A common way for AI agents to reach company databases, files and internal tools has gained backing across rival tech firms, reducing the need for endless custom links.
How the Shift Took Shape
Over the past two years, every AI agent required its own set of connections to the systems firms already use. This created a growing pile of separate code that had to be maintained and updated whenever any part changed. Anthropic first built its Model Context Protocol as an internal fix. It has since been placed under the Linux Foundation, where a neutral group now oversees it. Members include AWS, Google, Microsoft, Cloudflare, JPMorgan, Oracle and Salesforce.
The move reflects a clear change in the market. Large buyers made it plain they would not accept being locked into one agent provider. Vendors responded by agreeing on one interface rather than competing on their own closed versions.
What It Means for Smaller Businesses
For firms without large IT teams, the practical difference is simpler setup and lower ongoing costs. Instead of paying for multiple bespoke connectors that break on updates, a business can use one standard link that works with any compliant agent. This cuts the risk of being tied to a single vendor and makes it easier to test different tools without starting from scratch each time.
The change also shifts attention away from integration work toward how data is organised and kept accurate inside the business.
Points to Watch Next
Adoption will show first in widely used business applications such as customer records and support systems. Firms should track which of their existing tools add support for the protocol and how quickly new agents appear that rely on it rather than custom work. The real test will be whether this shared layer lasts longer than the models built on top of it.