Overview

CID222 is a multi-tenant AI gateway. It sits between the people and applications in your organisation and the large language models they use, inspects what crosses in both directions, and applies your policy before the provider sees the request and before the answer reaches the user.

Inspection covers personally identifiable information, toxic and harmful content, prompt injection and jailbreak attempts, and — on the response side — hallucination against supplied context. Every verdict is attributed to a tenant and recorded, so the same appliance answers both "was this blocked?" and "who has been sending this?".

What this guide contains

• What CID222 is — the problem it solves, what it does to a request, and the things it is not.
• Deployment modes — the six interception points compared on what each one sees and what it needs on the client.
• Architecture and services — the gateway, every container behind it, and the path a request takes.
• Multi-tenancy — tenants, tenant groups, how credentials resolve, and what isolation actually guarantees.
• Packages and licence tiers — what each package includes, how an entitlement is resolved at runtime, and what the trial gives you.
• Detection engines at a glance — the seven engines, what each finds, and which service runs it.
• Roles at a glance — the five fixed roles and how much data each one sees.

Where to go next

• Deploying for the first time → Install and activate.
• Running an appliance day to day → Administer.
• Calling the API from your own code → Develop.
• Something is broken → Troubleshoot and diagnose.