How to Connect SharePoint to Azure AI Search and Use It in Knots Flow

This guide walks you through connecting Azure AI Search to a SharePoint site to create a knowledge base that stays automatically up to date as your documents change. Once set up, you can connect it to Knots Flow and use it as a knowledge source in your AI Node.

For more details you can also refer to this official Azure AI Search article.

Requirements to follow this guide:

• A SharePoint Online site where you store your knowledge documents
• An Azure deployment of the "text-embedding-3-large" model

Create an app to allow the indexer to connect to SharePoint:

1. Go to https://portal.azure.com/#view/Microsoft_AAD_IAM/ActiveDirectoryMenuBlade/~/Overview
2. Click "Add" > "App registration"
3. Give it a name, select "Single tenant", leave the redirect URI empty, and click "Register"
4. In your app go to "Manage" > "API permissions"
5. Click "+ Add a permission" > Click "Microsoft Graph" > "Application permission"
6. Pick "Sites.Read.All" and "Files.Read.All", or "Sites.Selected" for more granular access control
7. Click "Add permissions" to confirm
8. Click "Grant admin consent for [your tenant name]"
9. In your app go to "Manage" > "Certificates & secrets"
10. Navigate to the "Client secrets" tab, click "+ New client secret"
11. Give it a description, pick an expiration date, and click "Add"
12. Copy the "Value" of your secret, you will need it later and it won't be shown again
13. In your app go to "Overview", copy "Application (client) ID" and "Directory (tenant) ID" for later

Create an AI search service:

1. Go to AI Search https://portal.azure.com/#view/Microsoft_Azure_ProjectOxford/CognitiveServicesHub/~/CognitiveSearch
2. Click Create
3. Give it a service name
4. Select a location close to Germany for better performance
5. Pick a pricing tier (At least "Basic")
6. Review + Create, then create the search service
7. Wait for the deployment to complete
8. Click "Go to resource"
9. Go to "Settings" > "Premium features" > Select the "Standard" plan

Create the AI knowledge source:

1. Go to "Agentic retrieval" > Knowledge source
2. Click "Add knowledge source"
3. Select "Microsoft SharePoint (Indexed)"
4. Fill in the name and description, select "Identity fields" for how to connect
5. SharePoint endpoint is the URL of your SharePoint site
6. Application ID is the Application (client) ID from the app you registered earlier
7. Application secret is the client secret you copied earlier
8. Tenant ID is the Directory (tenant) ID from the app you registered earlier
9. "Container name" allows you to pick the default library, all libraries, or a specific set using a query (see details here: https://learn.microsoft.com/en-us/azure/search/search-how-to-index-sharepoint-online#query)
10. "Mode": select "Standard" if your documents may include scanned pages, images, or tables
11. In "Enable text vectorization" click "Add vectorizer"
12. Select your text-embedding model (the one from the requirements) and click "Save"
13. Click "Create"

By default the indexer runs once a day. To configure the re-indexing schedule:

• Go to "Agentic retrieval" > "Knowledge sources" > select your source > pick a schedule > click "Save"

Gather the credentials for your search service:

• Go to "Search management" > "Indexes" and copy the index name
• Go to "Settings" > "Keys" and create or copy a key from the "Manage query keys" table
• Go to "Overview" and copy the "Url" field

Create a connector with Knots

• Navigate to this url: https://app.knots.io/modules/connectors
• In the "Add new connector" section select the system Azure ai search
• Pick a connector name and fill the form as indicated in this screenshot
• The vector field name and content field name created when following this guide are snippet_vector and snippet