How do I build a searchable knowledge base from my documents?
Use FormKiQ's KnowledgeBase offering to prepare a selected document collection for retrieval through a configured Amazon Bedrock knowledge base. Preserve source identity, update/deletion behavior, and citations. Enforce the caller's permitted document scope before retrieval and generation: FormKiQ permissions are not automatically copied into a vector store, and denying a citation afterward cannot undo a disclosed answer.
Define the collection and audience
Choose which sites, documents, revisions, and audiences belong in the collection. Exclude unapproved, deleted, staging, or restricted content unless the retrieval entitlement design explicitly handles it. Keep source document identity and revision metadata so answers can cite and open the correct source through authorized application paths.
The public v1.19.2 FormKiQ API does not expose KnowledgeBase management/query endpoints. Provisioning, ingestion, and the retrieval integration follow the selected module's delivered configuration. The documented query example uses the AWS Bedrock API after provisioning.
Test authorization before generation
Use a collection authorized in full for the caller, or a verified entitlement filter enforced by the query service. Do not trust arbitrary caller-supplied filters to restrict a privileged retrieval role. Test unauthorized users, revoked grants, changed folder restrictions, deleted documents, and stale chunks.
Show citations, missing-evidence outcomes, and freshness where useful. Protect retrieved snippets and generated answers as document-derived data. Reconcile permission and source changes across the entire retrieval path.
Edition and configuration
KnowledgeBase is a selected Advanced/Enterprise extension. Bedrock, vector storage, model/inference configuration, ingestion, authorization design, and AWS usage are part of the agreed deployment scope.