Pre-process

Bulk jobs at half price

Async delayed queue for backlogs and nightly runs up to 5GB. Same output, same evidence — 50% fewer credits.

Batch, without the babysitting

Drag a job through the queue — webhooks tell your system when each lands.

With a card focused, Alt and an arrow key moves it.

Open1
  • 1043 · Lending archive · 1,200pp — Queued · −50%

    −50%
In review1
  • 1044 · Nightly AP backlog · 480pp — Queued · −50%

    −50%
Resolved1
  • 1042 · Freight week · 200pp — Done · Done

    Done

5GB pipelines

Thousands of pages per job; split and classify run inside automatically.

Webhooks on complete

Signed callbacks per job and per document, with retries.

Half-price credits

Delayed scheduling trades immediacy for 50% fewer credits.

Same evidence

Identical output format — sync, async and batch are interchangeable.

From backlog to posted data

  1. 01

    Submit the pile

    One call, thousands of pages, idempotency key included.

  2. 02

    Queue drains

    Jobs process when capacity frees — status live, webhook at the end.

  3. 03

    Data posts

    Results identical to sync; credits show the 50% line item.

Backlogs love batch.

Nightly runs at half price, evidence intact.

See the API

Batch questions, answered

The questions teams ask most about batch.

Talk to us

50% fewer credits versus sync/async on the same pages. Tables +2 still applies where tables are used.

Job status endpoint plus signed webhooks per job and per document, with 5 retries and replay.

5GB per batch job. Split (200pp/100MB per file) and classify run inside automatically.

When you need answers now — use sync ?wait (≤25s) or async 202 for interactive flows.

Run bulk at half price

Async queue with webhooks.

Friday's 2,000-page backlog drains overnight. Monday starts with posted data at half credits.