Arcade, a two-year-old startup, has updated the Model Context Protocol (MCP), a foundational AI standard that enables secure access to external data sources. The changes, which have been publicly available since May, aim to simplify how servers manage session IDs, making it easier to scale AI systems across multiple servers. According to Arcade, the new version of MCP adopts a stateless approach to session IDs, similar to how most websites function, reducing the complexity of maintaining user sessions in distributed environments.
The current MCP system requires servers to track session IDs across multiple machines, which creates challenges for companies operating at scale. Under the old system, when an AI model like Claude connects to a server, it sends a 'hello' message with its capabilities, and the server responds with its own and issues a session ID. This ID must be included in every subsequent request, which becomes problematic when traffic is distributed across multiple servers. The new approach eliminates the need for servers to remember session IDs, allowing them to operate more efficiently and with less overhead.
Arcade raised $60 million in June with the idea that most AI agents fail not due to weak models but because the infrastructure isn't ready. The MCP update is part of an effort to address these infrastructure challenges, making it easier for companies to deploy large-scale AI systems. The changes are expected to reduce operational complexity and potentially lower costs for maintaining AI services at scale.
Source: techcrunch
