In the past week, the development of Artificial Intelligence agent technology witnessed a surge as leading companies including Google, Atlassian, and Microsoft rolled out new tools, frameworks, and collaborative standards. Google announced its open-source Agent Development Kit (ADK) at Google Cloud Next 2025, allowing developers to build multi-agent systems with modular workflows, large language model (LLM) routing, and cross-model compatibility. The ADK, underpinning Google’s internal tools, supports dynamic orchestration using frameworks like LangGraph and facilitates integration with platforms such as Vertex AI and LiteLLM. Additionally, over 50 major companies, including Atlassian and Salesforce, endorsed Google’s Agent2Agent (A2A) protocol to create secure standards for inter-agent communication.
Atlassian also made headlines at its Team ´25 conference by introducing Rovo AI agents designed to automate the software development lifecycle. The suite comprises agents for code planning, generation, review, and deployment, tightly integrated with Jira. A no-code Agent Designer was revealed to further ease development pain points and foster transparency, enabling trust in automated workflows for developers and project managers alike.
Microsoft extended its Azure AI Foundry with new orchestration features, powering multi-agent system coordination and real-time feedback loops via its Semantic Kernel technology. The company held an Agent Development Kit Hands-on Workshop in India and continued its Microsoft AI Agents Hackathon, focusing on best practices around multi-agent orchestration and leveraging new capabilities in the Azure AI Agents SDK, including real-time telemetry for agent improvement. Notably, partners such as KPMG and Fujitsu have adopted these Microsoft-powered workflows, claiming a 40% reduction in development complexity.
Other notable updates included Google’s release of Gemini 2.5 Pro Experimental, which topped chatbot leaderboards for coding and reasoning, and OpenAI unveiling its o3 and o4-mini inference models with advanced image reasoning capabilities. Conferences throughout the week, including Microsoft’s AI Expo Austin and Shanghai’s 36Kr AI Partner Conference, underscored intense global activity in both foundational agent research and enterprise adoption. Academic input was strong as well, with Penn State University and industry leaders discussing governance, trust, and security—particularly as research emerged highlighting vulnerabilities in autonomous Artificial Intelligence agent systems. Meanwhile, innovation continued in specialized industrial, infrastructure, and marketing domains, indicating growing adoption across sectors.
