One live model from everything you have.
The technical core, in plain language: how iDBQuery connects your databases, spreadsheets, PDFs and APIs into a single queryable model you can ask in plain language — no warehouse project, no pipelines.
The problem.
Your answers live in many systems at once. A figure is a row in Postgres, a column in a spreadsheet, a line in a PDF, and a field returned by an API. Without a single model on top, those views never reconnect, and every question turns into an export-and-VLOOKUP afternoon.
Connect anything.
MySQL, Postgres, MongoDB, Excel, CSV, PDFs and REST APIs all connect directly. iDBQuery reads each source as it is, introspects its structure, and builds one queryable model — without copying your data into yet another warehouse.
The semantic layer.
Real databases use cryptic table and column names. The semantic layer maps them to plain business meaning, so iDBQuery understands an obscure schema and answers correctly. Corrections are remembered, and the model gets sharper over time.
Cited answers.
Ask a question in plain language and get an answer where every number traces back to its source row. Click any figure to see exactly where it came from and the query behind it. No black box — the working is always one click away.
The engine.
Queries run on a fast in-process engine. Each conversation keeps a warm session, so follow-up questions stay quick. Sources that cannot be queried in place are pulled in on demand, with safeguards so a large source cannot overwhelm memory.
Surfaces, driven by chat.
One answer becomes a chart, a table, a live dashboard, or a shareable report. Asking a question in chat and building a surface from it are the same action — describe what you need and iDBQuery assembles it.
What iDBQuery connects
| Source type | How it connects |
|---|---|
| Databases | MySQL, Postgres and MongoDB, introspected live |
| Spreadsheets | Excel and CSV, read as queryable tables |
| Documents | PDF and Word, indexed page by page with citations |
| APIs | REST endpoints, mapped into the model on a schedule |