RBAC and data access scoping for finance AI
An AI workflow inherits risk from its permissions. If it reads more than the role it serves, it becomes a data-exfiltration path regardless of how well it behaves.
Octopus AI holds read-only access, scoped per source system, under role-based access control.
Sending is governed separately from reading: every outbound message is approval-gated by a named human.
Questions and answers
How is read access scoped?
Per source system and per role. An AFI capability cannot read data that its assigned role cannot read, and access is read-only.
Can an AI workflow escalate its own permissions?
No. Scope is configuration owned by the customer. A request outside documented scope is declined and escalated to a named human.
How is outbound communication controlled?
Separately from read access. Recipients, templates, thresholds, and cadence are configured by the customer, and each message requires human approval before it sends.
Can a workflow be stopped without stopping the platform?
Yes. Any AFI capability can be paused independently.
What is logged about access?
Every action, with who, what, and when, exportable for audit alongside approvals and version history.
Proof
- Reference client: $523K total contract value over 3 years.
- Approximately 430 analyst hours saved.
- Forecast cycle cut from about 60 hours to under 6 hours (90% faster).
- Error rate below 0.1%.
- 2,500+ autonomous tasks completed in month one.