The world of AI agent collaboration took big steps forward this week with new tools and standards announced by major tech companies.

Microsoft revealed multi-agent orchestration capabilities in its Copilot Studio platform. This allows multiple AI agents to work together on complex tasks, similar to how human teams collaborate. For example, one agent could analyze data while another creates reports, with both checking each other's work. The company also launched NLWeb, a new open standard that acts like HTML for AI agents. NLWeb lets websites offer conversational interfaces that different AI systems can understand, making it easier for agents to share information.

SAS introduced a governed AI agent framework in its Viya platform, emphasizing safe collaboration between human and artificial intelligence. Their system includes built-in checks for bias and compliance, ensuring agents make decisions that follow company rules. SAS plans to release industry-specific agents later this year that can automatically work with existing business tools.

Amazon Web Services contributed to the collaborative AI landscape by open-sourcing its Strands Agents SDK. This toolkit helps developers create AI agents that can split tasks into smaller pieces and work on them simultaneously, then combine the results. It's like having a team of digital workers that can divide and conquer complex problems.

Key partnerships emerged as Microsoft joined the Model Context Protocol (MCP) steering committee, working with GitHub to improve how AI agents securely access online services. Meanwhile, SAS announced collaborations with Microsoft, Intel and AWS at its Innovate conference, showing how big tech companies are teaming up to shape the future of AI teamwork.

These developments show a clear trend toward interoperable AI systems that can work together across different platforms. Just like email works whether you use Gmail or Outlook, new standards like MCP and NLWeb aim to let AI agents collaborate regardless of which company created them. As these technologies mature, we'll likely see more examples of AI teams handling tasks that currently require human specialists.