• by Laurie Sullivan , 4 hours ago

Anthropic has introduced an extension to its Model Context Protocol (MCP), an open-source standard for connecting artificial intelligence (AI) applications to external systems.

Using MCP, AI applications like Anthropic's chatbot Claude can bring apps into its platform, similar to the way OpenAI did last year with ChatGPT.

Anthropic’s MCP is the foundation technology for automated ad media buying. In October 2025, Yahoo, PubMatic, and Magnite and others launched the Ad Context Protocol (AdCP), which is built directly on the MCP standard. 

This integration enables agentic media buying, where AI agents can plan and execute campaigns without continuous human intervention. 

Claude acts as a conversational front door, which is key to these app integrations, and can pull real-time insights from a variety of apps.

It allows users to connect Claude with tools such as Asana, which turns chats into projects and tasks; Box, which searches for files and previews documents; Slack, which searches and retrieves conversations for context to generate messages; and many other integrations.  The tools serve up in the conversation, so users can collaborate with them in real time. 

Anthropic and Salesforce also have deepened their partnership by using MCP to integrate Claude with Salesforce's Agentforce 360 platform.

In a blog post, Anthropic described how Salesforce will bring enterprise context to Claude with Agentforce 360 to enable teams to reason, collaborate, and act from one connected interface.

MCP is supported by several technologies such as Ad Context Protocol (AdCP), the industry standard for agentic advertising; Agent-to-Agent (A2A) Protocols, which allows a buyer's AI agent to negotiate directly with a publisher’s AI agent, and IAB Tech Lab Roadmap, which makes it possible to build on these protocols.

PubMatic, Yahoo, and others have been testing MCP servers that allow buy-side AI agents to communicate and execute campaigns directly within their systems.

Earlier this month, built into the Claude desktop app, Anthropic announced a new tool called Cowork. It lets users designate a specific folder where Claude can read or modify files, with further instructions given through the standard chat interface.

This integration across apps and AI will prove successful for Anthropic as it works to build ethics and thrust into its technology.

Last week Anthropic published what it calls a Constitution for its AI model Claude. This is a detailed description of Anthropic’s vision, values and behavior; a document that explains the context in which Claude operates such as ethics and safety, and the kind of entity it will become.