Data ingestion¶
Data ingestion is the entry point from your business environment into MOI. After ingestion, data is registered in the Catalog; later processing and use start from that registration.
Two ingestion methods¶
Method |
Suitable data |
After it enters the platform |
|---|---|---|
Upload files |
Unstructured files such as documents, spreadsheets, presentations, images, audio, and video. |
Stored in a Catalog volume; file content is stored in object storage. |
Connect a data source |
Structured data in a business database. |
Enters the Catalog as tables that can be queried and processed. |
Whichever method you use, incoming data has a clear location: workspace → Catalog → database → table / volume.
Two decisions when ingesting¶
One-time or continuous. Upload a single batch directly. For continually generated files, bind a workflow to the volume so each later upload automatically triggers processing. See Data processing.
No preprocessing is needed. Upload PDF, audio/video, and similar formats directly. A workflow handles parsing and structuring during processing. Ingestion does one thing: bring data in and register it.
Next steps¶
What you want to do |
Go to |
|---|---|
Understand how incoming data is registered and traced |
|
Process ingested data |
|
Walk through the complete path from upload to Q&A |
For AI agents: ingestion produces either files in a volume or tables in the Catalog. File content is stored in object storage, not in the database, and is read by reference during processing.