News Hub

Google introduced Agent Plugins that package skills, tools, and related functionality.

Google released MCP Stateless updates to improve scaling of AI agent infrastructure.

Google describes session-aware load balancing techniques for scaling real-time AI agents.

Google Developers announced Agent Skills in Genkit Go for enabling on-demand expertise.

Agent and Model Evaluations in Gemini Enterprise Agent Platform reached general availability.

Google provides microbenchmarks for evaluating TPU performance.

Ray AI libraries now run on TPU infrastructure.

Google presents Tunix, a method for high-throughput agentic reinforcement learning training.

Google provides foundations for running Ray distributed computing framework on TPU hardware.

Google describes modular prompt transpilation techniques for building scalable AI agents.

Gemini Enterprise Agent Platform now supports grounding with parallel web search.

Google documents optimization of Qwen 3.5-397B MoE model on TPU7x hardware.

Google released LiteRT.js for high performance web AI inference.

Google built an AI race coach application using Antigravity and Gemini.

Google demonstrates elastic training with MaxText that recovers from terminated TPU in seconds.

Google Cloud Workbench Extension for ML development in VS Code now available.

Google Genkit enables building agentic full-stack applications.

Title alone insufficient to summarize; article concerns ADK 2.0 from Google Developers.

Google Developers discusses driving agent quality flywheel from coding agents.

Google released ADK Go 2.0 with graph-based workflow engine, human-in-the-loop, and dynamic orchestration for multi-agent applications.