Definition
A centralised traffic controller that sits between an organisation's users or applications and one or more AI models. Like a network firewall but purpose-built for AI, it enforces security policies — blocking dangerous inputs, preventing data leakage, monitoring usage, and managing credentials — across every AI interaction in one place.
Why it matters
Organisations often deploy dozens of different AI models across many teams. An AI gateway prevents each team from having to build its own safety controls, and ensures that if a policy changes — or a model is compromised — it can be enforced everywhere instantly.