The .AI layer runs across the whole stack — configurable AI agents for servicing, AML triage, adviser assist and comms; Document AI reading paper; a durable memory holding what agents learn; and an action layer that lets agents work through the portals with the same guardrails as a human. Configured per journey, not one-size-fits-all.
The spine — agents tuned per journey rather than a single generic assistant.
Where agents are built, configured and connected into the rest of the stack.
A durable, governed place agents read from and write to — separate from the systems of record they support.
Reads paper and handwritten forms into the validated pipeline.
The safe interface through which agents actually take action, under the same guardrails as a human user.
Each agent is scoped to a specific job and a specific part of the stack — so value is measurable and behaviour is auditable.
Answers member questions and drafts actions inside the portals.
Pre-sorts and prioritises KYC/CTF cases before human review.
Surfaces insights and next-best-actions in the adviser portal.
OCR+ extraction that normalises paper into structured data.
Drafts and drives customer messaging through the comms system.
Clears repetitive steps on the forms & analytics dashboard.
Most platforms add AI as a sidebar. Here, agents operate under the same identity controls that govern humans — the same SSO, MFA and role scoping. That's what makes an AML-triage or servicing agent safe to put in production: it can't do anything a permitted user couldn't, and every action is traceable.
Because the .AI Suite runs inside the StreamlineID scope on GeoLocked Australian AWS, agent activity stays within the same APRA-aligned boundary as the rest of the platform — and because agents are configured per journey rather than switched on globally, you can roll them out where the ROI is clear and hold the line everywhere else.
StreamlineID ✕