How It Works

Upload Your File & Choose Analysis Type

Start by uploading your Excel file (.xlsx, .xls) or CSV file. Your file is stored locally on your machine - never uploaded to any cloud server.

What Happens Behind the Scenes:

• File Processing: Your file is read and parsed using Python's Pandas library
• Local Storage: The file is temporarily stored in a local folder on your machine
• Data Validation: Columns are detected and validated for the selected analysis type
• Sheet Selection: If your Excel file has multiple sheets, you choose which one to analyze

AI Processes Your Data

Once you start the analysis, Aidstack Insights sends your data to OpenAI's API for intelligent processing. Here's the exact flow:

The Processing Pipeline:

1. Data Chunking: Your data is intelligently split into manageable chunks (to stay within AI model limits)
2. Context Building: The system creates rich context including:
   • Sample data from your spreadsheet
   • Column names and data types
   • Statistical summaries (if applicable)
   • Your selected analysis type
3. AI Analysis: OpenAI's GPT-4o-mini model (128K context window) analyzes your data:
   • Detects patterns and trends
   • Identifies categories and themes
   • Extracts sentiment from text
   • Translates content (if requested)
4. Result Aggregation: Results from multiple chunks are combined into a coherent analysis
5. Local Storage: Analyzed results are saved to a new Excel file on your local machine

Review Your Results

After processing, your analyzed data is displayed in an interactive table. Each row shows the original data alongside AI-generated insights.

What You Get:

• Interactive Table: Browse and search through your analyzed data
• New Columns: AI-generated insights appear as new columns alongside your original data
• Downloadable Results: Export to Excel with all original and analyzed data
• Progress Tracking: Real-time progress bar shows analysis status

Reading the Confidence Columns (Jev Classification):

If you ticked Add confidence columns, each result is followed by a __confidence value between 0 and 1. It measures how clear-cut the answer was, not how likely it is to be correct. Jev scores every option you supplied: a number near 1 means the winning option scored far above the rest, while a number near 0 means several options scored alike and the pick was a close call.

Use it to decide what to check first — it sorts your rows from clear-cut to borderline. Read it as a ranking, not as a percentage: 0.80 does not mean “80% likely correct”.

These are starting points from the model's documentation, not guarantees. To set a threshold you can defend, hand-code a sample of rows, compare them to the results, and see where the disagreements actually fall on the scale.

• Yes/No questions report no confidence — that column stays empty. Use the __score column instead (the probability of “yes”), where distance from 0.5 carries the same meaning.
• A low score is information, not a failure. Feedback that genuinely spans two categories should score low — that is exactly the row a person most needs to look at.

Chat with Your Data

Ask questions about your data in natural language and get instant answers with streaming AI responses.

How the Chatbot Works:

1. Question Detection: The system analyzes your question to understand what you're asking:
   • List questions: "Show me all...", "What are the unique..."
   • Analytical questions: "Which has the most...", "How many..."
   • Statistical questions: "What's the average...", "Calculate..."
2. Direct Data Analysis: Before asking the AI, the system performs actual Pandas operations:
   • Extracts unique values (up to 100)
   • Calculates value counts and frequencies
   • Computes statistics (mean, median, etc.)
   • Finds correlations between columns
3. Context Building: The system provides the AI with:
   • Sample of your data (up to 8,000 characters)
   • Pre-computed analysis results
   • Column names and data types
4. AI Response Streaming: OpenAI GPT-4o-mini generates the response:
   • Streams word-by-word using Server-Sent Events (SSE)
   • Renders markdown formatting in real-time
   • Provides natural language answers with complete data

Quick Question Templates:

Click on pre-made questions to get instant insights:

Privacy & Security

Where Is Your Data Stored?

• Uploaded Files: Stored locally in a folder on your machine (/uploads directory)
• Analyzed Results: Saved locally in a folder on your machine (/analyzed directory)
• API Key: Optionally saved in your browser; sent to the server for processing
• No Cloud Storage: We never store your files or data on any remote server

What About OpenAI?

When you use OpenAI's API:

• ✅ Your data is NOT used for training OpenAI's models
• ⏰ Data is retained for 30 days only for abuse monitoring
• 🗑️ After 30 days, your data is automatically deleted
• 📄 Full details: OpenAI API Data Usage Policies

Data Flow Summary: