> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.umnai.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.umnai.com/_mcp/server.

# Ingestion

Ingestion is the stage that reads source files, applies a schema, and creates a datasource that can be used for onboarding.

It connects the data you prepared in storage with the schema that describes it. The result is a datasource: a validated, platform-managed representation of the source data.

## How ingestion works

Source files are read from a connected storage location.

The schema defines how each column should be interpreted during ingestion.

The ingestion configuration combines the schema with ingestion-specific settings.

The ingestion job runs against a storage path using a processing compute configuration.

The completed job creates or updates the datasource used by onboarding.

![](https://files.buildwithfern.com/umnai.docs.buildwithfern.com/927a0cc8012b4e59aa22eb50d50c554f9ca919ed8fe98e1fc5790e62bdfa4b87/assets/images/ingestion_entities.png)

## Initial and incremental ingestion

The first ingestion job for a source creates a datasource.

Later ingestion jobs can update that datasource with additional data. Updates point to the existing datasource `id` and can include an `increment_label` to identify the new increment.