> ## Documentation Index
> Fetch the complete documentation index at: https://docs.shasta.health/llms.txt
> Use this file to discover all available pages before exploring further.

# Guidance

> How to configure AI assistant behavior and customer-specific instructions.

## Overview

Use **Guidance** to configure how the AI assistant should communicate, ask follow-up questions, route conversations, use knowledge sources, and produce outputs. Open **Guidance** from the Shasta provider portal sidebar, or go directly to [app.shasta.health/provider/guidance](https://app.shasta.health/provider/guidance).

Guidance is available to accounts with the required administrative access.

## Main guidance areas

* **Tasks** contains the primary guidance settings for assistant behavior.
* **Files** manages content and files the assistant can use.
* **Knowledge** controls communication style and knowledge instructions.
* **Outputs** configures structured information the assistant should extract.
* **Voices** manages voice-related assistant configuration.

Your account may show additional Guidance areas depending on your permissions and Shasta Health configuration.

## Tasks guidance

The main **Guidance** page includes sections for:

* **Basics**: tone and response length.
* **Communication style**: vocabulary, terms, and phrasing the assistant should use.
* **Context and clarification**: follow-up questions the assistant should ask.
* **Handover and escalation**: when the assistant should route directly to your team.
* **Content and sources**: when and how the assistant should use specific articles or sources.

Click a section to add or edit saved guidance. Guidance changes should be specific, concise, and written in the language you want the assistant to follow.

## Outputs guidance

Use **Outputs** when you want the assistant to extract structured fields from calls or conversations. These fields can appear in reports, including the Calls Report, and may power filters, charts, and call quality review workflows.

## Tips

* Use guidance for durable rules that should apply across many conversations.
* Keep instructions focused on what the assistant should do, say, collect, or avoid.
* Use handover guidance for scenarios that should go to staff instead of being handled by the AI.
* Review reports after changing guidance to confirm the assistant is behaving as expected.