Скачать или смотреть Interoperability in AI Agents on NANDA - Rakesh Gohel, JUTEQ

Rakesh Gohel, founder of JUTEQ Inc. and co-author of "Agentic Artificial Intelligence," highlights the critical need for interoperability among diverse AI agents, arguing that current frameworks operate in isolated "islands" without universal communication standards.

He explains that while single agents are limited, multi-agent systems offer specialized expertise and real-time collaboration, necessitating universal "translator" protocols like MCP (agent-to-tool communication) and A2A (agent-to-agent communication), which he sees as complementary rather than competing.

Gohel emphasizes that MIT's Project NANDA is building the foundational infrastructure for this decentralized "Internet of Agents," akin to a "Linux moment" for AI, by creating a new registry system and secure frameworks that enable discovery, search, authentication, and verifiable interactions based on principles of privacy and verifiability.

Ultimately, he predicts that the future of AI lies in "agentic composability," where businesses and individuals will combine specialized agents from an open, decentralized ecosystem rather than relying on single, proprietary solutions.