Does iDBQuery only work with clean, structured data?
No — iDBQuery works with messy, inconsistent and semi-structured data, not just tidy tables. It reads databases, spreadsheets, PDFs and folders of files, uses OCR for scanned documents, and its semantic layer maps cryptic fields, so imperfect real-world data still returns useful cited answers.
Not all data lives in a clean relational table, and iDBQuery does not assume it does.
- Structured sources. Databases, warehouses, spreadsheets and CSVs are queried directly with generated SQL.
- Documents. PDFs and scanned files are ingested and made queryable, with OCR turning images of text into data you can ask about.
- Folders of mixed files. Point iDBQuery at a whole folder and it processes the contents so you can ask questions across them.
- Cryptic schemas. The semantic layer interprets abbreviations, codes and inconsistent naming, so you are not blocked by a poorly documented database.
For example, a contracts folder full of scanned PDFs can become answerable — 'which contracts renew in the next 90 days?' — even though none of it started as a neat table. And a legacy database with columns like `f_amt` and `dt_1` can still answer 'total amount by month' because iDBQuery maps those fields to meaning. Every answer is still cited back to the underlying source, whether that source is a row in a table or a page in a document. So structured, semi-structured and unstructured data can all sit in one live model, and you query them the same way: by asking.
Updated 2026-06-22