Tag: MCP

Broad credentials made sense when fixed application logic stood between users and actions, but AI agents have changed that equation.
What began as a way for agents to call tools is expanding into infrastructure for longer-running, governed interactions across enterprise systems.
XAA removes a growing source of agent friction, but uneven support across the enterprise ecosystem makes the transition the harder problem to solve.
Centralized authorization eliminates repeated user consent, but autonomous agents still need runtime policy, short-lived credentials, and audit.
Already run CrowdStrike AIDR? See how it builds on Aembit’s identity control plane for AI agents. Read the full breakdown.
Stateless requests, explicit state, and routing metadata bring familiar distributed-systems patterns to agent infrastructure at scale.
The gateway label now covers several very different jobs. The useful question is what traffic each gateway handles, what decision it supports, and where identity and access fit in the architecture.
The second in a five-part series on how MCP is moving beyond tool calling, and what that shift means for agent workflows, interoperability, and enterprise use.
Aembit now integrates with CrowdStrike AI Detection and Response to secure MCP connections and inspect what happens after an AI agent connects.
The first in a five-part series on how MCP is changing for real-world use, and what those changes mean for teams building and securing agent systems.