Skip to content

The platform

Upload

Send documents to ByteIt as a file upload or straight from your own S3 bucket.

Every job starts from a document. You upload it in the same request that creates the job, so there is no separate upload step: a parse, classification or custom job takes the file directly.

Send one document as file, or up to 25 in one request as files. The two are mutually exclusive.

curl -X POST https://byteit.ai/v1/jobs/parse-jobs/ \
-H "X-API-Key: $BYTEIT_API_KEY" \
-F "file=@invoice.pdf"

One file returns the job as parse_job. Several files return one job per document in parse_jobs, with their count, and list any document that was rejected in failed_files with its file_name and error. The others are processed normally.

Limit Value
One request, all its files together 50 MB
Files per request 25
Pages per document 250

A single document can use the whole request limit. A request over it returns 413. A document with more pages fails its job, and a file type ByteIt cannot read fails with UNSUPPORTED_FILE_TYPE.

Set input_connector to s3 and say where the documents are in input_connection_data.

Field Meaning
source_bucket The S3 bucket that holds the documents.
source_path_inside_bucket The key of one document.
source_paths The keys of several documents. Use this or source_path_inside_bucket, not both.

To have the result written to your bucket instead of downloading it, set output_connector to s3 and output_connection_data to the destination bucket and path.

  • nickname gives the job a name you recognise in the job list and in Studio.
  • queue_for_batch: true accepts the job and holds it in queued instead of starting it right away.