- AI PlanetX
- Posts
- Google AI Browses Web Like Humans
In partnership with
Welcome to another edition of AI PlanetX.
Google’s AI learns to browse the web; Samsung’s mini model tops intelligence tests; SoftBank invests $5.4B in robotics and “Physical AI.”
Inside This Edition: 💎
Hottest AI News
Top AI & SaaS Tools
AI Tutorial: Turn Spreadsheet into Smart Data Analyst
Top AI & Tech News
AI Art Spotlight
Prompt of the Day: Sora 2 Prompting Guide by OpenAI
AI Video Tutorial
Course of the Day: Building AI - Part II by Elements of AI
Hottest AI News
DeepMind
Google’s Latest AI Browses the Web Just Like You Do
Google just launched Gemini 2.5 Computer Use — an AI model that can browse like a human by clicking, scrolling, and typing directly in a web window. It lets AI interact with any website without needing special permissions or APIs.
Details:
The model uses visual understanding to perform tasks like filling forms and navigating interfaces. It powers AI Mode and Project Mariner, which can complete web actions like adding shopping items
The launch came a day after OpenAI’s Dev Day, fueling the AI agent race. While OpenAI’s ChatGPT Agents and Anthropic’s Claude focus on full computer use, Google’s version is browser-based. It supports 13 actions such as opening tabs, typing, and drag-and-drop for web automation
Developers can use it via Google AI Studio and Vertex AI. Demos on Browserbase show it playing 2048 or browsing Hacker News. The videos run 3× quicker for effect, and Google says desktop control isn’t ready yet
This browser-native approach could revolutionize web automation and testing, making AI assistance available on any website regardless of whether it offers API integration.
Smart dictation that understands you
Typeless turns your raw, unfiltered voice into beautifully polished writing - in real time.
It works like magic, feels like cheating, and allows your thoughts to flow more freely than ever before.
With Typeless, you become more creative. More inspired. And more in-tune with your own ideas.
Your voice is your strength. Typeless turns it into a superpower.
Samsung
Samsung's Mini AI Model Demolishes Industry Giants on Intelligence Tests
Samsung just flipped the AI playbook. While others chase trillion-parameter giants, its team proved a 7M-parameter model can beat models thousands of times larger on tough reasoning tasks. It’s not a minor upgrade—it’s a full reset of what we thought scale meant for intelligence.
Details:
The old strategy of endlessly scaling language models has hit a wall in reasoning tasks. These massive models predict words one by one, so an early mistake can ruin the entire answer. Even methods like Chain-of-Thought need huge compute and massive datasets that often don’t exist
Samsung’s Tiny Recursive Model (TRM) flips that logic. It’s a small network that refines its reasoning up to 16 times in one run. Surprisingly, a two-layer version outperformed a four-layer one by avoiding overfitting. By back-propagating through the full recursion instead of using complex math steps, TRM’s accuracy on Sudoku-Extreme jumped from 56.5% to 87.4%
The results are stunning: 87.4% on Sudoku-Extreme with only 1,000 examples, 85.3% on maze navigation, and 44.6% on the ARC-AGI test. Its 7M-parameter model even beats Google’s Gemini 2.5 Pro (4.9% on ARC-AGI-2) while using a fraction of the resources
This research represents more than just a technical breakthrough – it's a fundamental challenge to the current trajectory of AI development.
Top AI & SaaS Tools
youbooks (Life-time Deal): Helps writers turn research and their authentic writing style into publish-ready nonfiction books — up to 300,000 words — quickly, without years of effort
Sorce App: AI-driven job-matching app that works like Tinder: candidates swipe right to apply and left to skip [F-R-E-E]
Hunyuan-Vision-1.5: Advanced vision–language model, currently ranked No. 3 on @ LMArena [F-R-E-E]
NeuTTS Air: Next-gen, fully open-source on-device text-to-speech with cloud-level quality, sized for mobile and edge [F-R-E-E]
Imagine v0.9: xAI released a major upgrade to its video generation model, with better visuals, motion, audio, and overall quality over v0.1 [F-R-E-E]
AI You’ll Actually Understand
Cut through the noise. The AI Report makes AI clear, practical, and useful—without needing a technical background.
Join 400,000+ professionals mastering AI in minutes a day.
Stay informed. Stay ahead.
No fluff—just results.
AI Tutorial
How to Set Up Your Personal AI Data Analyst in Julius AI
Most people stare at their spreadsheets and wonder what story the numbers are hiding. With Julius AI, you can upload your dataset and get answers in seconds — charts, insights, and summaries from plain-English questions.
Visit Julius AI
Go to Julius AI and sign in securely using your account credentials.
Start Chat
Tap on New and give it a precise, memorable title, for example: Sales Q3 drivers analysis.
Upload Dataset
Attach your CSV or Excel file for analysis.
If there are multiple sheets, explicitly choose the correct tab.
If your CSV uses a custom delimiter or day-first dates, mention it on upload.
Verify Ingest
Ask: Show row count, nulls by column, and inferred data types for a quick sanity check.
If needed, cast Revenue to numeric, parse Order Date as a date, and drop duplicates by Order ID.
Define Scope and Rules
State the goal and metric rules in one clear line, for example:
Analyze monthly revenue for the last 12 months; revenue equals unit_price × quantity; exclude rows where status equals Refunded.Ask Focused First Question
Example: Return revenue by calendar month with month-over-month percent change and a line chart from 2024-10-01 to 2025-09-30, inclusive.
Shape Output
Specify usability details so results are useful, for example:
format currency, sort months ascending, add a 3-month moving average, and include a concise table with month, revenue, mom_change_pct, rolling_3m_avg plus a one-paragraph insight.Validate and Export
Cross-check the grand total versus the sum of monthly totals and ask Julius to show the exact steps or query used; then download the chart as PNG, export the tables as CSV, and save the chat for future follow-ups.
Note: Always clean your dataset before uploading. Simple fixes like consistent column names and correct date formats prevent errors later.
Top AI & Tech News
SoftBank is buying ABB’s robotics division for $5.4 billion (pending regulator approval) to deepen its push into robotics and “Physical AI”
Figure AI will unveil its Figure 03 humanoid on Thursday, teasing a knitwear-clad home robot powered by Helix AI for learning and household chores
Google has expanded Opal, its AI-powered "vibe-coding" app for creating mini web apps from text prompts, to 15 more countries
IBM teamed with Anthropic to bring Claude LLMs to its enterprise tools, starting with an AI-first IDE
AI Art Spotlight
LEFT: Original photo | RIGHT: Generated by Nano Banana
Model: Gemini 2.5 Flash Image (Nano Banana)
Prompt:
Make this a studio portrait. Adjust the skin tone and apply a cinematic black-and-white grade. The man is wearing a black shirt and sitting on the back of a chair. Change the camera angle from eye level. Use a black background. The photo should look realistic in 8K with no pixel loss.Prompt of the Day
Sora 2 Prompting Guide by OpenAI
The Sora 2 Prompting Guide shows creators how to write precise, cinematic prompts to generate high-quality videos with Sora 2. It explains how to control motion, lighting, and style, use image references, and iterate effectively—helping filmmakers and AI creators turn ideas into visually consistent, professional-grade videos.
Top AI Video Tutorial
47 F-R-E-E Things You Can Do with Gemini
Complimentary AI Course of the Day
Building AI - Part II
Building AI, the second part of the Elements of AI series, is a course designed to help learners build their own AI models and applications. The course is flexible, allowing participants to choose among three difficulty levels—from multiple-choice exercises for non-programmers to Python-based projects for those with programming experience.