AIPOOOL Weekly AI Digest – 29 June 2025

Happy Sunday! This is AIPOOOL. The email that tells you what’s going on in Artificial Intelligence space in simple blocks. Get ready to have your mind blown by the sheer power of AI!

In Today’s Email :

• 🔥AI News Flash: OpenAI taps Google’s TPUs for ChatGPT, Salesforce’s AI sales surge, Hollywood-backed Studio bets big on AI video tools, Gemini boosts Google Photos speed, Nvidia rides an AI-fueled stock high, and DeepMind puts Gemini brains into on-device robots!

• ⛏️ Trending Tools: dbrief for Interview Automation, Scottie for faster news & many more …

• 🔰 Quick Grab: Agent S2: Open‑Source AI That Finally Uses Your Computer—Like a Human!

• 🎆Creators Corner: Top Picks from Hugging Face: Trending AI Applications You Can't Miss!

• 🥼 From Lab to Layman: Introducing Gemini CLI: AI in Your Terminal

Browse AI Tools | Instagram | Advertise

AI Happenings You Don’t Want To Miss

✨ OpenAI Leans on Google’s TPUs to Power ChatGPT
OpenAI is now renting Google’s tensor processing units (TPUs) to run ChatGPT and its services—marking a shift from reliance on NVIDIA chips and easing operational costs.

✨ Salesforce CEO Points to AI as Key Profit Driver
Salesforce’s Marc Benioff states that AI sales are becoming the company’s fastest-growing revenue stream, with adoption soaring across business tools.

✨ Studio Launches AI Startup Backed by Film Industry Execs
A new AI startup named Studio, supported by top film execs, plans to revolutionize video creation with tools that generate trailers, storyboards, and casting recommendations.

✨ Google Speeds Up “Ask Photos” with Gemini Power
Google Photos’ AI-powered “Ask Photos” feature is rolling out again—with faster response times for simple queries thanks to Gemini enhancements.

✨ Nvidia at Record High on AI Boom Expectation
Nvidia shares hit record levels as analysts predict an AI “golden wave,” fueled by surging demand for its chips in AI infrastructure.

✨ DeepMind Brings Gemini AI to On‑Device Robots
Google DeepMind is testing Gemini AI to power robots locally, without cloud computing—an advancement toward smarter, autonomous machines.

Free & Useful AI Tools -

1. dbrief- Automate expert interviews for content at scale.

2. TheLibrarian.io - Supercharge your day with a WhatsApp AI Assistant.

3. Scottie - Read the news 30x faster.

4. Kolena AI - Unlock faster financial workflows with AI automation.

🧠 Agent S2: Open‑Source AI That Finally Uses Your Computer—Like a Human!

Simular AI has unveiled Agent S2, a groundbreaking open‑source framework that allows AI to perform real-world tasks directly on your machine—automating GUI actions on Mac, Windows, Linux, even Android.

Built like a mini‑team of models, Agent S2 blends a powerful planner with specialist modules for precise GUI interaction, memory, and knowledge retrieval. Thanks to its Proactive Hierarchical Planning, it breaks tasks into manageable steps and continually refines its approach. It leverages visual grounding via UI‑TARS and web data from Perplexica, learning from experience as it goes.

On leading benchmarks like OSWorld (50-step desktop tasks), Agent S2 achieves around 34.5% success, outpacing OpenAI’s agent (~32%). On AndroidWorld (smartphone tasks), it hits 54%, beating previous record holders. It also supports seamless cross‑platform use via a unified API.

Already in practical use—booking flights, shopping online, even complex research—Agent S2 offers a flexible, privacy-friendly way to build automation, QA, and accessibility tools that actually work, without relying on cloud APIs.

For teams building real-world computer-use AI, Agent S2 delivers a modular, memory-enabled, and extensible toolkit—bringing us closer to true digital assistants that operate like real colleagues.

🤖 Top Picks from Hugging Face: Trending AI Applications You Can't Miss!

✨ OmniGen2: Unified Multimodal Magic
A powerful open-source model with separate pathways for text and images—offering top-tier performance in generating visuals, editing pictures, and blending text-and-image inputs.

✨ VideoRefer-VideoLLaMA3: Intelligent Video Q&A
Upload a video, highlight an area, and this AI will generate captions or answer questions about specific objects—ideal for nuanced spatial-temporal understanding.

✨ Self‑Forcing: Real-Time, Smooth Video Creation
A breakthrough in video diffusion, training the model to simulate inference during training—enabling streaming-quality video generation in real time on a single GPU.

✨ Sparc3D: High-Fidelity 3D from a Single Image
Transforms one photo into a detailed 3D model using sparse voxel techniques—capturing intricate geometry and preserving fine surface details efficiently.

🛠️ Introducing Gemini CLI: AI in Your Terminal

Google just launched Gemini CLI, an open-source tool that brings the cutting-edge Gemini 2.5 Pro AI directly into your command line. Built for Windows, Mac, and Linux, it empowers developers to write code, fix bugs, research, generate content, and even summarize images or videos—all via natural language prompts, straight from the terminal. It connects to AI tools like Imagen and Veo, and integrates with Google Search and Model Context Protocol for real-time data. Best of all? It’s free with a personal Google account: up to 60 model calls per minute and 1,000 daily—no payment or API key needed.

This is more than a helper—it’s an agent capable of executing tasks, editing files, and planning workflows. With shared tech between Gemini CLI and Gemini Code Assist in IDEs like VS Code, your prompts seamlessly translate between environments. And because it’s open-source, developers can inspect, tweak, and contribute under an Apache 2.0 license.

For anyone working in software, research, content creation, or data-driven environments, Gemini CLI offers a powerful, flexible, and free way to bring AI smarts right into your workflow.

We’re Curious… What we should cover more?

Click HERE to provide your feedback.

Do us a favor? Reply to this email and tell us what you'd like to see more (or less) of!

How did we do?

Click HERE to provide your feedback.