Product documentation
Understand AI-assisted workflows
How Aiviance separates questions, proposals and consequential actions while preserving entity scope and human authority.
Open
Aiviance uses AI to help people understand information and prepare work inside a defined financial scope. The assistant does not receive unlimited access, and a conversational answer is not the same thing as an authorised financial action.
Questions and actions follow different paths
Read-only questions can use information already available to the signed-in user and active entity. A workflow that would create, update, submit, post, send or otherwise affect a record follows the product's action controls instead.
Depending on the workflow, Aiviance can show a draft, request missing information or stop at a confirmation step. Approval must refer to the pending action; unrelated conversation is not treated as permission.
Scope is established before work begins
The active person, family or business context determines which records and capabilities are available. Aiviance should not combine information from another entity merely because the same user can access both. Role, relationship and regional availability can further limit an action.
People retain authority
AI can classify, compare, explain or prepare a proposed next step where the product supports it. Users can review, change or reject a proposal. An action can also be refused when required data, authority or confirmation is missing.
Confirmed and posted financial history follows the product's correction rules. AI assistance does not turn an immutable record into an editable one or remove the audit boundary around a consequential decision.
Availability is specific to the signed-in workspace
The exact tools, approval steps and supported workflows depend on the active platform, entity, permissions, region and plan. This public page describes the control model, not a complete list of internal tools. Signed-in users should use the Aiviance Guide for capabilities available in their current workspace.
For model and data boundaries, read AI processing and human review and Access controls.