This approach gives you the best of both worlds:

• Cognitive Search = retrieves relevant orders/comments from your database index.

• Azure OpenAI = interprets the user’s question, analyses retrieved documents, and generates a summarised or aggregated answer.

Flow:

1. User enters a natural language question in your frontend.

2. Cognitive Search retrieves the most relevant Orders.Comments.

3. The retrieved results are passed as context to Azure OpenAI (GPT model).

4. OpenAI generates a concise, human-friendly response (summary, count, or explanation).

Example C# Implementation (RAG Pattern):

using System;
using System.Linq;
using System.Threading.Tasks;
using Azure;
using Azure.Search.Documents;
using Azure.Search.Documents.Models;
using Azure.AI.OpenAI;class Program
{
    static async Task Main(string[] args)
    {
        // ====== Cognitive Search Setup ======
        string searchEndpoint = "https://<your-search-service>.search.windows.net";
        string searchApiKey = "<your-search-key>";
        string indexName = "orders";       

             var searchClient = new SearchClient(
            new Uri(searchEndpoint),
            indexName,
            new AzureKeyCredential(searchApiKey));       

             // ====== OpenAI Setup ======
        string openaiEndpoint = "https://<your-openai-resource>.openai.azure.com/";
        string openaiKey = "<your-openai-key>";
        string deploymentName = "<your-gpt-deployment>"; // e.g. "gpt-35-turbo"       

             var openaiClient = new OpenAIClient(new Uri(openaiEndpoint), new AzureKeyCredential(openaiKey));       

             // ====== User Input ======
        Console.WriteLine("Enter your question:");
        string userInput = Console.ReadLine();       

             // Step 1: Retrieve relevant documents from Cognitive Search
        var searchOptions = new SearchOptions()
        {
            Size = 5,
            QueryType = SearchQueryType.Semantic,
            QueryLanguage = "en-us"
        };
        searchOptions.Select.Add("orderId");
        searchOptions.Select.Add("comments");       

             var searchResults = searchClient.Search<SearchDocument>(userInput, searchOptions);       

             string retrievedText = string.Join("\n",
            searchResults.GetResults()
                         .Select(r => $"Order {r.Document["orderId"]}: {r.Document["comments"]}"));       

             // Step 2: Send retrieved docs + user query to Azure OpenAI
        string systemPrompt = @"
            You are a helpful assistant that analyzes order comments.
            Use the provided context to answer the user’s question.
            If context is insufficient, say you don’t know.
        ";        var chatOptions = new ChatCompletionsOptions()
        {
            Temperature = 0.2f,
            MaxTokens = 800
        };       

             chatOptions.Messages.Add(new ChatMessage(ChatRole.System, systemPrompt));
        chatOptions.Messages.Add(new ChatMessage(ChatRole.User,

             $"Question: {userInput}\n\nContext:\n{retrievedText}"));       

             var response = await openaiClient.GetChatCompletionsAsync(deploymentName, chatOptions);       

             // Step 3: Output final answer
        Console.WriteLine("\nAnswer:");
        Console.WriteLine(response.Value.Choices[0].Message.Content);
    }
}

Example Use Cases

User query:
“Show me all the orders where delivery is delayed due to an invalid address.”

• Cognitive Search: retrieves 5 orders with matching comments.

• OpenAI: formats a human-readable list (Order IDs + short summary).

User query:
“How many customers mentioned weather problems?”

• Cognitive Search: retrieves comments mentioning "weather".

• OpenAI: counts them and responds like: “3 orders reported weather-related delays.”

Benefits of RAG

• Keeps OpenAI responses grounded in your data (avoids hallucinations).

• Works on large databases by only sending relevant slices to OpenAI.

• Flexible: can answer fact-based queries (counts, filters) and generate natural summaries.