Brilo AI supports a Knowledge Approval approach where knowledge changes can be staged and held for human review before they are published to the live Brilo AI voice agent. With Knowledge Approval configured, content edits, new FAQ entries, or updated documents can be routed into a staging queue, reviewed by one or more approvers, and published only after sign-off, preserving version control and an audit log. This reduces risk from accidental or unvetted updates while keeping knowledge up to date through controlled content ingestion. Related terms include knowledge base, approval workflow, staging environment, publish, version control, audit log, and webhook endpoint.

Can knowledge changes require approval before going live? — Yes. When enabled, Brilo AI can hold knowledge updates in a staging queue until an assigned approver publishes them.

Do I need manual sign-off for every edit? — You can configure Brilo AI to require sign-off for all changes or only for changes to specific knowledge sources.

Can I preview changes before publish? — Brilo AI supports staged previews so reviewers can test changes against sample calls before publishing.

Enterprises ask about Knowledge Approval because phone-facing knowledge directly affects what callers hear and how sensitive information is handled. For healthcare, banking, and insurance organizations, an incorrect response can escalate risk, create regulatory exposure, or trigger needless escalations. Buyers want to know whether Brilo AI allows a predictable, auditable path for knowledge updates so that legal, clinical, or compliance teams can review changes before they reach customers.

When Knowledge Approval is enabled, Brilo AI separates content edits from the production knowledge base. Changes are made in a staging area and recorded as a draft version. An approval workflow assigns reviewers and enforces sign-off rules; only after an approver publishes the draft does Brilo AI merge the new content into the live knowledge base used by the Brilo AI voice agent. Brilo AI maintains version control and an audit log so teams can trace who changed what and when. In Brilo AI, knowledge change is a recorded edit to a knowledge source that can be staged, reviewed, and published. For guidance on what to do when the voice agent cannot answer because of pending changes, see the Brilo AI article on how the agent handles uncertainty: Brilo AI “What happens when the AI is unsure?”

Brilo AI enforces safety boundaries to limit the scope of what a knowledge approver can publish. Approval workflows should be used for content that affects caller safety, regulatory responses, or transactional logic. Brilo AI will not automatically publish changes that are flagged by content rules or that fail validation checks; such items stay in the staging queue until corrected. In Brilo AI, approval workflow is the configured process that requires one or more named approvers to review and publish staged knowledge. In Brilo AI, audit log is the immutable record of drafts, approvals, publishes, and rollbacks that supports traceability and incident review. For patterns on end-to-end deployments that include staged testing and guardrails, review Brilo AI’s guidance on agent scope and deployment: Brilo AI “Can the AI voice agent answer calls end-to-end?”

Brilo AI voice agent workflows can be configured to hand off to a live agent or to a secondary workflow if a caller reaches content gated by pending approval or if the agent detects uncertainty. Typical patterns:

Using Brilo AI Knowledge Approval reduces the chance of publishing incorrect or non-compliant content to callers, lowering operational risk and limiting costly escalations. Controlled publishing increases reviewer confidence and shortens the review-to-publish cycle by formalizing who can approve changes. Organizations gain predictable change windows and an auditable trail that supports governance and periodic reviews.

Who can be an approver?

Approvers are named users or groups configured in your Brilo AI tenant. Typical approvers are subject-matter experts, compliance officers, or product owners who have access to the staging area and the audit log.

Can I require multiple approvals for high-risk content?

Yes. Brilo AI’s approval workflow can be configured to require sequential or parallel approvals for content tagged as high risk.

What happens if an approver rejects a change?

Rejected drafts remain in the staging queue with reviewer comments. Authors can edit the draft and resubmit it for approval; Brilo AI preserves previous draft versions for traceability.

Can I roll back a published change?

Yes. Brilo AI records versions and allows you to revert to a previous published version, while keeping the rollback action in the audit log.

Does Knowledge Approval affect automated model training?

When Knowledge Approval is enabled, only published knowledge is used for live agent responses and model tuning. Draft content stays isolated until published to avoid polluting production behavior.