# News: April 23, 2026

> 9 curated links on AI agents, Python, security, Google Cloud, and engineering, selected by Médéric Hurier.

- Author: [Médéric Hurier (Fmind)](https://www.fmind.dev/)
- Published: 2026-04-23
- URL: https://www.fmind.dev/news/2026-04-23/

- [Deep Research Max: a step change for autonomous research agents](<https://blog.google/innovation-and-ai/models-and-research/gemini-models/next-generation-gemini-deep-research/>): The new Deep Research and Deep Research Max agents offer MCP support and native visualizations for long-horizon, autonomous research workflows.
- [Subagents have arrived in Gemini CLI](<https://developers.googleblog.com/subagents-have-arrived-in-gemini-cli/>): Gemini CLI now supports specialized subagents that handle complex, high-volume tasks in isolated context windows, allowing the primary agent to act as a strategic orchestrator.
- [Claude Opus 4.7 on Vertex AI](<https://cloud.google.com/blog/products/ai-machine-learning/claude-opus-4-7-on-vertex-ai/>): Anthropic's Claude Opus 4.7 is now generally available on Vertex AI, offering improved performance in coding, long-horizon agentic tasks, and complex document vision capabilities.
- [A2UI v0.9: The New Standard for Portable, Framework-Agnostic Generative UI](<https://developers.googleblog.com/a2ui-v0-9-generative-ui/>): A2UI v0.9 introduces a framework-agnostic standard and SDK for Generative UI, enabling AI agents to dynamically generate tailored UI widgets using a project's existing component catalog.
- [Gemini Robotics-ER 1.6: Powering real-world robotics tasks through enhanced embodied reasoning](<https://deepmind.google/blog/gemini-robotics-er-1-6/>): This reasoning-first model upgrade for physical agents enhances spatial understanding and multi-view reasoning for complex tasks like instrument reading and success detection.
- [Gemini 3.1 Flash TTS on Google Cloud](<https://cloud.google.com/blog/products/ai-machine-learning/gemini-3-1-flash-tts-on-google-cloud/>): The new Gemini 3.1 Flash TTS model provides developers with precise controllability over AI-generated speech using over 200 audio tags for pacing and expressivity across 70+ languages.
- [Stitch's DESIGN.md format is now open-source so you can use it across platforms](<https://blog.google/innovation-and-ai/models-and-research/google-labs/stitch-design-md/>): Google Labs has open-sourced the DESIGN.md format, a shared visual language that allows AI agents to understand and validate design rules for generated UIs.
- [Turn your best AI prompts into one-click tools in Chrome](<https://blog.google/products-and-platforms/products/chrome/skills-in-chrome/>): "Skills in Chrome" allows users to save, remix, and instantly reuse AI prompts as one-click workflows directly within the browser's Gemini interface.
- [The Friction is Your Judgment — Armin Ronacher & Cristina Poncela Cubeiro, Earendil](<https://www.youtube.com/watch?v=_Zcw_sVF6hU&si=kAoex1qKFPIVp-7Q&themeRefresh=1>): This video features a discussion on the "friction" in AI development and judgment, focusing on the human-AI interaction and engineering challenges.