Advanced Settings

Customize AI behavior, model selection, function schemas, and advanced parameters

Assistant Type

Select which assistant to configure

Configure the SharePoint voice and text assistant for CRM data.

AI Instructions

System Instructions

Define how the AI assistant should behave and respond. Used by text chat and as fallback for voice.

Voice-specific Instructions (Realtime)

These instructions are used only for the Realtime (voice) assistant. Keep them short and very explicit.

Model Parameters

Text Chat Temperature: 0.8

Controls randomness for text chat responses. Lower values are more focused, higher values are more creative.

Voice Activity Detection (VAD)

Threshold: 0.5

Activation threshold for voice detection (0-1).

Prefix Padding: 300 ms

How much audio before speech detection to include.

Silence Duration: 500 ms

How long to wait before detecting end of speech.

Power Automate Integration

Configure your Power Automate flow URLs and function definitions. Define the input parameter schemas and descriptions for each function.

Fetch SharePoint Data - Flow URL (D1 Flow)

Power Automate flow URL for fetching SharePoint data.

Function Description

Tell the AI what this function does and when to call it.

Fetch SharePoint Data - Input Parameters Schema

JSON schema defining input parameters the AI can send to this flow.

Perform SharePoint Action - Flow URL (D2 Flow)

Power Automate flow URL for performing SharePoint actions.

Function Description

Tell the AI what this function does and when to call it.

Perform SharePoint Action - Input Parameters Schema

JSON schema defining input parameters the AI can send to this flow.

Manage SharePoint Item - Flow URL (D3 Flow)

Power Automate flow URL for managing SharePoint items.

Function Description

Tell the AI what this function does and when to call it.

Manage SharePoint Item - Input Parameters Schema

JSON schema defining input parameters the AI can send to this flow.

Text Chat

Enable Text Chat Interface

When enabled, adds a text input area to the assistant page for typing messages instead of using voice.

Text Chat Model

The main model for text chat. With cost optimization enabled, simple queries are routed to cheaper models automatically.

Cost Optimization

Cost optimization can reduce your API costs by 50-80% by intelligently routing simple queries to cheaper models and caching SharePoint data.

Text Chat

Enable Cost Optimization for Text Chat

When OFF, uses original behavior with your configured model for all queries.

Optimization Level

SharePoint Voice Assistant

Enable Cost Optimization for SharePoint Voice

When OFF, uses original behavior without session tracking or cost monitoring.

Optimization Level

Enable Smart Filtering for Voice Queries

When ON, filters SharePoint records before sending to GPT (faster responses, lower token usage). When OFF, sends all records (current behavior, useful for complex queries).

Estimated Cost Comparison (per 1000 queries)

Without optimization (gpt-5.1):~$15-20

With balanced optimization:~$4-6

With aggressive optimization:~$2-4

Voice sessions: ~$0.30/min (standard) or ~$0.10/min (mini model)

Back to SharePoint Assistant