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

# Responsible use

> Safety, privacy, and workplace-use boundaries for enterprise workspaces.

BehavIQ is designed to help people think through communication and behavioural-risk situations. It can help organise observations, draft language, and suggest next steps. It does not know the full context around a workplace event.

## Intended use

Use BehavIQ for support tasks such as:

* preparing for a difficult conversation
* reviewing a situation before escalating it through normal channels
* drafting calm, practical communication
* checking what information is missing
* using approved workspace knowledge-base material during chat

Users should still follow their organisation's policies, training, legal obligations, and safety procedures.

## Uses that need human review

Do not use BehavIQ as the only basis for:

* emergency response
* clinical or mental-health decisions
* legal advice
* HR, disciplinary, hiring, termination, rostering, or performance decisions
* safety-critical operational decisions
* covert monitoring or individual surveillance

If there is immediate danger, use emergency services and the organisation's incident procedure.

## What admins can see

Workspace admins can manage members, roles, invitations, SSO, SCIM, API access, knowledge-base files, usage, billing, settings, and audit logs.

The ordinary admin tools are not designed to give employers general access to every individual chat. Limited content or metadata review may still happen for support, security, abuse investigation, legal compliance, or a written enterprise process.

## Model and provider limits

AI responses can be wrong, incomplete, stale, or too generic for the situation. Users should treat responses as decision support, not instructions that override policy or human judgment.

Prompts, uploaded files, retrieved knowledge-base content, organisation instructions, and safety guidance may be sent to AI providers where needed to run the feature. Optional web search may send part of a request outside BehavIQ through the AI provider.

## Retention limits

Enterprise conversation retention can be indefinite or set to 30, 90, 180, or 365 days. A finite setting controls saved chat history in BehavIQ application storage. It does not automatically delete every provider record, audit log, billing record, security log, backup, or support record.

Knowledge-base files remain available until an admin removes them. Remove files when they should no longer influence answers.

## Pilot checklist

Before adding real workplace users, the organisation should have:

* a participant notice that explains what BehavIQ is for
* a clear list of prohibited uses
* an admin visibility statement
* a support and incident contact path
* a retention and offboarding plan
* a process for handling unsafe, incorrect, or concerning outputs
* a record of enabled features such as web search, uploads, SSO, SCIM, and API access

Keep this short enough that participants can read it. The point is clarity, not paperwork.