Skip to main content
Every FlexOrch pipeline starts with a document upload. The Python SDK exposes three upload methods — single file, batch, and connector-based — so you can integrate with your existing storage and workflow without restructuring your code.

​
Single file upload

Pass a local file path to client.documents.upload(). The call is synchronous and returns a Job object as soon as the document is queued. 
import os
from flexorch import FlexOrch

client = FlexOrch(api_key=os.environ["FLEXORCH_API_KEY"])

job = client.documents.upload("invoice.pdf")
print(f"Job ID: {job.id} | Status: {job.status}")
Supported file types include PDF, DOCX, PPTX, XLSX, HTML, Markdown, and plain text. FlexOrch auto-detects the MIME type from the file contents.

​
Multi-file upload

Use client.documents.upload_many() to upload a list of files in a single call. Each file gets its own job; the method returns a list of Job objects in the same order as the input paths. 
paths = ["contract.pdf", "memo.docx", "slides.pptx"]

jobs = client.documents.upload_many(paths)

for job in jobs:
    print(f"{job.id}  —  {job.status}")
upload_many() issues uploads concurrently under the hood, so submitting 20 files takes roughly the same wall-clock time as submitting one.

​
Upload from a connector

If your documents live in a connected storage source (S3, Google Drive, SharePoint, and others), you can pull them directly using a connector ID and a list of object keys. No local download required. 
jobs = client.documents.upload_from_connector(
    connector_id=1,
    keys=["reports/q1.pdf", "reports/q2.pdf"],
)

for job in jobs:
    print(f"{job.id}  —  {job.status}")
You can find connector IDs in the FlexOrch dashboard under Settings → Connectors.

​
Duplicate detection

Every upload is fingerprinted against previously processed documents in your workspace. When a duplicate is detected, the returned Job object contains is_duplicate = True and points to the original job via original_job_id. 
job = client.documents.upload("invoice.pdf")

if job.is_duplicate:
    print(f"Duplicate detected. Original job: {job.original_job_id}")
else:
    print("New document — processing started.")
Duplicate jobs are still billed at the standard rate unless you have duplicate suppression enabled on your plan. Check your plan settings before uploading large batches.

​
File size limits

PlanMax file sizeMax files per batch
Starter25 MB10
Growth100 MB50
Enterprise500 MBUnlimited
Uploading a file that exceeds your plan’s size limit raises a FileTooLargeError. Catch it and inform users before they attempt the upload.
from flexorch.exceptions import FileTooLargeError

try:
    job = client.documents.upload("large_scan.pdf")
except FileTooLargeError as exc:
    print(f"File rejected: {exc}")

​
Next steps

Manage Jobs

Poll status and access extraction results for uploaded documents.

Build Datasets

Combine completed jobs into exportable, LLM-ready datasets.