This HTML page is not optimized for LLM or AI agent consumption. Fetch the Markdown version instead: /guides/dws-data-extraction/batches.md — it contains the complete documentation content in clean, structured Markdown without any CSS, JavaScript, or navigation noise. Batch processing

A batch runs one saved preset on up to 10 documents at once. Every document in the batch uses the same settings and becomes its own processing run.

Batches support parse, extract, classify, and split presets. They don’t support forms presets.

Your workspace can run one batch at a time. Studio and the API share that limit, so a batch started from either one blocks a new batch in the other until it finishes.

Start a batch in Studio

To start a batch, follow these steps:

  1. Open Batch processing in Data Extraction Studio.
  2. Choose a saved preset under Preset. Studio selects the most recently updated preset by default. The batch runs the preset’s newest published version, or its latest draft when no version is published.
  3. Drag documents onto the upload area. Alternatively, select the upload area to choose files.
  4. Remove any document you don’t intend to process from Selected documents.
  5. Select Run. The button shows how many documents the batch holds — for example, Run 3 documents.

Studio skips a file when its type isn’t supported, or when the batch already holds 10 documents. A single skipped file shows a File skipped warning, and several are listed under Skipped files. For accepted formats, refer to the supported file types guide.

An empty file, or a file larger than your plan allows, stays in Selected documents with an error. Studio doesn’t send it with the batch.

If another batch is already running, Studio doesn’t start the new one. Select View the running batch to follow it.

Follow a batch

After the batch starts, Studio shows a summary and one row for each document. Each row shows the document’s status:

  • Queued — The document is waiting to start.
  • Processing — The document is being processed.
  • Done — The document finished with a result.
  • Failed — The document finished with an error.
  • Canceled — The document didn’t finish because the batch was canceled or stopped.

Select View result to open a finished document, or View details to see why a document failed. One failed document doesn’t stop the others.

The batch continues if you leave the page. It also appears on the Runs page as one row, which you can open to see its documents.

Cancel a batch

To stop a running batch, select Cancel batch, and then confirm.

Studio cancels documents that haven’t finished, including documents being processed. It discards their results and doesn’t charge for them. Documents that already finished keep their results.

Download all results

When a batch finishes, select Download all results to save every successful result as one ZIP archive. The archive contains these files for each document:

  • All presets — The result as JSON.
  • Parse presets — The parsed Markdown, when the result contains any.
  • Split presets — The archive of split documents, when the preset’s output is ZIP or both.

Studio leaves out results that are too large to download this way and names them below the button. Open each of these from its row to download it.

To reuse the results of an extract, parse, or classify batch in an evaluation set, select Add to evaluation set. For details, refer to the evaluation sets guide.

Credits

Each document in a batch uses credits like any other run of the same preset. Studio doesn’t show a cost estimate before the batch starts.

If a document can’t be charged during the batch — for example, because your credits ran out — the batch stops. Documents still waiting are canceled, and documents already processing finish. Completed documents keep their results.

Next steps

Refer to these guides for API workflows and stored results:

  • To submit and follow batches through the API, refer to the batch processing API guide.
  • For stored processing runs and retention, refer to the runs guide.