MCP (Model Context Protocol) is a protocol for AI agents to discover and use tools more flexibly, originally introduced by Anthropic. Recent interest in MCP has exploded, with Google Trends showing a sharp rise in searches following OpenAI's support announcement. A hands-on test using Twilio's MCP implementation found that MCP-enabled runs completed tasks ~20.5% faster on average, made 19.2% fewer API calls, and had slightly fewer user interactions. However, the use of MCP came with added cost, introducing overhead in cache reads and writes, which increased by ~28.5% and ~53.7%, respectively. The test also found that while MCP makes the agent smarter and faster at the cost of munching more tokens, there's reason to believe this cost burden will decrease over time. MCP is beneficial for generalist tasks but may not offer significant improvements for niche or proprietary scenarios. To make the most of MCP, developers should include only necessary API details and data, use tool filtering, monitor token usage, and experiment with models and settings.