Approving and governing MCP servers is only part of the picture — AI Stewards also need visibility into what's happening once those servers are live. This video walks through the observability features available in AI Control Tower for MCP servers running through AI Gateway. Starting from the AI Gateway tab on the AI Control Tower landing page, we look at gateway-level transaction counts and success rates across all active servers, then drill into the Linear Prod record to see tool-level KPIs including request counts, success rates, and latency at P50, P90, and P95. We finish in the Security & Privacy tab, where access visualizations surface authorized and failed connection attempts over time — giving AI Stewards the signals they need to act. Part of the AI Gateway Centralized MCP Server Governance capability.