Private processing
Australia only keeps private AI processing with Australian-served models. Global permits the wider provider catalogue, including providers outside Australia.
Data & AI for Australian RTOs
RTEO separates public, non-sensitive work from private, sensitive data so your workspace can make a deliberate choice about where each kind of work runs.
Workspace setting
Choose the lane.
Private and public work are classified before a model is selected.
Australia only
Separate public control
Two data classes
RTEO treats a call as private whenever it contains personal, customer or workspace-held material. Public means openly available content that does not identify a person. The label belongs to the call, not to the AI model.
Students, contacts, enrolments, email content, conversations over workspace data and client-authored drafts. These follow the private processing setting.
Your public website, public competitor pages, search data, open-web research and generated public assets. These follow the separate public setting.
If a call is not explicitly classified, RTEO uses the safer private route. It never guesses that a call is safe to send overseas.
Workspace controls
Your workspace chooses where private data may run. When private processing stays Australia-only, the public setting can keep public work in Australia where a suitable model exists or use Global for the wider model catalogue. If private processing is Global, public work is Global too.
Australia only keeps private AI processing with Australian-served models. Global permits the wider provider catalogue, including providers outside Australia.
Australia where available keeps public work on an Australian model when one exists. Global keeps more public features available.
Workspace owners can tighten private processing to Australia only; enabling Global private processing requires RTEO support. Members can choose the separate public lane in Settings > Data & AI.
The router uses the call's classification and the selected lane before it chooses a provider or model.
The settings page explains what runs where, and why a feature is unavailable when Australia has no suitable model.
What runs where
The exact model depends on the AI role, the data class and the workspace setting. The useful promise is simple: Australia-only private work does not silently fall back to a Global model.
Private calls use Australian-served models through AWS Bedrock when the role has one. Public calls can use the same lane when the public setting is Australia where available.
Global processing uses the wider model catalogue through the configured gateway and named providers described in the DPA.
If Australia-only processing is selected and a role has no Australian model, that feature reports unavailable instead of routing overseas. The rest of the workspace keeps working.
For your client review
The DPA sets out the data classes, processing modes, subprocessors, security measures and deletion commitments in detail. It is public so your team can review it before signup.