Skip to content
← All news

October 28, 2026

7 curated links on AI agents, Python, security, Google Cloud, and engineering, selected by Médéric Hurier.

  1. Genie 4: A new generation of world simulation

    deepmind.google

    Introduces Google DeepMind's foundational world model built on Gemini Omni, with native spatial audio, live natural-language scene steering, and Gemini 4 Argon reasoning for embodied agent training.

  2. Gemini Robotics Embodied Reasoning 2 is now available on Google Cloud

    cloud.google.com

    Launches public preview of the Gemini Robotics ER 2 model on Google Cloud and AI Studio, bringing video stream analysis, long-horizon planning, tool calling, and multi-robot coordination.

  3. 8 examples of Gemini 4 Argon in action

    blog.google

    Demonstrates Gemini 4 Argon's long-horizon reasoning across eight workloads, including a 5,000-line zero-dependency 3D renderer and driving Blender via MCP to turn desk photos into a 3D game.

  4. Fast, reliable A2UI, with swappable inference formats

    developers.googleblog.com

    Upgrades the open A2UI protocol across Python, TypeScript, and Dart SDKs with swappable DSLs, cutting layout latency 3x with A2UI Express and up to 80x with controlled generation.

  5. Building Conversational Commerce with Multi-Agent Swarms on GCP: A Working Prototype

    developers.googleblog.com

    Details a Google Cloud reference architecture that replaces retail chatbots with multi-agent swarms built on ADK, Vertex AI Agent Engine, and MCP/A2A to coordinate discovery through checkout.

  6. Gemini Updates: Powering Enterprise AI Agents

    youtube.com

    Showcases Gemini Enterprise capabilities for building, governing, and orchestrating multi-step autonomous agents across corporate data and SaaS ecosystems with the Agent2Agent (A2A) protocol and role-based access.

  7. The AI bubble is about to pop...

    youtube.com

    Compares current hyperscaler AI capital expenditures and chip infrastructure buildouts against historical railroad, telecom, and dot-com boom-bust cycles to evaluate the financial risks of an AI market correction.