source: techcrunch ai: ai’s most important protocol is getting a little bit easier to use
level: technical
the model context protocol, or mcp, is a standard that lets ai models securely connect to external tools like calendars, databases, and business apps. instead of building custom integrations for every service, mcp provides a common way for chatbots and agents to access data and take actions. a new version of the protocol, set to roll out next week, aims to fix a key scaling problem that has held back wider adoption.
currently, mcp relies on session ids that tie a client to a specific server. when a client like claude connects, the server issues an id that must be sent with every request. this works fine for small setups, but in real deployments with millions of users and load balancers spreading traffic across many servers, keeping track of those ids becomes a major headache. each server has to know about sessions created on other machines, adding complexity and cost.
the update introduces a stateless approach, similar to how most websites handle sessions. servers will no longer need to maintain persistent session state, making it easier to run mcp at scale behind load balancers. this change should lower the barrier for companies to offer first-party mcp integrations, helping the ecosystem move beyond hype and into practical, large-scale use.
why it matters: simpler scaling for mcp means more companies can deploy ai agents that reliably connect to real business tools, turning agentic ai from a demo into a practical reality.
source: techcrunch ai: ai’s most important protocol is getting a little bit easier to use