Build an AI-Powered Excel Chatbot in C#: Turn Spreadsheets into Smart Assistants

private static string BuildCsvContext(string excelFilePath)
{
    // Initialize Syncfusion Excel engine
    using ExcelEngine excelEngine = new ExcelEngine();

    // Set default version to XLSX
    IApplication excelApp = excelEngine.Excel;
    excelApp.DefaultVersion = ExcelVersion.Xlsx;

    // Open the Excel file
    using FileStream fileStream = File.OpenRead(excelFilePath);
    IWorkbook workbook = excelApp.Workbooks.Open(fileStream);

    // Build CSV-like context
    StringBuilder stringBuilder = new StringBuilder();

    // Add file and sheet info
    stringBuilder.AppendLine($"File: {excelFilePath}");
    stringBuilder.AppendLine($"Sheets: {workbook.Worksheets.Count}");
    stringBuilder.AppendLine();

    // Convert each worksheet to CSV format and append to the context
    foreach (IWorksheet worksheet in workbook.Worksheets)
    {
        MemoryStream csvStream = new MemoryStream();

        // Save workbook as CSV
        worksheet.SaveAs(csvStream, ",");

        // Convert CSV to text
        string excelData = Encoding.UTF8.GetString(csvStream.ToArray());
        stringBuilder.AppendLine("Sheet Name : " + worksheet.Name);
        stringBuilder.AppendLine();
        stringBuilder.AppendLine(excelData);
        stringBuilder.AppendLine();
    }

    return stringBuilder.ToString();
}

private static async Task<string> AskOpenAIAsync(
    string apiKey,
    string model,
    string systemPrompt,
    string userContent,
    List<ChatMessage> chatHistory)
{
    // Initialize OpenAI client
    OpenAIClient openAIClient = new OpenAIClient(apiKey);

    // Create chat client for the specified model
    ChatClient chatClient = openAIClient.GetChatClient(model);

    // Append system and user messages to chat history
    chatHistory.Add(new SystemChatMessage(systemPrompt));

    // Add user message to chat history
    chatHistory.Add(new UserChatMessage(userContent));

    // Get AI response
    ClientResult<ChatCompletion> chatResult = await chatClient.CompleteChatAsync(chatHistory);

    string response = chatResult.Value.Content[0].Text ?? string.Empty;

    // Add assistant response to chat history
    chatHistory.Add(new AssistantChatMessage(response));

    return response;
}

private static async Task ExecuteChatWithExcel(string openAIApiKey)
{
    Console.WriteLine("AI Powered Excel ChatBot");

    Console.WriteLine("Enter full Excel file path (e.g., C:\\Data\\report.xlsx):");

    // Read user input for Excel file path
    string? excelFilePath = Console.ReadLine()?.Trim().Trim('"');

    if (string.IsNullOrWhiteSpace(excelFilePath) || !File.Exists(excelFilePath))
    {
        Console.WriteLine("Invalid path. Exiting.");
        return;
    }

    if (string.IsNullOrWhiteSpace(openAIApiKey))
    {
        Console.WriteLine("OPENAI_API_KEY not set. Exiting.");
        return;
    }

    string csvContext;
    try
    {
        // Convert Excel to CSV-like context
        csvContext = BuildCsvContext(excelFilePath);
    }
    catch (Exception ex)
    {
        Console.WriteLine($"Failed to read Excel: {ex.Message}");
        return;
    }

    Console.WriteLine("\nOverview (AI):");

    List<ChatMessage> chatHistory = new List<ChatMessage>
    {
        new SystemChatMessage("You are an assistant that helps analyze Excel data.")
    };

    try
    {
        // Get an overview of the Excel data
        string overview = await AskOpenAIAsync(
            openAIApiKey,
            model: "gpt-5", // replace with a model available in your account
            systemPrompt: "You are a helpful assistant. Provide a concise 3-6 bullet summary of the workbook data.",
            userContent: csvContext,
            chatHistory: chatHistory);

        Console.WriteLine(overview);
    }
    catch (Exception ex)
    {
        Console.WriteLine($"OpenAI overview failed: {ex.Message}");
    }

    Console.WriteLine("\nDo you want any specific details? Type 'Stop' to exit.");

    while (true)
    {
        // Read user question
        Console.Write("\nYou: ");

        string? userQuestion = Console.ReadLine();

        if (userQuestion == null)
            continue;

        if (userQuestion.Trim().Equals("Stop", StringComparison.OrdinalIgnoreCase))
            break;

        // Define system prompt for the chatbot
        string systemPrompt = @"You are an intelligent Excel assistant built using Syncfusion XlsIO and OpenAI. Your job is to help users understand, analyze, and interact with Excel data through natural language. You are integrated into a C# application and can access structured Excel content such as tables, charts, formulas, and cell values.
            Your capabilities include:
            - Summarizing Excel sheets and ranges
            - Answering questions about data trends, totals, averages, and comparisons
            - Explaining formulas and calculations
            - Suggesting improvements or insights based on the data
            - Responding in a friendly, professional, and concise manner
            Constraints:
            You do not generate or modify Excel files directly. Instead, you interpret and explain the data provided to you by the application.
            Always assume the user is referring to the most recent Excel content unless stated otherwise. Maintain context across the conversation to provide coherent and helpful responses.
            If the user asks something outside the scope of Excel data, politely redirect them back to Excel-related tasks.";

        string userPrompt =
            "Question:\n" + userQuestion;

        try
        {
            // Get answer from OpenAI based on user question and chat history
            string answer = await AskOpenAIAsync(
                openAIApiKey,
                "gpt-5",
                systemPrompt,
                userPrompt,
                chatHistory);

            Console.WriteLine("ChatBot: " + answer);
        }
        catch (Exception ex)
        {
            Console.WriteLine($"OpenAI error: {ex.Message}");
        }
    }
}

AI-Powered Excel ChatBot

Enter full Excel file path (e.g., C:\Data\report.xlsx):

D:\VS Projects\ExcelChatBot\ExcelChatBot\Employees Performance Report.xlsx

Overview (AI):

- Scope: 100 employee review records covering all quarters of 2025 across six departments (Support, HR, Engineering, Sales, Marketing, Finance); Employee IDs are unique, with several employees appearing multiple times across periods/departments (likely transfers or multiple reviews).

- Performance scores range from 2.50 to 4.96 (top: Helen, Finance, Q1 2025, 4.96; Paula, Marketing, Q4 2025, 4.95; Quentin, Marketing, Q1 2025, 4.91; lowest: Laura, HR, Q4 2025, 2.50). Many entries are above 4.0, indicating generally strong performance.

- Goals Met ranges from 0-10, with weak alignment to scores, e.g., high scores with 0 goals: Mike, HR, Q1 2025, 4.82; Eva, Support, Q2 2025, 4.71; low/medium scores with 10 goals also occur), suggesting goals tracking may not fully reflect overall performance ratings.

- Top-scoring entries span multiple departments (notably Finance, Marketing, Support), indicating high performers are distributed rather than concentrated in a single department.

- Manager comments use recurring themes/templates: “Outstanding performance,” “Exceeded expectations,” “Consistently delivers high-quality work,” alongside improvement flags like time management, deadlines, and attention to detail; several note strong leadership and problem-solving.

- Data note: The sheet includes Syncfusion trial/license text at the top and bottom (non-data content) that should be ignored or removed for clean analysis.

Do you want any specific details? Type ‘Stop’ to exit.

You: Who are the top 5 performers this quarter?

ChatBot: Assuming “this quarter” = the most recent period in the sheet (Q4 2025), ranked by Performance Score:

- Paula (E0089) – Marketing – 4.95.
- Jane (E0034) – Finance – 4.82.
- Eva (E0023) – HR – 4.65.
- Tina (E0070) – Engineering – 4.53.
- Alice (E0033) – Marketing – 4.47.

Note: Eva appears twice in Q4; I used her higher score. If you meant a different quarter or want per-record ranking, let me know.

You: Which department has the highest average performance score?bnmkl;’

ChatBot: Finance has the highest average performance score: 4.15 (13 reviews).

- Department averages for context: Marketing 3.90; HR 3.84; Sales 3.64; Engineering 3.64; Support 3.47.

- If you meant a specific quarter (e.g., Q4 2025), let me know, and I’ll filter and recompute.

You: Summarize the manager’s feedback for the HR team.

ChatBot: Most common development theme: time management and deadlines (5 of 10 HR reviews mention “needs improvement in time management” or “struggled with meeting deadlines”).

- Strong praise also present: 2 “Outstanding performance,” 1 “Exceeded expectations,” and 1 “Consistently delivers high-quality work.”

- One review is mixed: “Met most goals with room for improvement.”

- Overall tone is balanced-about half recognition and half developmental feedback-spanning Q1-Q4 2025.

- Actionable takeaway: Prioritize time-management and deadline-support initiatives while continuing to recognize top performers.