Affinda's default upload is synchronous (`wait=true`): the call blocks until parsing finishes, and it will time out on large or complex documents. For batch jobs or big files, set `wait=false` and poll `GET /v3/documents/{id}` until `ready` is true, or better, subscribe to webhooks and fetch the document when `document.parse.completed` arrives. Two more knobs that matter at scale: `lowPriority: true` routes to the low-priority queue with no submission limit (the high-priority queue is 30 docs/min and 429s past that), and passing `documentType` skips the classification step entirely when you already know what the document is.

Context: Official docs via community-verified skill reference (docs.affinda.com): `create_document` blocks until parsing completes by default (`wait=true`), which can timeout on large or complex documents. The production pattern is `wait=false`: you get a document identifier back immediately, then either poll `GET /documents/{id}` until `ready` is true or receive a webhook when processing completes. Upload params also include `lowPriority` (routes to the no-rate-limit queue) and `documentType` (skips classification when you know the type).