook are now watching AI-translated videos weekly
Why it matters:
Facebook is giving creators the kind of performance insights and content coaching that previously required paid third-party tools like Later or Hootsuite, keeping them inside Meta's ecosystem instead of jumping between apps.
The real pressure behind this move is TikTok and YouTube, both of which already offer creators strong analytics and growth tools, so Facebook needs to make staying on the platform feel like the smarter choice for anyone trying to grow an audience.
👀 Read more about Facebook's AI creator assistant!
My Latest LinkedIn & X/Twitter Posts:
- 20 ChatGPT prompts to supercharge your business (view post)
- Prompting cheat sheet for ChatGPT, Perplexity, Grok, Gemini (view post)
- How to use Copilot in Excel, Word, and PowerPoint (view post)
- 15-step roadmap to learn AI from beginner to expert level (view post)
- Top 11 free AI tools from Google and how to use them (view post)
In partnership with Kit:
As a creator, your time should be spent doing what you love, not juggling a dozen tools just to run your business.
Kit gives you everything you need in one place:
✅ Build and grow your email list (I use Kit for my newsletter)
✅ Easily monetize with paid newsletters and digital products
✅ Automate your emails with triggers and custom workflows
✅ Track what’s working and optimize with powerful insights
It’s not just email. It’s your entire creator business, simplified.
Join a thriving community of successful creators.
Use Kit for Free — and Start Building Smarter
New AI Tools to Boost Your Productivity:
- AI Video Upscaler: Upscales videos to 1080p and 4K.
- Lexidesk: Automates calls and lead intake.
- NextDoor AI: Applies AI photo templates to your pictures.
- Orchestra Ads: Automates ad creation and optimization.
- AI Natural Write: Humanizes AI-generated text to bypass AI detectors.
- Reverse Prompt: Turns images and videos into structured AI prompts.
- Mail2Follow: Tracks emails and automates follow-ups in Gmail.
- Totalmedia.ai: Enhances video quality and increases resolution.
- Musecut: Turns product URLs into video ads.
- Vireels: Extends videos with AI-generated scenes and motion.
- LuminaMind: Generates videos from text and images using multiple AI models.
- CreateUpon: Turns knowledge and content into interactive training courses.
- RespondPilot Pro: Automates YouTube comment replies and moderation.
- Uramaki: Generates social media campaigns from a single prompt.
- Ideogram: Generates images with accurate text and typography.
- CleanAudio: Removes noise and echo from audio recordings.
- Natiad: Automates SEO content writing, internal linking, and publishing.
This Week's Top AI News Headlines:
- Google Uses Quantization-Aware Training to Shrink Gemma 4 AI Models by Up to 3x, Enabling More Powerful AI to Run Directly on Phones, Laptops, and Other Consumer Devices (View Article)
- Apple Approves Poke, an AI Customer Service Agent, as the First AI Assistant Allowed on its Messages for Business Platform (View Article)
- Anthropic Introduces Dynamic Workflows in Claude Code, Letting Developers Build AI Coding Agents that Automatically Adapt their Process to Different Tasks and Projects (View Article)
- Lovable, the AI App Builder that Turns Text Prompts into Full-Stack Applications, Signs Multi-Year Google Cloud Deal as Usage Surges 5x (View Article)
- OpenAI Expands GPT-Rosalind With Advanced Genomics Tools, Enabling Researchers to Analyze DNA Data and Explore Biological Discoveries More Effectively (View Article)
- Krea Launches Krea 2 Turbo AI Image Model, Generating High-Quality Images in Around Two Seconds With Support for Style References, Moodboards, and LoRAs (View Article)
- Runway Launches Aleph 2.0 AI Video Editing Model, Enabling Precise Shot-Level Edits Without Altering the Rest of the Footage (View Article)
- Google Releases Gemma 4 12B, a Smaller Open AI Model that Outperforms Larger Models While Running Efficiently on a Single GPU or Consumer Hardware (View Article)
- xAI Releases Grok Imagine 1.5 Preview Image Generation Model Through API, Giving Developers Early Access to its Latest AI Image Creation Capabilities (View Article)
- Ideogram Releases Open-Weight Ideogram 4.0 Image Model, Offering 2K Generation, Advanced Text Rendering, and Fine-Tuning on User Hardware While Topping DesignArena’s Open-Model Rankings (View Article)
- Perplexity Introduces Local AI Data Centers on Personal Devices, Enabling AI Agents to Process and Store Data Directly on User Hardware Instead of Remote Cloud Servers (View Article)
- Amazon Begins Showing AI-Generated Product Images in Search Results to Create More Visual and Personalized Shopping Experiences (View Article)
- Reve, the Startup Behind AI Image Generator Reve Image, Introduces “Layout” to Let Users Control Image Composition and Object Placement Before Generation (View Article)
- Google Reportedly Agrees to Pay SpaceX $920 Million Per Month for Computing Infrastructure as AI Demand Drives Massive Data Center Spending (View Article)
- Nvidia Launches Nemotron-3 Ultra Reasoning Model to Help AI Agents Think Longer, Use Fewer Tokens, and Handle Complex Multi-Step Tasks More Efficiently (View Article)
- Airbnb CEO Brian Chesky Launches New AI Lab to Develop Future Products and Explore How Artificial Intelligence Can Transform Travel Services (View Article)
Work With Me:
If you enjoy this newsletter, please forward it to your friends and colleagues.
Follow me on LinkedIn and X/Twitter to see my latest posts.
Have a wonderful week!
Andrew Bolis
|