|
In this newsletter:
- Google launches Gemini desktop app for AI help anywhere
- Meta launches Muse AI agent that shops, emails, and plans trips for you
- Apple Watch can now turn conversations into notes and transcripts
Plus, you’ll find new AI tools and this week’s top AI news headlines!
In partnership with 1440:
Get a daily digest covering top stories across business, science and world news...
💻 Google Launches Gemini Desktop App That Assists You Anywhere on Your Computer
Google released Gemini for desktop on Mac and Windows, giving users a keyboard shortcut to summon AI assistance from any app, alongside Gemini Spark for autonomous background tasks and creative tools for images and video.
What's new:
- Available now for download on macOS and Windows
- Press Alt + Space on Windows or Option + Space on Mac to summon Gemini over any active task
- Ask questions, verify facts, or draft text without leaving what you're working on
- Select features available to users over 18 with a Google AI subscription
What it can do:
- Pull information from Gmail, Google Drive, Docs, and Calendar directly into your desktop workflow
- Find details buried in email threads, summarize shared documents, check meeting notes against other meeting notes
- Share what's on your screen, and Gemini uses context to debug errors, explain charts, or understand what you're working on
- Speak your thoughts into any application and Gemini automatically formats and polishes your dictation in real time
- Generate custom images with Nano Banana and high-quality videos with Gemini Omni from a dedicated creative workspace
Gemini Spark integration:
- Works as your 24/7 personal AI agent for multi-step research, planning, and information integration
- Runs autonomously in the background under your direction
- Cleans up folders or summarizes local files into Google Docs and Sheets
Why it matters:
Every major AI company has a browser-based assistant, but Google is the only one that already lives inside the apps people actually use daily: Gmail, Drive, Docs, Calendar.
Bringing that context awareness to a desktop app means Gemini knows your work history before you even ask a question, which is an advantage no standalone AI assistant can replicate without years of your data behind it.
👀 Read more about Gemini for desktop!
🤖 Meta's New Muse App Handles Your Shopping, Emails, and Travel Plans for You
Meta launched Muse, its first standalone AI agent app designed to complete real tasks for you rather than just answer questions.
What Muse is:
- A personal AI agent that runs on a dedicated, secure virtual machine in Meta's cloud
- Powered by Meta's Muse Spark 1.3 model
- Available on iOS, Android, web at muse. ai, and WhatsApp
- Currently available only to adults 18 and older in the US
- AI glasses support reportedly in the pipeline
What it can do:
- Sort through your Gmail inbox and delete emails you don't need
- Shop on Amazon based on your size, style, and color preferences, and complete the purchase on your behalf
- Book travel, manage calendars, find and cancel forgotten subscriptions
- Generate AI podcasts, images, videos, and interactive documents called artifacts
- Create a personalized news feed based on your interests
- Maintains a running list of goals and personalized suggestions based on your conversations
How it stays in control:
- Review and approve critical actions like sending emails and making purchases before they happen
- Complete audit trail of everything your agent has done and is planning to do
- Logins kept in a secure credential store the agent cannot read, with 1Password integration coming soon
- Generates a one-time card number at checkout so your real card is never seen by the merchant or agent
- Eligible purchases covered by Link's purchase protections
- Conversations not shared with Meta's ad systems
- Connect your email, calendar, Instagram, and other apps
The privacy concern:
- Muse pulled very specific personal interests from Instagram's API data, including anime, CrossFit, Labrador retrievers, Florida wildlife, and 90s nostalgia that aren't even visible to users in their own Instagram settings
- Meta says it only exchanges data with third-party apps that are needed for them to work, and does not share information with advertisers
How it's performing:
- Reached No. 2 on the US App Store within 48 hours of its September 8 launch with over 83,000 iOS downloads
- Meta described it as a slower start compared to Threads (4.3 million downloads on day one) and Meta AI (108,000 downloads at launch)
- Ranked No. 338 in Productivity on Google Play, significantly lower than iOS performance
Why it matters:
Meta entering the AI agent space with a tool that actually completes tasks rather than just answering questions is a significant shift for a company that built its empire on ad targeting.
The one-time card numbers, purchase protections, and no-ad-sharing promise are clearly designed to address the trust problem head-on, but the bigger question is whether those safeguards are enough to convince users who already handed their data to Meta through Instagram and Facebook for years, and got burned, to now hand over their bank accounts and inboxes too.
👀 Read more about Meta's Muse AI agent!
⌚ Apple Watch's New AI Features Can Transcribe Conversations & Take Meeting Notes Automatically
Apple introduced three AI audio features to its new Apple Watch lineup that process live audio, transcribe conversations, and summarize meetings directly on the device.
What's new:
- Audio Intelligence: alerts wearers to sounds like sirens, alarms, doorbells, babies crying, and more using on-device AI; works even without an iPhone nearby
- Live Rewind: double-press the Digital Crown to go back 15 seconds and transcribe what was just said; transcription saves to the Siri app for later
- Siri Recap: uses ambient listening to take high-level notes of conversations, automatically generates a title, summary, and key points, not exact transcripts
- Apple does not create or store audio recordings; raw audio is not accessible even to Apple
- Does not identify or attribute speakers
- Generated transcripts and recaps are protected with end-to-end encryption
- Siri Recap is not always-on by default; users choose when it's enabled (only at work, only during the day), and it can be turned off anytime from Control Center
The concern:
- Live Rewind lets you transcribe something anyone said without them knowing, triggered by a subtle double-press rather than holding up a phone
- Apple adds an audible chime and full-screen microphone animation when recording
- Raises consent questions both ethically and legally, as recording laws vary by state
- Text-only transcripts with no audio may complicate legal use as evidence
Why it matters:
Apple has spent years positioning itself as the privacy-first tech company while quietly expanding what its devices can do with your data. These three features mark the most significant shift yet because a watch is something people forget they are wearing, which makes ambient transcription fundamentally different from holding up a phone to record.
Apple's encryption and no-storage promises address the legal risk but not the social one. People already changed how they behave knowing cameras are everywhere. Adding always-available transcription to a device worn on millions of wrists raises the same question all over again, just in a smaller package that is far easier to ignore.
👀 Read more about Apple Watch's new AI audio features!
My Latest LinkedIn & X/Twitter Posts:
- 9 tips for writing expert-level ChatGPT prompts (view post)
- How to go from basic to expert-level prompting (view post)
- 10 ChatGPT prompt templates to get better results faster (view post)
- 15 ChatGPT prompts to help you structure research (view post)
- Step-by-step guide for using Perplexity AI (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:
- Tables: Finds, qualifies, and enriches B2B leads.
- Widgo: Qualifies website visitors and books demos.
- Zoomaa AI: Automatically sends lead magnets to post commenters.
- SchedulifyX: Creates, schedules, and publishes social media content.
- Intepra.com: AI social media post generator.
- Jupitrr AI: Creates and repurposes videos for social media.
- AI Toolbox: Organizes, searches, and exports your AI chats.
- Tucky: Combines desktop notes with an AI assistant.
- Articos: Tests marketing ideas with simulated customers.
- Schedulenchill: Schedules LinkedIn posts drafted by your existing AI tool.
- Typewise: Automates customer service and resolves requests.
- LiftmyCV: Finds jobs and automatically applies for you.
- Tadata: Automates repetitive work from Slack.
- Nex: Automates prospecting, outreach, and CRM tasks.
- ChatGPT Images 2.5: Creates and edits images from prompts and sketches.
- Databox: Analyzes performance data, creates reports, and recommends actions.
- Kombai: Provides ready-to-use website and app design inspiration.
This Week's Top AI News Headlines:
- OpenAI Launches the Agents API in Public Beta, Letting Developers Build and Run Long-Running Cloud AI Agents with the Same Codex Harness, Tool Use, and Multi-Agent Infrastructure Behind Codex (View Article)
- Google Labs Expands Dreambeans, a Personalized AI Story App, to All U.S. Adults on Android and iOS, Using Connected Google Apps to Generate Daily Stories Tailored to their Lives and Interests (View Article)
- OpenAI Launches ChatGPT for Financial Services, Combining GPT-6 Astra with Premium Financial Data and Tools for Investment Banking, Equity Research, Financial Modeling, and Earnings Analysis (View Article)
- Anthropic Finds its Most Advanced Claude Models can Perform Near-Superhuman Intelligence Targeting Tasks, Including Photo Geolocation and Linking Anonymous Social Media Accounts, Prompting New Safeguards Against Misuse (View Article)
- OpenAI Launches GPT-Live-1 in the API, a Full-Duplex Voice Model that can Listen and Speak Simultaneously to Enable Faster, More Natural AI Voice Conversations (View Article)
- ElevenLabs and Universal Music Group Partner on a Licensed AI Music Platform that will Let Fans Create Remixes, Mashups, and New Interpretations of Participating Artists’ Music (View Article)
- OpenAI Launches a Data Agent in ChatGPT Work, Letting Businesses Analyze Company Data, Build Interactive Dashboards, and Take Actions Across Data Sources Using Natural Language (View Article)
- DeepSeek Launches V4.1-Flash, a 552-Billion-Parameter Multimodal AI Model Designed to Deliver Better Performance, Faster Speeds, and Lower API Costs than its Previous V4-Pro Model (View Article)
- OpenAI Expands ChatGPT Access for U.S. Governments Through a Multi-Year Agreement Offering Free Licenses and Discounted AI Usage to Federal, State, Local, and Tribal Agencies (View Article)
- Anthropic Says it Blocked Attempts to Misuse Claude for Potential Biological Weapons Research, Cyberattacks, Surveillance, and Propaganda as More Capable AI Models Create New Security Risks (View Article)
- Moonshot AI, the Chinese Startup Behind the Kimi AI Models, is Reportedly Exploring Dual Listings in Hong Kong and Shanghai as it looks to Raise Billions from Public Markets (View Article)
- OpenAI Calls for Mandatory AI Safety Regulation and Backs Four California Bills Covering Independent Safety Assessments, AI Auditors, Youth Protections, and Biological Threat Safeguards (View Article)
- Positron, an AI Chip Startup Focused on High-Memory Processors for AI Inference, Raises $875 Million at a $5 Billion Valuation as Demand for Alternatives to Nvidia Accelerates (View Article)
- Andrew Tulloch, a Prominent AI Researcher Who Previously Worked on OpenAI Models and Co-Founded Thinking Machines Lab, Leaves Meta to Join Anthropic’s AI Training and Inference Team (View Article)
- Anthropic Says Alibaba, Moonshot, DeepSeek, Zhipu, and Xiaomi Generated Nearly 190 Million Claude Interactions in Attempts to Distill its AI Capabilities, with Some Requests Potentially Exposing Sensitive Government and Military Data (View Article)
- Enflame, a Tencent-Backed Chinese AI Chipmaker, Surges 179% in its Shanghai Market Debut After a $912 Million IPO as Investors Bet on Domestic Alternatives to Nvidia (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
|