Studio Reference

Everything you need to run and use Datagrunt Studio — a web-based workbench for exploring, cleaning, and joining CSV and Excel files, powered by the Datagrunt library and an embedded DuckDB engine.

Studio runs entirely on your machine: a Next.js frontend talks to a FastAPI sidecar that wraps Datagrunt and DuckDB. There are no accounts and no cloud backend — your session’s tables live in a local DuckDB file, and nothing leaves your machine unless you explicitly export it.

Studio follows the same design principle as the library: load losslessly, transform explicitly. Datagrunt parses your files without silently coercing or dropping anything; all type conversion and cleanup happens in Studio’s DuckDB layer, in the open, when you decide it should.

• Setup & Running — Prerequisites, running locally, running with containers, and configuration (.env, Google Cloud credentials, local Ollama).
• Importing Data — Uploading files, the staged import preview, per-file parse options, Google Cloud Storage import, and session management.
• Querying & Cleansing — The Query Editor (DuckDB SQL console), the results grid, joins, and the visual Cleanse pipeline.
• AI PDF Extraction — The two-step Extract → Rationalize workflow, and choosing between Gemini, Vertex AI, and local Ollama models.
• Exporting Data — Downloading results locally and exporting to Google Cloud Storage.

For the Datagrunt library itself — the parsing engine behind Studio’s imports — see the API Reference and Guides.