Skip to main content

Grounded Knowledge Index Setup Wizard

The Grounded Knowledge Index Setup Wizard guides you through creating new Grounded Knowledge Indexes by configuring the required Azure resources. Outermind provides two wizard variants: the Combined Wizard (recommended) for provisioning multiple index types at once, and a Single-Index Wizard for creating one index at a time.

Index Types

The wizard supports creating the following types of knowledge indexes:

TypePurposeTopics Step
Email ArchiveIndex and search historical email communicationsIncluded
Corporate KnowledgeIndex internal company documentation, policies, and proceduresIncluded
SharePoint DocumentsIndex documents from SharePoint Online libraries-
Google Drive DocumentsIndex documents from Google Workspace Shared Drives-
Activity Log SearchIndex agent execution activity summariesIncluded

Select the appropriate types based on what content you want to make searchable for your AI agents.

Prerequisites

Before starting the wizard, ensure you have:

  • An Azure account with an active subscription
  • Owner or Contributor access to create resources
  • Access to create Azure OpenAI and Azure AI Search services

The Combined Wizard allows you to provision multiple index types in a single session. Azure providers (OpenAI and AI Search) are shared across all selected indexes, reducing setup time.

Step 1: Welcome - Select Index Types

Select one or more index types to create. The wizard includes fast-path provider detection that automatically checks for existing Azure OpenAI and Azure AI Search providers in your account. If providers are already configured, the wizard can skip those setup steps entirely.

tip

If you have already created an index, the wizard detects your existing providers and lets you reuse them. You only need to configure providers once.

Step 2: Azure OpenAI Resource

Configure the Azure OpenAI resource that will power the AI capabilities. This resource provides:

  • LLM for answer synthesis - Reads retrieved documents and generates contextual answers
  • Embedding model - Converts documents into vectors for semantic search

Option A: Use Existing Provider If providers were detected in Step 1, your existing Azure OpenAI provider is pre-selected. You can proceed without re-entering credentials.

Option B: Create New Provider

  • Enter your Azure OpenAI endpoint URL
  • Enter your API key
  • Test the connection before proceeding

See the detailed Azure OpenAI setup guide for step-by-step instructions.

Step 3: Deploy Model

Select or configure the model deployment to use:

  • Deployment Name: The name of your deployed model (e.g., gpt-4.1-mini)
  • Model: The underlying model (e.g., GPT-4.1-mini, GPT-4o)

The wizard tests the connection to verify your credentials work with the specified deployment.

Step 4: Azure AI Search Service

Configure the Azure AI Search service that will store your document indexes:

Option A: Use Existing Service If detected, your existing Azure AI Search service is pre-selected with stored credentials.

Option B: Create New Service

  • Enter your search service endpoint URL
  • Enter your Admin API key
  • Test the connection before proceeding

See the detailed Azure AI Search setup guide for step-by-step instructions.

Step 5: Configure Indexes

For each selected index type, configure:

  • Name - Display name for your index (auto-populated based on index type)
  • Description - Optional description of what this index contains
  • Output Mode - How the AI should format responses (Answer Synthesis recommended)
  • Retrieval Reasoning Effort - How thoroughly the AI should analyze retrieved documents (Low, Medium, or High)

Advanced options specific to certain index types may also be available (e.g., enabling AI Knowledge Indexing for Corporate Knowledge indexes).

Step 6: Provisioning

The system sequentially provisions the required resources for each selected index:

  • Creates the Azure AI Search index
  • Configures the knowledge source
  • Sets up the knowledge base
  • Verifies all connections

Progress indicators show each step's status. If provisioning fails, you can view error details and retry.

Step 7: Grant Tool Access

After provisioning completes, you can bulk-grant access to the newly created knowledge index tools:

  • AICOS - Grant access to the AI Chief of Staff if configured
  • SMEs - Select which Specialist Agents should have access
  • Agents - Choose which standard agents can use these knowledge indexes

You can skip this step and configure tool access later from individual agent settings.

For Email Archive, Corporate Knowledge, and Activity Log Search type indexes, configure the topic taxonomy:

  • Select a Template - Choose from predefined topic structures
    • "Email Archive - Standard" for email indexes
    • "Corporate Knowledge - Standard" for corporate indexes (7 categories and 25+ subcategories)
    • Activity Log indexes use a default activity template applied automatically
  • Create Topics - Import topics from the selected template
  • Skip for Now - Configure topics later from the index details page

For Activity Log Search indexes, the wizard automatically applies the default activity topic template without requiring manual selection.

Step 9: Complete

Review the completed setup:

  • Index status and configuration summary for each provisioned index
  • Test each index with a sample query
  • Navigate to index details pages

Single-Index Wizard

The Single-Index Wizard follows the same steps but creates only one index at a time. Use this when you want to add a single additional index to your existing configuration.

The steps are identical to the Combined Wizard, but the Welcome step allows selecting only one index type.

After Provisioning

Once provisioning completes, your Grounded Knowledge Indexes will:

  1. Appear in the Knowledge Indexes list with status "Active"
  2. Be available as tools that can be assigned to agents
  3. Be ready to receive documents for indexing
  4. Be automatically configured in Account Settings:
    • Email Archive indexes are set as the Email History Index, allowing knowledge source scans to work immediately
    • Corporate Knowledge indexes are set as the Corporate Knowledge Index for SME access

Cost Considerations

Running a Grounded Knowledge Index involves Azure resource costs:

ResourceApproximate Cost
Azure OpenAI (GPT-4.1-mini)Pay-per-token usage
Azure AI Search (Basic)~$75/month
Azure AI Search (Standard)~$250/month

Choose the Azure AI Search tier based on your expected document volume and query frequency. Azure providers are shared across all indexes, so adding additional indexes does not require additional Azure resource subscriptions.