This requirement is very specific: the agent must automate tasks by simulating user interactions across third-party apps and websites , including:

clicking buttons

entering text

extracting information from screens

That is exactly what Computer Use in Copilot Studio is designed for.

Why C is correct:

Computer Use enables agents to interact with user interfaces in a human-like way.

It is intended for scenarios where the agent must work across apps and websites that may not expose direct APIs.

It supports low-code business automation where the AI performs on-screen actions rather than just answering questions.

Why the other options are not correct:

A. a natural language understanding+ (NLU+) model in Copilot Studio NLU+ helps with understanding user intent, not with UI interaction automation.

B. Copilot skills Skills extend capabilities, but the question is specifically about simulating on-screen actions across apps and websites.

D. Model Context Protocol (MCP) MCP is for standardized tool and context integration between models and external systems. It is not the primary feature for direct UI simulation.