AInfraMinds

🧠 What is Custom Question Answering?

Custom Question Answering (CQA) is a feature under Azure AI Language Service that allows you to build a knowledge-based Q&A system. Instead of writing code or training ML models, you simply provide documents or URLs — Azure does the rest.

Think of it like building your own version of ChatGPT — but trained on your company’s internal documents, policies, FAQs, or even PDFs.

📦 What You’ll Build

A simple Question Answering model that:

• Ingests your documents (PDFs, Word, or URLs)
• Extracts Q&A pairs using Azure’s built-in AI
• Can be queried through a web interface or REST API

🧰 Prerequisites

• An Azure subscription
• Contributor/Owner access to create resources
• Some personal documents to test with (PDFs, FAQs, etc.)

🚀 Step-by-Step Setup

🔹 Step 1: Create an Azure AI Language Resource

1. Go to Azure Portal
2. Search for Language and choose Azure AI Language
3. Click Create
4. Fill in the basics:
   • Subscription & Resource Group
   • Region: e.g., East US
   • Name: language-cqa-demo
   • Pricing Tier: Choose F0 (Free) or S tier
5. Click Review + Create → then Create

🔹 Step 2: Create a Custom Question Answering Project

1. Go to the Language Studio
2. Sign in with the same Azure account
3. Click Custom Question Answering under the “Build” section
4. Choose your newly created resource
5. Click + New Project
   • Project Name: faq-ai-bot
   • Language: English
   • Description: “My first custom AI bot”
   • Click Create

🔹 Step 3: Add Knowledge Sources

You can add sources by:

📁 Uploading files:

• PDFs, DOCX, TXT files
• Click + Add Source → Files
• Upload your own company FAQ or document

🌐 Adding URLs:

• Add a webpage (like your company policy or GitHub Wiki)
• Language Studio will auto-scrape Q&A pairs

✍️ Manual Q&A:

• Add custom question-answer pairs directly
• Helpful for adding targeted responses

🔹 Step 4: Train the Project

1. Once your sources are added, click Train
2. Wait a few seconds — training is fast!
3. You’ll get a notification when training completes

🔹 Step 5: Test Your Model

1. Go to the Test tab
2. Ask questions based on your content:
   • “What is our refund policy?”
   • “How do I reset my password?”
3. See how your model responds!
4. Tweak answers as needed by editing the extracted Q&A pairs

🔹 Step 6: Deploy the Project

1. Click the Deploy tab
2. Click Deploy project
3. This will create a public endpoint and assign a deployment name (e.g., faq-bot-prod)

🔑 Using the API

Once deployed, you can call your Q&A bot via REST API:

POST https://<your-resource-name>.cognitiveservices.azure.com/language/:query-knowledgebases?projectName=faq-ai-bot&deploymentName=faq-bot-prod&api-version=2021-10-01

Add headers:

    Ocp-Apim-Subscription-Key: <your-key>
    Content-Type: application/json

And body:

    {
        "question": "How do I reset my password?"
    }

💡 Real World Use Cases

• Internal IT helpdesk bot
• HR FAQ bot for employees
• Customer service agent for common questions
• Secure chatbot trained on policy docs

📎 Resources

• Get started with Azure Cognitive Services for Language
• What is custom question answering?
• Azure AI Language pricing
• Tutorial: Create a FAQ bot

🧠 Final Thoughts

If you’re new to Azure or AI, this is a perfect hands-on start. No code, no hassle — just smart responses powered by your own data.