Direct Answer (TL;DR)
Yes. Brilo AI supports using Approved Answers so certain workflows only return pre-approved responses. Approved Answers let you lock an agent to a controlled set of answer templates (approved responses), use conservative fallback responses when no approved answer matches, and route uncertain cases to a human. When enabled, Approved Answers operate alongside confidence thresholds, session limits, and explicit escalation triggers to keep calls auditable and predictable.
Can only pre-approved responses be used in specific flows? — Yes. Brilo AI can be configured so selected workflows serve only Approved Answers and never generate free-form replies.
Do Approved Answers force human review on every uncertain call? — No. Brilo AI can auto-apply fallback phrasing or escalate based on configurable confidence thresholds and clarification limits.
Can I mix Approved Answers with dynamic data reads? — Yes. Brilo AI can merge approved templates with authorized data fields while preserving the approved phrasing.
Why This Question Comes Up (problem context)
Enterprises ask this because regulated teams need predictable, auditable voice replies. Security, legal, and compliance teams in healthcare, banking, and insurance want to avoid unscripted language that could misstate a policy, disclose protected information, or create record-keeping gaps. Buyers also need clarity about when the voice agent is allowed to improvise versus when it must use only approved language.
How It Works (High-Level)
Approved Answers is a configuration that restricts a voice agent’s output to a curated set of answer templates and response scripts. When a call enters a workflow that uses Approved Answers, the agent attempts to match intent and contextual data to an approved template before generating any free-form text. If no template matches, Brilo AI applies a configured fallback or triggers an escalation.
A fallback response is a short, conservative message the agent uses when no approved answer applies or confidence is low.
For implementation details about grounding and conservative fallbacks, see the Brilo AI article on preventing wrong or made-up answers: Brilo AI guidance on preventing wrong or made-up answers.
Related technical terms used: approved responses, answer templates, fallback responses, confidence threshold, escalation trigger, human handoff.
Guardrails & Boundaries
Approved Answers should be paired with explicit guardrails so the agent stays safe and auditable:
Only use Approved Answers in workflows that have a clearly documented scope (topics, allowed data reads, and prohibited actions).
Block free-form generation in Approved Answers workflows; never allow the agent to add unscripted material beyond filled data fields.
Use confidence thresholds and clarification limits: if intent matching confidence is below the threshold, the agent should use a fallback response or escalate to a human rather than guess.
Limit session persistence to avoid context drift across calls.
Record and audit all matches, template use, and escalations for compliance review.
An escalation trigger is a configurable rule that forces immediate human handoff when certain keywords, low confidence, or policy conditions occur. For recommended escalation and uncertainty behavior, see Brilo AI’s guidance on what happens when the AI is unsure: Brilo AI uncertainty & escalation guide.
Do not use Approved Answers as a substitute for missing policies: Approved Answers enforce approved phrasing but do not replace governance on which topics should be automated.
Applied Examples
Healthcare example — A hospital’s appointment-line workflow uses Approved Answers to provide only pre-approved scheduling confirmations and prep instructions. Brilo AI fills patient name, appointment time, and location into templates but never paraphrases clinical instructions. If the caller asks an out-of-scope clinical question, the agent issues a fallback response and routes the call to clinical staff.
Banking / Financial services example — A retail bank configures Approved Answers for routine balance and payment-status inquiries. Brilo AI uses approved scripts to disclose balances and next steps while pulling data from the authorized account record. Requests involving loan terms or dispute resolution trigger an escalation to a specialist.
Insurance example — A claim status workflow uses Approved Answers to read standardized claim statuses and next steps. If the caller asks about policy exclusions, the agent returns a fallback response and connects to a claims adjuster.
Human Handoff & Escalation
Brilo AI supports multiple handoff paths from Approved Answers workflows:
Warm transfer: transfer to a live agent with context and the last matched template shown to the human.
Silent queue: place the caller in a queue and create a human ticket with the transcript and approved-answer history.
Callback request: schedule a human callback when specialist review is required.
When Approved Answers are active, Brilo AI includes the matched template ID, confidence score, and the reason for any fallback in the handoff payload so the receiving human sees why the agent escalated.
Setup Requirements
Define: Create and approve the set of answer templates you want to use for the workflow (Approved Answers).
Configure: Apply the Approved Answers policy to the target Brilo AI voice agent workflow and mark it as template-only.
Integrate: Provide authorized data sources the agent may read for template fields (your CRM or your webhook endpoint).
Set: Establish confidence thresholds and clarification limits that trigger fallback responses or escalations.
Test: Validate common and edge-case call flows in a staging environment and review audit logs.
Enable: Turn on recording and audit logging for the workflow to capture template use and escalations.
Train: Share the approved templates and escalation rules with your human teams so handoffs are handled consistently.
For routing and end-to-end behavior when you enable template-only workflows, see Brilo AI’s end-to-end call handling guide: Brilo AI end-to-end call handling guide. For guidance on long or multi-turn conversations while using constrained responses, see: Brilo AI on handling long conversations.
Business Outcomes
Using Approved Answers reduces legal and compliance risk by limiting agent language to vetted scripts. It simplifies auditing because each spoken reply maps to a template ID and an audit trail. Operationally, Approved Answers can speed resolution for high-volume, low-risk interactions while ensuring that high-risk or ambiguous calls are escalated to humans with context.
FAQs
What happens if no approved answer matches the caller’s question?
Brilo AI uses the configured fallback response or applies the escalation trigger. The agent can attempt a single clarifying question if allowed by the workflow, but otherwise it will route the call to a human or schedule a callback.
Can Approved Answers include dynamic customer data?
Yes. In Brilo AI, templates can include placeholders for authorized data reads (for example, account balance or appointment time). The agent fills only the approved fields; templates control phrasing.
Are Approved Answers reversible if we want the agent to generate responses later?
Yes. You can disable the template-only policy on a workflow to allow generative replies. Brilo AI recommends re-testing and gradually expanding scope rather than switching broadly in production.
How are Approved Answers audited?
Brilo AI logs template IDs, timestamps, confidence scores, and any fallback or escalation events. Those logs can be used for compliance reviews and quality sampling.
Do Approved Answers affect transcription or recording?
No. Transcription and recording remain available per your workflow settings; audit logs will link recordings to the used template and handoff data.
Next Step