Skip to content
← All news

January 29, 2026

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

  1. The new era of browsing: Putting Gemini to work in Chrome

    blog.google

    Google details Gemini 3-powered Chrome features like the side panel assistant, connected apps, and auto browse for agentic multi-step tasks.

  2. Kimi K2.5 (moonshotai) on Hugging Face

    huggingface.co

    Moonshot AI releases Kimi K2.5, an open-source native multimodal MoE model with agentic tooling support, long context, and deployment guidance for vLLM/SGLang.

  3. Tailor Gemini CLI to your workflow with hooks

    developers.googleblog.com

    Google introduces Gemini CLI hooks so developers can inject context, enforce policies, and add safety checks like secret scanning without modifying the CLI source.

  4. Spanner in 2025: Innovations powering intelligent, multi-model AI applications

    cloud.google.com

    Google summarizes 2025 Spanner advances for AI workloads, including multi-model (vector/graph/text) support, Vertex AI integrations, and performance and safety improvements.

  5. Future Mode Part 1: The intelligent and secure browser for enterprises

    cloud.google.com

    Chrome Enterprise outlines its roadmap for agentic browsing at work with security controls, policy enforcement, and Gemini 3 integration.

  6. Foundations of Secure MCP: Architecture and Threat Model

    youtube.com

    Google Cloud Tech explains security architecture and threat modeling considerations for Model Context Protocol deployments.

  7. Artificial Analysis: Kimi K2.5 results for you to swipe through

    reddit.com

    A Reddit post sharing benchmark slides from Artificial Analysis about Kimi K2.5 performance.

  8. Master 80% of Google AntiGravity in 24 Minutes

    youtube.com

    A tutorial-style video covering the core workflows and agentic features of Google AntiGravity.