Import settings

Stockly → Settings → Import.

Reading documents

• Default analysis engine — which route a document takes by default: the classic reader or a language model. → AI import

The card Native AI Import (OpenAI) below it belongs to the AI extension and appears only when that is installed. → How a document is chunked

Variants

Variant stock update mode decides what to do when an invoice names a product that has variants.

PDF extraction

A separate section with the extraction service’s connection data: Client ID, Client Secret, Split mode and Enable Adobe API logs.

The split mode determines who divides the file into individual pages before the analysis begins: Internal by FPDI does it inside the shop and spares the credits of the account, External by Adobe leaves it to the service. A failed internal split is forwarded to the service so that the analysis continues.

The same section carries an API Usage Report: how much of the extraction service has been used. On a plan billed per page that is the only place the remaining allowance is visible.

Note

Without these the classic reader has no means of reading: the engine is listed as unconfigured, and language-model analysis is what remains. The module has no local PDF reader, so this holds for every document, not only for scans with no text layer.

Logs

Enabled logs record the requests and responses of external services, which is used when diagnosing a reading problem. Retention is set by Retention (days). → API logs