Brilo AI Knowledge Base Import lets you ingest your existing documents, FAQs, and support articles so the Brilo AI voice agent can answer calls using your content. The import process accepts structured files (CSV/JSON), document collections (PDF, DOCX), and API/webhook feeds, then runs document parsing, embedding, and semantic indexing to make content searchable during calls. You can map fields to your CRM, set update schedules, and test answers before going live. For enterprise environments, plan content cleaning and access controls before import to protect sensitive data.

How do I bring my FAQ content into Brilo AI? — Import your FAQ as CSV or DOCX, map fields, and run a test ingest; Brilo AI will parse and index it.

Can Brilo AI read our support articles? — Yes. Upload documents or connect a webhook/API and enable content ingestion and semantic indexing.

What formats does Brilo AI accept for KB import? — Common formats include CSV, JSON, PDF, and DOCX, or live API/webhook feeds for continuous sync.

Enterprises already maintain large, regulated knowledge bases and want to reuse that content with Brilo AI voice agents to avoid reauthoring. Buyers ask about import because they need predictable behavior for call automation, clear data lineage for compliance, and low-friction integration with existing CRMs and ticketing systems. They also want to understand how Brilo AI handles updates, semantic search, and content pruning so agent answers stay accurate and auditable.

When you run a Knowledge Base Import, Brilo AI performs content ingestion, document parsing, and embedding generation to create a semantic index the voice agent uses at runtime. The high-level steps are: upload or connect your source, map metadata (title, topic, intent, confidence), run a controlled ingest, and validate sample conversations in the test console. Brilo AI supports bulk uploads and incremental sync from a webhook or API when enabled.

In Brilo AI, knowledge base import is an ingestion workflow that converts your documents into queryable vectors and metadata for the voice agent to use.

In Brilo AI, semantic indexing is the process that organizes embedded content so the voice agent can find the most relevant passages during a call.

Brilo AI enforces guardrails during import to limit what content the voice agent will surface. You configure allow-lists and deny-lists, answer confidence thresholds, and maximum passage lengths to reduce hallucination risk. Brilo AI will not publish imported content to callers until you run the validation step and flip the environment from staging to production.

An answer confidence threshold is a configuration that prevents low-confidence matches from being returned to callers; instead those calls can trigger a fallback or human handoff.

Do not import unvetted or raw PII/PHI without first applying your organization’s redaction and access policies. Brilo AI expects the customer to provide cleaned, permissioned content and to define retention rules and user access controls before live deployment.

Brilo AI voice agent workflows can escalate to a live agent or a separate workflow when configured. Typical handoff triggers include low answer confidence, caller requests for a live agent, or detection of regulatory-sensitive content. Handoff options include warm transfer with context (call summary and matched passages) or scheduled call-back by a human team. Configure the routing to include CRM case creation or a ticket with the matched KB passage so agents see the exact content that led to the transfer.

Importing an existing knowledge base into Brilo AI reduces time to value by reusing approved enterprise content, shortens agent handle time by surfacing exact passages, and improves caller experience through consistent answers. For regulated teams, imports enable traceability: every answer can be traced back to a specific document and ingestion timestamp for auditing and quality review.

What file formats can I import?

Brilo AI accepts common structured and document formats such as CSV, JSON, PDF, and DOCX for bulk import; you can also connect a webhook or API for continuous sync. Check your content for consistent fields and clean formatting before ingest.

How do I prevent the agent from sharing confidential content?

Apply deny-lists during import, remove or redact confidential fields before uploading, and set conservative confidence thresholds so uncertain matches trigger human handoff rather than direct answers.

How often should I re-import or sync my knowledge base?

Re-import frequency depends on how often content changes. Many teams run scheduled incremental syncs or enable webhook-driven updates for near real-time content refresh. Validate major changes in a staging environment before publishing.

Can Brilo AI map imported content to my CRM records?

Yes. During setup you can map imported metadata to your CRM fields so matched passages and the call summary can be attached to customer records. Provide your CRM mapping and endpoint details as part of the setup process.

What happens if the agent returns an incorrect answer from imported content?

If an incorrect answer is returned, update the source document or adjust mapping and confidence thresholds, then re-ingest the updated content. Use the test console to verify the corrected behavior before publishing.