Simple Chat

Knowledge settings

What this group controls

Knowledge controls web research, URL access, Azure AI Search, extraction, chunking, multimodal processing, audio/video processing, and file sync.

Why it matters

These settings decide what evidence enters the system and how it becomes searchable. Endpoint, chunking, and sync choices affect answer quality, indexing cost, and document freshness.

Video planned

Knowledge settings walkthrough

Recording planned. Walk through each tab in the Knowledge group and explain when to change each setting.

To publish this video:

  1. Upload the recording to YouTube or Microsoft Stream.
  2. Save a poster frame as docs/images/video-posters/admin-knowledge.png.
  3. Add the watch URL to this slot in docs/_data/media.yml.

Before you change anything

  • Provision Azure AI Search and Document Intelligence before enabling dependent features.
  • Decide whether web, URL, and deep research access is approved.
  • Define allowed sync source types and workspace scopes.

Web & Research

Web Search

Web Search lets chat ground a single message in current public web results. SimpleChat does not call a search API directly. It calls an Azure AI Foundry agent that you create and configure, and that agent uses the Grounding with Bing Search tool to run the query and return results with citations.

Two things gate the capability, and both are required. enable_web_search turns it on, and web_search_consent_accepted records that an administrator acknowledged that Grounding with Bing Search moves data outside the Azure compliance boundary and operates under separate terms. Enabling the toggle without accepting the consent leaves the feature off.

Only the user’s current chat message is sent to the agent. Conversation history, workspace documents, attached file contents, and system prompts are never included in the outbound query. Decide your rollout policy on that basis, and use the notice settings below if you want users reminded of it in the composer.

An agent ID is not optional. If it is missing, chat tells the user web search is unavailable rather than silently answering from training data, so a half-finished configuration fails loudly instead of quietly degrading.

URL Access

The URL Access section belongs to the Web & Research tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.

Deep Research

The Deep Research section belongs to the Web & Research tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.

Settings

Setting What it does Default Notes
Enable Web Search via Foundry Agent Adds web search through the configured Azure AI Foundry agent for approved chat flows. Requires accepted consent to take effect. Off enable_web_search; capability toggle
Show data notice to users when web search is used Shows a dismissible banner above the chat composer while web search is active, so users know the message will leave the tenant. Off enable_web_search_user_notice; capability toggle
Notice Text The banner wording. Shown once per browser session, from the first time the user activates web search until they dismiss it. N/A (runtime control) web_search_user_notice_text
Foundry Project Endpoint Project endpoint format: https://.services.ai.azure.com/api/projects/ (not the inference endpoint). N/A (runtime control) web_search_foundry_endpoint
Foundry API Version Pins the service API version SimpleChat sends with requests for this feature. N/A (runtime control) web_search_foundry_api_version
Foundry Agent ID Identifies the agent that carries the Grounding with Bing Search tool. Without it, chat reports web search as unavailable. N/A (runtime control) web_search_foundry_agent_id
Authentication Type Selects how SimpleChat authenticates to the Foundry project. The identity must have Cognitive Services User and AI Developer roles on that project. N/A (runtime control) web_search_foundry_auth_type
Cloud Selects the Azure cloud whose endpoints and authority are used to reach the Foundry project. N/A (runtime control) web_search_foundry_cloud
Managed Identity Type Chooses between the system-assigned identity and a user-assigned identity when authenticating with a managed identity. N/A (runtime control) web_search_foundry_managed_identity_type
Authority Endpoint (Custom Cloud) Overrides the login authority when the selected cloud is a custom or sovereign environment. N/A (runtime control) web_search_foundry_authority
Managed Identity Client ID (UAMI) Identifies which user-assigned managed identity to authenticate with. N/A (runtime control) web_search_foundry_managed_identity_client_id
Tenant ID Directory the service principal authenticates against. N/A (runtime control) web_search_foundry_tenant_id
Client ID Application ID of the service principal used for authentication. N/A (runtime control) web_search_foundry_client_id
Client Secret Provides the secret credential used when the selected authentication mode requires one. N/A (runtime control) web_search_foundry_client_secret
Enable URL Access for chat and workflows Allows chat and workflows to inspect user-provided URLs within the configured URL limits and domain policy. Off enable_url_access; capability toggle
Require UrlAccessUser App Role Required app role value: UrlAccessUser. Assign this role to users or groups in the Enterprise App before enabling the requirement. When enabled, only assigned users can use URL Access in chat or enable it for workflows. Off require_member_of_url_access_user
Chat URL Limit Hard limit: 100 direct URLs per chat message. 10 url_access_max_chat_urls_per_turn
Workflow URL Limit Hard limit: 500 direct URLs per workflow prompt. 50 url_access_max_workflow_urls_per_run
Url Access Allowed Domains Lists the approved IDs, domains, groups, workspaces, or sources that may use this feature. Empty list url_access_allowed_domains
Allowed Domains Lists the approved IDs, domains, groups, workspaces, or sources that may use this feature. N/A (runtime control) url_access_allowed_domains_new
Url Access Blocked Domains Lists the domains, users, or destinations that this feature must not use. Empty list url_access_blocked_domains
Blocked Domains Lists the domains, users, or destinations that this feature must not use. N/A (runtime control) url_access_blocked_domains_new
Enable Deep Research for chat Enables Deep Research in chat so SimpleChat can inspect search results and linked source pages within the configured crawl limits. Off enable_source_review; capability toggle
Allow internal network hostnames Allows DNS hostnames that resolve to private/internal addresses. Literal IP URL targets, localhost, metadata hosts, link-local addresses, and reserved addresses remain blocked. Off source_review_allow_internal_hosts
Activation Mode Defines behavior for the related admin workflow; verify the affected feature after saving. manual source_review_default_mode
Max Pages per Turn Hard limit: 10 pages. 10 source_review_max_pages_per_turn
Max Seed Pages per Turn Limits initial search-result and direct URL pages so budget remains for child pages. 10 source_review_max_seed_pages_per_turn
Timeout per Turn Hard limit: 30 seconds. 30 source_review_timeout_seconds
Max Redirects Every redirect target is revalidated. 5 source_review_max_redirects
Max MB per Page Hard limit: 5 MB. 5 MB source_review_max_bytes_per_page converted from source_review_max_bytes_per_page bytes
Source Traversal Depth Depth 2 follows selected links from seed and child pages. 2 source_review_max_depth
Inspect linked source pages Exposes the capability after required services, permissions, and rollout policy are ready. On enable_deep_source_review; capability toggle
Use model-assisted source link planning Defines behavior for the related admin workflow; verify the affected feature after saving. On source_review_enable_llm_planning
Allow JavaScript rendering fallback Defines behavior for the related admin workflow; verify the affected feature after saving. On source_review_allow_js_rendering
Rendered Load More Clicks When JavaScript rendering is enabled, Deep Research can click visible Load More controls until this cap is reached. 12 source_review_js_load_more_clicks
Respect robots.txt Defines behavior for the related admin workflow; verify the affected feature after saving. On source_review_respect_robots_txt
Log Deep Research activity Defines behavior for the related admin workflow; verify the affected feature after saving. On source_review_audit_logging
Test Prompt Narrows the admin list shown for test prompt. N/A (runtime control) web_search_test_query
URL Defines behavior for the related admin workflow; verify the affected feature after saving. Not specified in defaults url_access_policy_test_url

Search Index

Azure AI Search

The Azure AI Search section belongs to the Search Index tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.

Settings

Setting What it does Default Notes
Use APIM instead of direct Azure AI Search Routes Azure AI Search calls through APIM instead of calling the Search endpoint directly. Off enable_ai_search_apim; capability toggle
Search Endpoint Provides the endpoint or route SimpleChat uses for this service. Empty azure_ai_search_endpoint
Authentication Type Chooses whether SimpleChat authenticates to this service with a key, managed identity, or another supported method. key azure_ai_search_authentication_type
Search Key Provides the secret credential used when the selected authentication mode requires one. Empty azure_ai_search_key
Azure APIM AI Search Endpoint Provides the endpoint or route SimpleChat uses for this service. Empty azure_apim_ai_search_endpoint
Azure APIM AI Search Subscription Key Provides the secret credential used when the selected authentication mode requires one. Empty azure_apim_ai_search_subscription_key

Document Extraction

Document Intelligence

The Document Intelligence section belongs to the Document Extraction tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.

Chunk Sizes

Documents are split into chunks before they are indexed, and each chunk is embedded as a single request. That makes chunk size a retrieval decision and a hard technical limit at the same time: smaller chunks return more precise citations, larger chunks keep more surrounding context in one result, and a chunk that does not fit in the embedding model’s context window cannot be indexed at all.

Because of that limit, overrides are capped per unit rather than by one shared number. Word fields and character fields have different ceilings, both derived from the embedding model’s context window, and the tab shows the current values. A value above the ceiling is reduced on save and the page reports which fields were changed.

Page and slide counts are structural: how much text a page holds is not known until extraction runs, so they are not capped here. If an extracted chunk still turns out to be too large to embed, its text is stored and remains searchable and citable while only the portion used to compute its vector is trimmed, and the event is logged.

Custom sizes apply to new uploads only. Existing documents keep the chunks they were indexed with until they are uploaded again.

Metadata Extraction

The Metadata Extraction section belongs to the Document Extraction tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.

Multi-Modal Vision Analysis

The Multi-Modal Vision Analysis section belongs to the Document Extraction tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.

Settings

Setting What it does Default Notes
Require DeepResearchUser App Role Required app role value: DeepResearchUser. Assign this role to users or groups in the Enterprise App before enabling the requirement. When enabled, only assigned users can use Deep Research. Off require_member_of_deep_research_user
Max User URLs per Turn Direct URLs beyond this cap are recorded as omitted in the ledger. 100 deep_research_max_user_urls_per_turn
Max Search Queries per Turn Includes the original current-message query. 8 deep_research_max_search_queries_per_turn
Plan multiple web search queries Narrows the admin list shown for plan multiple web search queries. On deep_research_enable_query_planning
Save research ledger artifacts Narrows the admin list shown for save research ledger artifacts. On deep_research_enable_ledger_artifact
Enable Enhanced extraction Enables the enhanced extraction path for richer PDF and image structure when the required services are configured. Off enable_enhanced_extraction; capability toggle
PDF and Image Extraction Mode Enhanced captures more document detail for PDFs and images, including tables, page structure, and checked or unchecked marks. It adds latency and has a 6X increase for every 1000 pages when selected. read document_intelligence_pdf_image_extraction_mode
Auto Sample Pages Auto samples this many first PDF pages with Document Intelligence Layout. If it detects tables, selection marks, or figures, the full PDF uses Enhanced; otherwise it finishes with Standard. Images use Enhanced in Auto mo Not specified in defaults document_intelligence_auto_sample_pages
Extract mathematical formulas Exposes the capability after required services, permissions, and rollout policy are ready. Off enable_document_intelligence_formula_extraction; capability toggle
Foundry Endpoint Your Microsoft Foundry resource endpoint, without a trailing path. Empty azure_content_understanding_endpoint
Authentication Type Managed identity requires the Cognitive Services User role on the Foundry resource. key azure_content_understanding_authentication_type
Content Understanding Key Provides the secret credential used when the selected authentication mode requires one. Empty azure_content_understanding_key
API Version Default: Not specified in defaults azure_content_understanding_api_version
Document Analyzer Default: Not specified in defaults azure_content_understanding_analyzer_id
Image Analyzer Default: Not specified in defaults azure_content_understanding_image_analyzer_id
Analyze images embedded in DOCX and PPTX files Exposes the capability after required services, permissions, and rollout policy are ready. On enable_office_embedded_image_analysis; capability toggle
Minimum Image Size (pixels) Images narrower or shorter than this are skipped as icons or spacers. Not specified in defaults office_embedded_image_min_pixels
Maximum Images Per Document Caps per-document cost. Duplicate images are analyzed once. Not specified in defaults office_embedded_image_max_per_document
Use APIM instead of direct Document Intelligence endpoint Routes Document Intelligence calls through APIM instead of calling the service endpoint directly. Off enable_document_intelligence_apim; capability toggle
Document Intelligence Endpoint Provides the endpoint or route SimpleChat uses for this service. Empty azure_document_intelligence_endpoint
Authentication Type Chooses whether SimpleChat authenticates to this service with a key, managed identity, or another supported method. key azure_document_intelligence_authentication_type
Document Intelligence Key Provides the secret credential used when the selected authentication mode requires one. Empty azure_document_intelligence_key
Azure APIM Document Intelligence Endpoint Provides the endpoint or route SimpleChat uses for this service. Empty azure_apim_document_intelligence_endpoint
Azure APIM Document Intelligence Subscription Key Provides the secret credential used when the selected authentication mode requires one. Empty azure_apim_document_intelligence_subscription_key
Enable custom chunk sizes by file type Exposes the capability after required services, permissions, and rollout policy are ready. Off enable_chunk_size_override; capability toggle
Summarize history for search context Adds a model-generated summary of recent conversation history into search context when hybrid document search is active, so follow-up searches can use prior conversational context. Off enable_summarize_content_history_for_search
Summarize older history beyond conversation limit Summarizes older conversation messages that fall outside the configured conversation-history window so long chats can retain condensed context instead of dropping all older content. Off enable_summarize_content_history_beyond_conversation_history_limit
TXT (words) Defines a capacity or timing boundary that keeps the feature inside supported limits. 400 words chunk_size_txt
LOG (words) Defines a capacity or timing boundary that keeps the feature inside supported limits. 1000 words chunk_size_log
DOC (words) Defines a capacity or timing boundary that keeps the feature inside supported limits. 400 words chunk_size_doc
DOCM (words) Defines a capacity or timing boundary that keeps the feature inside supported limits. 400 words chunk_size_docm
DOCX (words) Defines a capacity or timing boundary that keeps the feature inside supported limits. configured WORD_CHUNK_SIZE words chunk_size_docx
HTML (words) Minimum enforced at 50% of target on merge. 1200 words chunk_size_html
Markdown (words) Target words per chunk. Heading sections larger than this are split, so a long section under one heading cannot become a single unindexable chunk. Minimum enforced at 50% of target on merge. 1200 words chunk_size_md
XML (characters) Defines a capacity or timing boundary that keeps the feature inside supported limits. 4000 characters chunk_size_xml
YAML (characters) Defines a capacity or timing boundary that keeps the feature inside supported limits. 4000 characters chunk_size_yaml
YML (characters) Defines a capacity or timing boundary that keeps the feature inside supported limits. 4000 characters chunk_size_yml
JSON (characters) Defines a capacity or timing boundary that keeps the feature inside supported limits. 4000 characters chunk_size_json
Transcripts (words) Applies to new audio transcripts. 400 words chunk_size_transcript
PDF (pages) Pages per chunk after extraction. 1 page chunk_size_pdf
PPT/PPTX (slides) Slides per chunk after extraction. 1 slide chunk_size_pptx

Audio & Video

AI Video Intelligence

The AI Video Intelligence section belongs to the Audio & Video tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.

AI Voice Conversations

The AI Voice Conversations section belongs to the Audio & Video tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.

Settings

Setting What it does Default Notes
Enable Video File Upload & Processing Allows users to upload video files for processing through the configured Video Indexer resource. Off enable_video_file_support; capability toggle
Cloud / Endpoint Mode Choose the endpoint family that matches your deployed cloud. Use Custom only when you need a non-standard Video Indexer endpoint. Not specified in defaults video_indexer_cloud
Custom API Endpoint Provides the endpoint or route SimpleChat uses for this service. Not specified in defaults video_indexer_custom_endpoint
Effective API Endpoint Provides the endpoint or route SimpleChat uses for this service. Not specified in defaults video_indexer_endpoint_display
Resource Group * The Azure resource group containing your Video Indexer account Empty video_indexer_resource_group
Subscription ID * Your Azure subscription ID Empty video_indexer_subscription_id
Account Name * The name of your Video Indexer account resource Empty video_indexer_account_name
Location * Azure region where your Video Indexer account is deployed (e.g., eastus, westus2, northeurope) Empty video_indexer_location
Account ID * Found in the Video Indexer account Overview page in Azure Portal Empty video_indexer_account_id
ARM API Version Default for : Not specified in defaults video_indexer_arm_api_version
Timeout (seconds) Defines a capacity or timing boundary that keeps the feature inside supported limits. 600 video_index_timeout
Enable AI Response Completion Audio Cues Exposes the capability after required services, permissions, and rollout policy are ready. Off enable_chat_completion_audio_cues; capability toggle
Enable Audio File Upload & Processing Allows users to upload audio files for transcription through the configured Speech service. Off enable_audio_file_support; capability toggle
Enable Voice Input (Speech-to-Text) Shows voice input controls in chat and sends captured speech to the configured Speech service. Off enable_speech_to_text_input; capability toggle
Enable Voice Responses (Text-to-Speech) Allows voice responses from chat output through the configured Speech service. Off enable_text_to_speech; capability toggle
Endpoint Use the resource-specific custom-domain endpoint when selecting Managed Identity. Empty speech_service_endpoint
Location Required for speech recognition locale defaults and for text-to-speech when using Managed Identity. Empty speech_service_location
Speech Subscription ID Defines behavior for the related admin workflow; verify the affected feature after saving. Empty speech_service_subscription_id
Speech Resource Group Defines behavior for the related admin workflow; verify the affected feature after saving. Empty speech_service_resource_group
Speech Resource Name If you use a custom-domain Speech endpoint, this is usually the first part of that hostname. Empty speech_service_resource_name
Speech Resource ID Provide Subscription ID, Resource Group, and Speech Resource Name to auto-build the ARM resource ID. Empty speech_service_resource_id
Locale Defines behavior for the related admin workflow; verify the affected feature after saving. en-US speech_service_locale
Authentication Type Chooses whether SimpleChat authenticates to this service with a key, managed identity, or another supported method. key speech_service_authentication_type
API Key Provides the secret credential used when the selected authentication mode requires one. Empty speech_service_key

File Sync

File Sync

The File Sync section belongs to the File Sync tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.

Visible Source Types

The Visible Source Types section belongs to the File Sync tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.

Personal Workspace Sync

The Personal Workspace Sync section belongs to the File Sync tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.

Group Workspace Sync

The Group Workspace Sync section belongs to the File Sync tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.

Public Workspace Sync

The Public Workspace Sync section belongs to the File Sync tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.

Settings

Setting What it does Default Notes
Enable File Sync Enables the File Sync feature so configured external sources can import files into workspaces. Off enable_file_sync; capability toggle
Max Sources Defines a capacity or timing boundary that keeps the feature inside supported limits. 10 file_sync_max_sources_per_scope
Min Schedule Minutes Defines a capacity or timing boundary that keeps the feature inside supported limits. 15 file_sync_min_schedule_interval_minutes
Max Files Per Run Defines a capacity or timing boundary that keeps the feature inside supported limits. 1000 file_sync_max_files_per_run
Max GB Per Run Defines a capacity or timing boundary that keeps the feature inside supported limits. 5 GB file_sync_max_gb_per_run
Max Concurrent Runs Defines a capacity or timing boundary that keeps the feature inside supported limits. 2 file_sync_max_concurrent_runs
Allow Recursive Sources Defines behavior for the related admin workflow; verify the affected feature after saving. On file_sync_allow_recursive_sources
SMB Share Available now. On file_sync_visible_source_types
OneDrive Coming Soon. Off file_sync_visible_source_type_onedrive
On-prem SharePoint Coming Soon. Off file_sync_visible_source_type_sharepoint_on_prem
Google Workspace Coming Soon. Off file_sync_visible_source_type_google_workspace
Enable personal sync Exposes the capability after required services, permissions, and rollout policy are ready. On enable_file_sync_personal; capability toggle
Admins manage sources only Defines behavior for the related admin workflow; verify the affected feature after saving. Off file_sync_personal_admin_only
Require PersonalFileSyncUser App Role Defines behavior for the related admin workflow; verify the affected feature after saving. Off file_sync_personal_require_app_role
Manage User Sources Defines behavior for the related admin workflow; verify the affected feature after saving. Not specified in defaults Runtime UI control
Enable group sync Exposes the capability after required services, permissions, and rollout policy are ready. On enable_file_sync_group; capability toggle
Admins manage sources only Defines behavior for the related admin workflow; verify the affected feature after saving. Off file_sync_group_admin_only
Require Group Assignment to Use File Sync Defines behavior for the related admin workflow; verify the affected feature after saving. Off require_group_assignment_for_file_sync
File Sync Allowed Group Ids Lists the approved IDs, domains, groups, workspaces, or sources that may use this feature. Empty list file_sync_allowed_group_ids
Manage Group Sources Defines behavior for the related admin workflow; verify the affected feature after saving. Not specified in defaults Runtime UI control
Enable public sync Exposes the capability after required services, permissions, and rollout policy are ready. Off enable_file_sync_public; capability toggle
Admins manage sources only Defines behavior for the related admin workflow; verify the affected feature after saving. Off file_sync_public_admin_only
Require Public Workspace Assignment to Use File Sync Defines behavior for the related admin workflow; verify the affected feature after saving. Off require_public_workspace_assignment_for_file_sync
File Sync Allowed Public Workspace Ids Lists the approved IDs, domains, groups, workspaces, or sources that may use this feature. Empty list file_sync_allowed_public_workspace_ids
Manage Public Workspace Sources Defines behavior for the related admin workflow; verify the affected feature after saving. Not specified in defaults Runtime UI control
Search Groups Defines behavior for the related admin workflow; verify the affected feature after saving. N/A (runtime control) Runtime UI control
Search Public Workspaces Defines behavior for the related admin workflow; verify the affected feature after saving. N/A (runtime control) Runtime UI control

Common tasks

  1. Configure document grounding. Set search and extraction dependencies, upload a small document, and ask a grounded question. Outcome to verify: The answer cites indexed content.
  2. Enable approved research. Enable only allowed web, URL, or deep research routes and test allowed and blocked cases. Outcome to verify: Research follows policy.
  3. Publish file sync. Enable sync, choose source types and scopes, then run a small source. Outcome to verify: Files appear in the intended workspace.

Troubleshooting

Symptom Likely cause Fix
A synced file is not searchable The source ran but extraction or indexing failed later. Check sync state, extraction settings, and Azure AI Search before rerunning.