How IMD works

A project is stored in the folder you select or open. That folder is the center of the workspace and should contain the data and outputs for one project.

Choose a folder location you can find later. Avoid temporary download folders for serious work because outputs and reports will be easier to lose.

Use column headers and field controls to see how IMD interpreted each field: text, number, date, boolean, enum, or other supported types.

Correct type interpretation matters because summaries, charts, cleaning suggestions, and AI context all depend on fields being understood correctly.

Start by opening the table and reviewing Diagnostics, Summary, and field-level signals. Use the Cleaning tab to prepare an editable cleaning plan.

Run the plan only after reviewing the steps. IMD updates the active table source file in place and keeps table versions in History so you can inspect or restore earlier versions.

Attach or rely on the current table context where the app provides it. Table names, field names, types, and sample values help the model avoid generic answers.

If the answer seems unrelated, restate the table and field names explicitly. Context can change when you switch tables or conversations.

After a run, compare the cleaned result with the original table. Check row count, field names, important categories, missing values, and any field used in later analysis.

Cleaning results appear in the active table. Use the History tab to review previous table versions or restore one when needed.