The latest wave of AI innovation is redefining collaboration, discovery, and embodied intelligence, as industry leaders push the boundaries of how humans and machines interact across digital and physical spaces.

🤖 Google DeepMind SIMA 2 – Google DeepMind unveils SIMA 2, a next-generation AI agent capable of reasoning, conversing, and generalizing across 3D virtual worlds. Designed as a cooperative companion rather than a simple executor, it adapts dynamically to new environments and games, offering a glimpse into the future of universal AI assistants and robotics.

🔍 LinkedIn AI People Search – LinkedIn introduces an AI-powered people search for Premium users, bringing conversational intelligence to professional discovery. By blending LinkedIn’s professional graph with AI comprehension, this feature streamlines networking and talent discovery, with global expansion planned in later phases.

💬 OpenAI ChatGPT Group Chats – OpenAI rolls out collaborative group chats in ChatGPT, allowing up to 20 participants to co-create and discuss in real time with adaptive AI support. Active across select Asia-Pacific regions, groups can share links, set custom instructions, mute notifications, and engage with GPT‑5.1 Auto, which tailors responses per participant’s tier.

Together, these developments point toward a more interactive, connected, and reasoning-driven era of AI, where digital companions, search engines, and chat platforms grow increasingly responsive to how people think, work, and communicate.

SIMA 2 Revolutionizes AI Agents in 3D Worlds

SIMA 2, unveiled by Google DeepMind, marks a major leap in AI agents for 3D virtual environments by merging scalable instruction-following with the advanced reasoning power of Gemini models. Unlike its predecessor, SIMA 2 can not only interpret and execute complex language-based instructions but also reason about goals, converse naturally, and generalize its learned skills across multiple games and brand-new, AI-generated worlds. Building on hundreds of skills learned from human demonstrations and evolving through self-improvement, SIMA 2 operates more like a collaborative companion than a simple command-execution bot, and even adapts to completely new games with human-like flexibility. Its progress demonstrates a path toward generalist embodied intelligence, crucial for future robotics and AI assistants, while responsible development remains central as it is released as a limited research preview to game developers and academic partners.

LinkedIn Launches AI People Search for Premium Users

LinkedIn has introduced an AI-powered people search feature for Premium users in the U.S., allowing professionals to find others using natural language queries instead of strict job titles or filters. Users can now describe what they’re looking for, such as “someone who’s grown a small business” or “an expert in digital marketing,” and the AI interprets context and intent using LinkedIn’s professional graph to surface the most relevant matches. This feature bridges conversational input with professional discovery, making it easier to uncover new connections, collaborators, or hires beyond traditional keyword searches. LinkedIn plans to expand the rollout to more regions and membership tiers, positioning AI as a key driver in how users explore and connect across its network more intelligently.

OpenAI Introduces Group Chats in ChatGPT

OpenAI has launched a new group chat feature in ChatGPT, allowing up to 20 users to collaborate in a single shared conversation with AI support. Currently in pilot across Japan, New Zealand, South Korea, and Taiwan, this feature lets users create groups via a shareable link, set custom instructions per group, and interact with the AI which now reacts contextually and selectively instead of responding to every message. Participants can see who’s typing, use emojis, rename chats, mute notifications, and enjoy tailored AI responses powered by GPT-5.1 Auto that adapt to each user’s subscription tier. This new capability enhances teamwork and social interactions through AI-mediated group brainstorming, planning, or decision-making, marking OpenAI's first shared ChatGPT experience.

In this video, we’ll look at how Google’s Gemini Nano Banana model edits complex images, replacing objects, improving vibes, restoring photos, and even making creative transformations.

Top AI Products from this week

• Nakora GitHub Repo Visibility - Simplify managing your GitHub repositories' visibility settings with Nakora's tool, allowing easy switching between public, private, and internal repo statuses while understanding the implications for project access and collaboration.

• Sendr - Scale personalized sales outreach effortlessly with AI-powered custom videos, dynamic landing pages, and enriched contact data; record a single video pitch and let AI tailor it for each contact to boost engagement and booking rates.

• PartGenie - Accelerate electronics component sourcing and BOM optimization using AI to instantly analyze datasheets, find in-stock alternatives, and validate specifications, cutting procurement time from weeks to minutes.

• Recap - Transform your knowledge into visual insights with AI by summarizing webpages, PDFs, and documents into viral posts, videos, mind maps, and tables, making complex content easy to understand and share.

• SpeedCode - Master high-performance parallel programming through hands-on coding challenges on an online platform, focusing on developing fast multicore solutions and evaluating their scalability in a cloud environment.

This week in AI

• LMArena Code Arena - LMArena has launched Code Arena, a next-generation platform for evaluating AI coding models through real-world development cycles. The platform features agentic tool calls, persistent sessions, shareable code generations, and a unified workflow for prompt, generation, and evaluation.

• MIT Self-Adapting Language Models (SEAL) – MIT researchers have developed SEAL, a novel framework that allows LLMs to update their weights using a trial-and-error reinforcement learning process that selects the best self-edits for improving performance on tasks like question answering.

• Even G2 and R1 – Even Realities has unveiled the Even G2 Display smart glasses and R1 smart ring, a quietly innovative pair of wearables designed to integrate seamlessly into daily life. The G2 glasses feature a dual-layer spatial display, natural gesture control via the R1 ring, and on-device AI like Conversate and Even AI for contextual assistance.

• Visa Expands Intelligent Commerce with AI Pilot - Visa is expanding its Intelligent Commerce initiative across Asia Pacific, accelerating the rise of agentic commerce where AI-powered agents shop and pay on behalf of consumers.

• Ask Oracle chatbot – Oracle has introduced the Ask Oracle chatbot, a sample Oracle APEX application that offers a no-code chatbot interface leveraging Autonomous AI Database Select AI capabilities.

• Google Private AI Compute - Built on Google’s custom TPU infrastructure and secured by Titanium Intelligence Enclaves (TIE), this service enables faster, smarter AI experiences without compromising data control. It bridges the gap between cloud AI power and local privacy, enhancing features like Pixel 10’s Magic Cue and Recorder app.

Paper of The Day

The research paper presents a comprehensive overview of how agentic AI systems, capable of reasoning, planning, and autonomous decision-making, are revolutionizing scientific research by automating tasks such as literature review, hypothesis generation, experiment execution, and result analysis. It categorizes current AI tools and systems employed in scientific discovery and highlights significant advancements in fields like chemistry, biology, and materials science. The paper discusses essential evaluation metrics, implementation frameworks, and datasets, addressing critical challenges including automation limits, system reliability, and ethical concerns. It also outlines future research directions that emphasize enhancing human-AI collaboration and improving system calibration to maximize the transformative potential of agentic AI in scientific innovation while ensuring ethical deployment.

To read the whole paper 👉️ here