A groundbreaking collaboration is taking place in the world of artificial intelligence, and it's set to revolutionize how AI agents interact and work together. The future of AI is here, and it's all about collaboration and openness.
OpenAI, Anthropic, and Block, three leading companies in the AI space, have joined forces to establish the Agentic AI Foundation, a new open-source organization dedicated to promoting standards for AI agents. But here's where it gets controversial: they're not just creating standards, they're transferring ownership of key agentic technologies to this foundation, opening up a whole new world of possibilities.
The technologies in question are game-changers. Anthropic's Model Context Protocol (MCP) allows AI agents to connect and communicate, while OpenAI's Agents.md sets the rules for coding these agents, and Block's Goose framework empowers the creation of AI agents that can perform tasks on computers. These tools were already free to use, but now, through the foundation, anyone can contribute to their development, taking AI collaboration to the next level.
"MCP is a powerful tool, but its adoption has been varied," says Nick Cooper, who leads work on MCP at OpenAI. "By making it an open standard, we're encouraging developers and companies to embrace it, building systems that integrate AI seamlessly. This open interoperability is a game-changer, allowing companies to communicate across different AI providers and systems."
The Agentic AI Foundation is an ambitious project, overseen by the Linux Foundation, which has a proven track record in managing open-source projects like the Linux operating system. Other big names in the tech industry, including Google, Microsoft, and AWS, have also signed on, showing their commitment to this new paradigm.
This shift towards agentic AI is a significant one. It moves us beyond simple chat-based systems to a world where AI agents take actions on our behalf, negotiating and interacting with each other to power a wide range of applications. Imagine AI assistants handling your bookings and purchases, while businesses use AI agents to manage transactions and customer interactions. It's a potentially lucrative and transformative shift.
Srinivas Narayanan, CTO of B2B applications at OpenAI, envisions a future where AI agents routinely communicate, and he believes that working across open standards will ensure these interactions are smooth and seamless. "Open source is key to shaping and adopting AI in the real world," Narayanan emphasizes.
The question of openness is a critical one for the AI industry. US companies have largely made their money through closed models, offering access via APIs. However, there's a growing trend towards open-source models, with companies like Meta releasing their best models for anyone to use. Chinese AI companies, such as DeepSeek and Alibaba, have also embraced open-source models, gaining popularity with developers and researchers. This raises concerns about a potential strategic advantage for Chinese firms over time.
OpenAI, Anthropic, and Block believe that openness is the key to unlocking the full potential of agentic AI. While their standards don't require the use of their foundation models, they believe that openness will ultimately benefit their businesses by removing barriers to deployment.
Manik Surtani, head of open source at Block, highlights the popularity of their Goose agent, which has seen a surge in use over the past year. "Donating Goose to the AAIF makes it easier for anyone to contribute to its development and build upon it," Surtani explains.
Jim Zemlin, executive director of the Linux Foundation, sums it up: "MCP, Agents.md, and Goose are essential tools for developers building agentic technologies. By bringing them together under the AAIF, we ensure their growth with the transparency and stability that open governance provides."
While open standards may be technologically neutral, the global dominance of these agentic tools could give the US companies behind them significant influence. Just as ICANN and WC3 standards have shaped the web, this new foundation could play a role in determining how AI is used worldwide.
This collaboration is a bold step towards a more open and collaborative future for AI. It's an exciting development, and one that will undoubtedly spark debate and discussion. What do you think? Will openness be the key to AI's success, or are there potential pitfalls we should consider? We'd love to hear your thoughts in the comments.
