Deploying a Smart Documentation Wiki with Azure AI Foundry and Private App Integration

What You’re Building ?

You’ll be creating a Wiki Knowledge Hub that:

• Stores technical and functional documentation in Blob Storage or SharePoint

• Uses Azure AI Foundry to index, version, and retrieve content

• Enables semantic search and generative Q&A using Azure AI Search

• Optionally integrates with Copilot Studio or Power Pages for front-end access

Step-by-Step Implementation

1. Organize Documentation in Blob Storage or SharePoint

Option A: Azure Blob Storage

• Create a Blob Storage account in Azure

• Organize documents by project, module, or domain

• Use folders for logical grouping (e.g., /Finance/FunctionalSpecs, /Tech/IntegrationDocs)

• Supported formats: PDF, DOCX, TXT, CSV, Markdown

Option B: SharePoint

• Create a SharePoint site for each project or domain

• Upload documents into document libraries

• Ensure metadata tagging (e.g., author, version, module) for better indexing

🔗 Reference: Introduction to Azure Blob Storage

2.  Set Up Azure AI Foundry Hub-Based Project

• Go to Azure Portal → Create a hub-based Azure AI Foundry project

• Ensure you have an active Azure subscription

• This hub will manage your data versions, lineage, and reproducibility

🔗 Reference: Create hub-based project in Azure AI Foundry

3. Add Data Sources to Azure AI Foundry

• In Foundry, go to Data → Add New Data

• Choose Blob Storage or SharePoint as source

• Select type: folder, file, or table

• Assign friendly names and version tags (e.g., FinanceSpecs_v1, TechDocs_2025Q3)

• Foundry supports versioning, auditability, and lineage tracking

🔗 Reference: Manage data in Azure AI Foundry

4. Configure Azure AI Search for Semantic Indexing

• Create an Azure AI Search service

• Connect it to your Blob Storage or SharePoint container

• Define index schema:

  • content (string, required)

  • title, tags, project, author, version (optional metadata)

• Run the indexer and validate that documents are searchable

🔗 Reference: Set up Azure AI Search with Blob Storage

5. Enable Generative Q&A (Optional)

• Use Azure OpenAI or Foundry Prompt Flow to build Q&A agents

• Connect the indexed data source to your agent

• Use embeddings and retrieval-augmented generation (RAG) for contextual answers

• Deploy via Copilot Studio, Power Pages, or Teams

6. Document the Wiki Creation Process

To ensure plagiarism-free documentation:

• Cite all Microsoft Learn and Azure Docs links used

• Use paraphrased summaries with references

• Include diagrams showing architecture (e.g., Blob → Foundry → Search → Q&A Agent)

• Add version history and author attribution for each document

References

• Azure AI Foundry: Add and manage data

• Azure AI Search with Blob Storage

• Azure AI Foundry GitHub Docs

Deploying the Wiki via a Private App

You can use Power Pages, Copilot Studio, or a custom web app to surface the Wiki. Here's how:

🔹 Option A: Power Pages (Low-Code Portal)

1. Create a Power Pages site

   • Go to Power Pages

   • Choose a template (e.g., Knowledge Base or Custom Portal)

2. Connect to Azure AI Search

   • Use REST API or embed a search component

   • Authenticate using Azure AD or API keys

3. Integrate Generative Q&A

   • Use Azure OpenAI or Prompt Flow to create a Q&A agent

   • Embed it via iframe or custom HTML component

4. Secure Access

   • Use role-based access control (RBAC)

   • Enable login via Microsoft Entra ID (formerly Azure AD)

🔹 Option B: Copilot Studio (Conversational Interface)

1. Create a Copilot in Copilot Studio

   • Go to Copilot Studio

   • Choose “Custom Copilot” and define intents like “Search Wiki”, “Ask about Finance Docs”

2. Connect to Azure AI Search and Foundry

   • Use Data Plugin to connect to indexed content

   • Use Generative Answers to enable semantic Q&A

3. Deploy to Teams, Web, or Mobile

   • Publish Copilot to Teams or embed in internal web apps

   • Use adaptive cards for rich responses

Security & Governance

• Use Microsoft Entra ID for authentication

• Enable audit logging in Azure AI Foundry

• Apply data sensitivity labels in SharePoint or Blob metadata