What Is Document Processing Automation?
Document processing automation uses software to capture documents, extract their text and data, and route the results to the right system automatically. With Filestack, a document is processed the moment it is uploaded: Workflows run virus scanning, document detection, OCR and conversion, branch on the results, and send a signed webhook to your application.
The Five Stages of Document Process Automation
-
Capture the document from an upload, a scan or a phone photo.
-
Classify it by type, such as invoice, ID or letter.
-
Extract the text and the values the process needs.
-
Validate the results against business rules.
-
Integrate them into the system of record.
OCR vs Intelligent Document Processing
OCR turns an image of text into machine-readable text. Intelligent document processing is automated document processing built around OCR: capture, detection, extraction, routing and delivery. See the intelligent document processing pipeline.
- Pipeline
- Capture, scan, detect, OCR, convert, route, notify, store
- Build
- In the visual Workflow builder, or as JSON
- Storage
- Your own S3, Google Cloud Storage or Azure bucket
- Plans
- OCR, document detection, virus detection and webhooks on paid plans. Pricing
How Filestack Automates Document Processing
For OCR automation on uploaded documents, attach a Workflow to the upload. Filestack scans the file, straightens phone photos, reads the text, converts the result and routes it.
-
01Capture File Picker, camera, URL and cloud sources
-
02Scan
virus_detectionon any file type -
03Detect
doc_detectioncrops and deskews phone photos -
04OCR
ocrreturns text and a bounding box per word -
05Convert Output to PDF and other formats
-
06Route Conditions branch on each result
-
07Notify A signed webhook to your endpoint
-
08Store Filestack storage or your own bucket
OCR and document detection run on paid plans, each with a free trial.
Build a Document Workflow in JSON
Filestack Workflows are a document automation API. Define the tasks and conditions once, in the visual builder or as JSON, then attach the Workflow ID to every upload.
A Workflow is a tree. Each task runs after its parent, and a
requirementsblock decides whether it runs at all. The builder restates the same tree in plain English.Pass the Workflow ID in
storeTo. Every file the picker uploads runs through it, and this document workflow automation needs no other trigger.
"name": "document-intake",
"next": [{
"name": "virus_detection",
"next": [{
"name": "doc_detection",
"requirements": [{ "type": "and", "conditions": [{
"task": "virus_detection", "path": ["infected"],
"condition": "eq", "value": false }] }],
"next": [{
"name": "ocr",
"next": [{ "name": "store" }]
}]
}]
}]
}
client.picker({
accept: ['image/*', '.pdf'],
storeTo: { workflows: ['YOUR_WORKFLOW_ID'] },
}).open();
Route Documents Automatically Based on Their Content
To route documents by what they contain, add conditions to the Workflow. Each condition reads a task’s result by path, compares it with an operator, and decides which task runs next.
"type": "and",
"conditions": [{
"task": "ocr",
"path": ["text"],
"condition": "incl",
"value": "INVOICE"
}]
}
Invoice text
OCR text that includes "INVOICE" goes to the accounts payable queue through an
inclcondition on theocrresult.Infected file
A file that
virus_detectionflags is stored in a quarantine path.Unclear photo
An unclear photo routes to a retake prompt in your app, so the user can shoot it again.
Ten operators compare results, combined with and or or.
ltltegtgteeqneqinclninclkexknex
Get Results Into Your Systems With Signed Webhooks
To add virus scanning to uploads, put virus_detection first in the Workflow. The fs.workflow webhook then delivers every task’s result to your endpoint, signed with FS-Signature.
A sample
fs.workflowevent for a clean file. Write the results to your database, queue or CRM.The signature is HMAC-SHA256 of the timestamp and the raw body, with your endpoint's secret. Webhooks
"action": "fs.workflow",
"text": {
"workflow": "687r07d2-5f84-44a0-b20b-1c29a1deb2ab",
"sources": ["VJQTvCboRVivWJ2BpKgd"],
"results": {
"virus_detection_1548213592150": {
"data": { "infected": false, "infections_list": [] }
}
},
"status": "Finished"
}
}
from filestack.helpers import verify_webhook_signature
app = Flask(__name__)
@app.post('/filestack')
def filestack_webhook():
valid, details = verify_webhook_signature(
'YOUR_WEBHOOK_SECRET', request.get_data(), {
'FS-Signature': request.headers['FS-Signature'],
'FS-Timestamp': request.headers['FS-Timestamp']})
if not valid:
abort(401)
return '', 200
Document Processing Automation Use Cases
Examples of document processing automation include invoice and receipt intake, KYC onboarding, mailroom digitization and resume intake. More use cases
Invoice and receipt intake
- Documents
- PDF invoices and receipt photos
- Workflow
virus_detection,doc_detection,ocr, route on text- Result
- Accounts payable receives the text and a clean PDF
KYC and onboarding documents
- Documents
- ID photos and proof of address
- Workflow
doc_detection,ocr,storewith private access- Result
- Straightened scans and text, ready for review
Mailroom digitization
- Documents
- Scanned envelopes and letters
- Workflow
ocr, route on sender and type- Result
- Searchable mail, as at Earth Class Mail
HR and resume intake
- Documents
- Resumes as PDF, Word files or photos
- Workflow
virus_detection, convert to PDF,ocron scans- Result
- One searchable PDF per candidate
Earth Class Mail Case Study
Earth Class Mail turns paper mail into searchable digital documents. Filestack OCR, then called Filestack Capture, read and sorted the mail automatically.
"A simple integration with Filestack's content intelligence API allowed us to offload the analysis of a large volume of mail quickly and reliably. This automation has proven critically important as our team has more time to focus on enhancing our core product."
- 70%of Earth Class Mail's mail processing completed through the Filestack API
- 50%+increase in the overall auto-delivery rate
- 6,000+customer inboxes served
- 10,000images processed in a day, at peak
Filestack vs an IDP Suite vs Building It Yourself
Building document automation in-house means running OCR, scanning, storage and routing yourself. An IDP suite sells trained extraction and review screens. Filestack gives developers the pipeline as an API.
| Build it yourself | IDP suite | Filestack | |
|---|---|---|---|
| Who it is for | Teams with ML and infrastructure capacity | Operations teams processing set document types | Developers adding document steps to their own app |
| Setup | OCR engine, scanner, storage and queues | Vendor onboarding and model training | A Workflow and an API key |
| OCR and extraction | Your own engine and field logic | Trained field extraction per document type | Full text and word coordinates; your app maps the fields |
| Routing | Your own rules engine | Built-in rules | Conditions on any task result |
| Review UI | One you build | Built-in review screen | Routes exceptions to your own review queue |
| Developer control | Complete | Within the suite's options | JSON Workflows, webhooks and your own storage |
| Pricing | Infrastructure plus engineering time | Per page or per document, usually by contract | Plans by usage. Pricing |
An IDP suite fits teams that need trained field extraction and a built-in review screen.
Security and Compliance
Filestack is SOC 2 compliant. Documents travel over TLS, requests carry signed policies, virus detection can scan every file before it reaches your systems, and storage can be your own bucket. Security at Filestack
Authorize
Signed policies set who can upload and which calls run.
Encrypt in transit
Documents travel to Filestack over TLS.
Inspect
virus_detection scans each file before your systems read it.
Store
Results land in your own S3, Google Cloud Storage or Azure bucket.
Automate document processing on every upload.
- SOC 2 compliant
- 70% of Earth Class Mail's mail processing completed through the Filestack API
- Most teams ship their first upload in under a day
Document Processing Automation FAQs
How do you automate document processing?
Capture documents at upload, then run them through a pipeline that scans, detects, reads, converts and routes them. With Filestack, one Workflow attached to the upload does all of it and sends the result to your system by signed webhook.
Which documents can be automated?
Any document your users upload: PDFs, office files, scans and phone photos. Common examples are invoices, receipts, IDs, onboarding forms, mail and resumes.
What is the difference between OCR and IDP?
OCR converts an image of text into machine-readable text. Intelligent document processing (IDP) is the whole flow around it: capturing, detecting, extracting, validating and routing the document.
Can Filestack extract specific fields such as invoice totals?
The ocr task returns the full text and a bounding box for every word. Your application maps those results to the fields it needs, such as an invoice total, and classifies the document.
How are exceptions reviewed?
Workflows route exceptions, such as an unclear photo or a document that needs a second look, to your own review queue or app, where your team handles them.
How long does it take to set up?
Most teams ship their first upload in under a day. A document Workflow is configured in the visual builder or as JSON, then attached to uploads with its ID.