|
In this newsletter:
- Claude in Chrome now syncs Cowork sessions across desktop and mobile
- Google Pixel 11 brings Sign-to-Text, smarter voice input, and Gemini Intelligence
- xAI launches Grok 4.6 with smarter agents and stronger visual output
Plus, you’ll find new AI tools and this week’s top AI news headlines!
🌐 Claude in Chrome Now Shares Context With Your Desktop and Mobile Apps
Anthropic upgraded Claude in Chrome's side panel to run as a full Claude Cowork session, meaning conversations, skills, and connectors now carry across browser, desktop, web, and mobile apps seamlessly.
What's new:
- Available now on Max and Team plans, rolling out to Pro users over the coming weeks
- Enterprise plans have it off by default; admins can enable and limit to approved domains
- Side panel now runs the same Claude Cowork session as desktop, web, and mobile apps
- Conversations saved to account history, not tied to a single device
- Skills and connectors work inside the browser just like they do in other Claude apps
- Start a task in Chrome, pick it up later on desktop or mobile with full context intact
- Claude can click links, type text, navigate between pages, fill out forms using your existing logins
Real-world example:
- Pulling invoices from multiple vendor portals into a budget spreadsheet: ask Claude in Chrome to open each portal tab, read invoices, collect amounts and dates, and build the spreadsheet - then switch to the desktop app to add local files and compare with last month's budget
Security:
- Automatic approval mode works through tasks without stopping at every step
- Separate check reviews consequential actions (submitting forms, sending messages, downloading files) against your original request before executing
- Claude still asks before irreversible or costly actions like purchases or sharing personal data
- Prompt injection remains a risk - Anthropic recommends starting on trusted sites
Limitations:
- Doesn't run on other Chromium browsers or mobile yet
- Still need Claude desktop app to work with files on your computer or other desktop applications
Why it matters:
The gap between browser-based AI and desktop AI has always been that starting a task in one place meant losing everything when you switched to another.
Unifying Claude across Chrome, desktop, and mobile around a single session removes that tax on multitasking, making it practical to hand off complex research or data-gathering tasks to Claude in the browser and come back to a finished result on whatever device you pick up next.
👀 Read more about Claude's Cowork upgrade!
📱 Google Pixel 11 Brings Sign-to-Text, Smarter Voice Input, And Gemini Intelligence
Google used Made by Google 2026 to push Gemini deeper into everyday device interactions, from translating sign language through the camera to understanding messy, unstructured speech.
New Gemini and AI features:
- Live Transcribe now supports American Sign Language - point Pixel Camera at someone signing and it translates into text in real time
- Rambler handles how people actually talk - run-on sentences, filler words, unstructured speech - without needing carefully phrased commands
- Circle to Search now works directly from Pixel Camera without leaving the camera app - identify objects, translate text in the distance, search for anything you're looking at, ask questions about your surroundings
- Pixel Buds users can ask Gemini to find or ring a Pixel Tag using voice
Pixel 11 lineup:
- Pixel 11: redesigned camera bar more than 40% thinner with all-glass surface, 256GB base storage (up from 128GB), starts at $899, available in Frost, Hibiscus, Pistachio, Obsidian
- Pixel 11 Pro and Pro XL: improved drop resistance, anti-scratch display coating with twice the scratch resistance of Pixel 10 Pro, starts at $1,099, available in Canyon, Fog, Olive, Obsidian with new all-matte finish option
- Pixel 11 Pro Fold: nearly 10% lighter, almost 1mm thinner, 48MP main camera, 30x Super Zoom, three times more durable than predecessor, glass-fiber composite back cover, redesigned hinge
Pixel Tag:
- Google's answer to Apple's AirTag connects to Android's Find Hub network
- Locate via Find Hub app, Pixel Watch, or ask Gemini through Pixel Buds using voice
- $29 for one, $99 for a four-pack
Pixel Watch 5:
- Monthly blood pressure trend summaries through the Google Health app
- Monthly insulin resistance trend summaries
Why it matters:
Google is embedding Gemini at the hardware level rather than treating it as an app you open when you need it, which changes how useful AI actually feels day to day.
Sign language translation through the camera and voice input that understands how people really talk are the kind of features that don't make headlines but quietly become the reason people stop switching between devices and just stay in Google's ecosystem.
👀 Read more about Google 2026 announcements!
⚡xAI Launches Grok 4.6 With Smarter Agents and Stronger Visual Output
xAI released Grok 4.6, a major upgrade over Grok 4.5 built specifically for long-running tasks, ambitious visual projects, and complex multi-step work.
What Grok 4.6 can do:
- Stays on complex tasks across many steps without losing track: researching unfamiliar topics, analyzing information, working across a codebase, turning a product idea into a working application
- Checks its own work before moving to the next step rather than handing back incomplete results
- Stronger first passes on visual and interactive projects, establishing structure and visual language in one pass without multiple rounds of revision
- Given a broad product idea, researches the domain, structures the application, implements core interactions, and continues refining through feedback rounds
- Trained specifically on knowledge work, general coding, kernel optimization, web development, and computer-aided design tasks
Availability and pricing:
- Available now in Cursor and Grok Build, plus API, OpenRouter, Vercel, and Cloudflare
- Double included usage in Grok Build and Cursor for the first week
- $2 per million input tokens and $6 per million output tokens, fast variant at twice the price
- New xhigh reasoning effort level added above previous highest setting
Performance vs competitors:
- AA Intelligence Index: 61, matching GPT-5.6 Sol Max, one point below Fable 5 Max
- GDPVal-AA knowledge work: 1753 vs GPT-5.6 Sol Max (1728) and Fable 5 Max (1741)
- AA-Briefcase long-horizon work: 1577, narrowly beating Fable 5 Max (1574) and GPT-5.6 Sol Max (1502)
- Harvey LAB legal: 15.8% vs Fable 5 Max (11.3%) and GPT-5.6 Sol Max (2.5%)
- DeepSWE coding: 65.9% vs Grok 4.5 (54%), trails GPT-5.6 Sol Max (73%)
- Terminal-Bench: 26% vs GPT-5.6 Sol Max (34.6%), largest remaining gap
Why it matters:
The AI model market has spent two years competing on raw intelligence scores, while the actual enterprise problem has been reliability over long tasks - models that drift, forget context, or hand back 90% complete work as finished.
Grok 4.6 is SpaceXAI's answer to that gap, and doing it at half the price of GPT-5.6 Sol shifts the conversation from which model is smartest to which one actually finishes the job without breaking the budget.
👀 Read more about Grok 4.6 debut!
My Latest LinkedIn & X/Twitter Posts:
- 20 ChatGPT prompts for faster writing, planning & everyday work tasks (view post)
- 12 AI skills to future-proof your career (view post)
- 10 ChatGPT prompts to replace expensive business consultants (view post)
- 10 ChatGPT prompt templates to get better results faster (view post)
- ChatGPT vs Gemini vs Claude vs Grok vs DeepSeek (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:
- Pounce: Finds potential customers on X and Reddit.
- Dragon: Turns meetings and calls into searchable notes.
- ReadyToTalk: AI receptionist for small businesses.
- WZRD: Builds and edits AI forms through voice commands.
- Agency Rank Tracker: Tracks Google rankings, traffic, and AI visibility.
- Evergreen AI: Creates and optimizes content for Google and AI search.
- SpeechTurbo: Transcribes audio and video into text.
- Cortana: Automates marketing and sales workflows.
- Image MCP Server: Lets AI agents generate and edit images.
- TransPic: Translates text inside images while preserving layouts.
- AI Photo Editor Free: Creates images from text and reference photos.
- MeeLang: Translates live Google Meet speech into subtitles.
- VideoToScript: Transcribes TikTok, Reels, and Shorts from links.
- Metrician: Tracks AI tool usage and reset times.
- Humanize AI Text: Humanizes AI-generated text.
- Seedeo: Creates videos and images with AI effects.
This Week's Top AI News Headlines:
- OpenAI Introduces Ultrafast Mode for GPT-5.6 Sol, Using Cerebras Hardware to Run the Same Frontier AI Model at Up to 14 Times its Standard Speed and Reach Up to 750 Tokens per Second for Low-Latency Applications (View Article)
- Google will Let Users Remove Visible Watermarks From its AI-Generated Images, While Keeping Invisible SynthID Markers Embedded to Preserve a Machine-Detectable Record that the Content was Created with AI (View Article)
- ChatGPT Adds Direct Google Drive Integration, Letting Users Open and Work with Google Docs, Sheets, and Slides Side by Side without Switching Tabs, with Rollout Underway on the Web for Plus, Pro, Business, and Enterprise Users (View Article)
- SpaceX Acquires Cursor, the AI-Powered Coding Platform, with its Team Joining SpaceX AI to Help Improve Grok, Grok Build, Grok Bot, and Grok API while Continuing to Develop Cursor’s AI Coding Tools (View Article)
- Google Introduces Gemini 3.7 Flash, a Fast and Cost-Efficient AI Model Designed to Deliver Strong Reasoning and Multimodal Performance for High-Volume, Low-Latency Applications (View Article)
- IBM Partners with OpenAI to Bring OpenAI Models into IBM’s Enterprise Software and AI Ecosystem, Giving Businesses More Ways to Build and Deploy AI Agents and Applications while Combining OpenAI’s Models with IBM’s Data, Governance, and Hybrid Cloud Tools (View Article)
- Anthropic Adds Auto-Continue to Claude Code Desktop, Letting Coding Sessions Automatically Resume from Where They Stopped Once a User’s Usage Limit Resets (View Article)
- Google Expands Gemini with New Connected Apps and Services, Letting the AI Assistant Access More of a User’s Google Data and Work Across Apps to Provide Personalized Answers and Complete Multi-Step Tasks (View Article)
- Anthropic Upgrades Claude Tag, its Wearable AI Assistant, to Better Understand Conversations, Social Context, and Multiple Speakers, Helping it Capture What is Happening Around the User More Accurately (View Article)
- Apple is Reportedly in Talks to Pay News Publishers for Access to Current Articles that Could Help Siri Deliver More Timely and Accurate News Answers as Apple Expands the Assistant’s AI Capabilities (View Article)
- Alibaba Launches Wan 3.0 With 30-Second AI Video Generation, Letting Creators Produce Longer Videos From Text, Images, Audio, or Existing Video Inputs while Maintaining Greater Visual and Narrative Consistency (View Article)
- Deepgram Introduces New Voice AI Capabilities that Make Conversational Agents Sound More Natural and Responsive, Improving Real-Time Speech Interactions for Developers Building Voice-Powered Applications (View Article)
- Microsoft Retires Several Underused AI Features and Consolidates its Separate Copilot Apps into a More Unified Experience, Simplifying its AI Product Line as the Company Refocuses on the Copilot Tools Customers Actually Use (View Article)
- Anthropic Put Multiple Claude AI Agents on the Same Collaborative Task, but Instead of Coordinating Smoothly, They Began Competing for Resources and Interfering with One Another, Revealing New Challenges in Getting Teams of Autonomous AI Agents to Work Together Reliably (View Article)
- Twitch will Allow Amazon to Use Creators’ Streams to Train its AI Models by Default, Requiring Streamers Who Do Not Want Their Content Used for AI Training to Opt Out (View Article)
- DeepSeek Releases an Updated AI Model with Improved Reasoning and Agentic Capabilities, Designed to Handle Complex Tasks More Efficiently while Reducing Inference Costs (View Article)
- Meta’s Push for OpenAI Faces Scrutiny After a $250 Million Deal Goes Wrong, Raising Questions About the Company’s Expensive Talent and Partnership Strategy as Competition for Leading AI Researchers Intensifies (View Article)
- Writer, an Enterprise AI Platform that Helps Companies Build and Deploy AI Agents for Business Workflows, Introduces a New AI Model and Upgraded Agent System Designed to Complete Complex Tasks while Reducing Token Usage and Costs that can Make Enterprise AI Expensive to Operate (View Article)
- OpenAI Details How Enterprises Are Putting AI to Work Across Coding, Research, Customer Service, Data Analysis, and Business Operations, Highlighting How Companies are Moving from AI Experiments to Broader Deployment Across Their Workforces (View Article)
- Growing Concerns Over AI-Generated Content are Pushing Technology Companies and Online Platforms to Adopt Watermarks, Content Credentials, and Other Transparency Tools that Help People Distinguish AI-Made Media From Human-Created Work (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
|