In late November, Anthropic introduced the Model Context Protocol, an open standard for connecting AI assistants to the systems where data lives. It did not make the same noise as a new model. I suspect it may matter just as much.
I have seen this movie
Much of my career has involved business telephony. In that world, standards are what make everything possible. Open protocols let equipment and software from different vendors talk to each other, and they turned phone systems into something you could build on with open-source platforms.
AI tools have the opposite problem today. Each assistant needs its own custom connection to each database, repository or service. Every new combination means another integration to write and maintain.
Why a standard matters
A shared protocol means a connector can be written once and used by many clients. That is the kind of unglamorous plumbing that makes an ecosystem grow.
The part engineers should not skip
Connecting a model to real systems also expands what can go wrong. A connector is an access path, and it deserves the treatment we give any other one: least privilege, logging and careful review. In telephony, a badly secured trunk can become an expensive lesson. An over-permissioned AI connector could be worse.
Standards tend to win quietly. This is one to keep watching.