befree-itr-automation
Janvi shahthis is scope of work I need proper designs
Comments (0)
Sign in to leave a comment
Internal Reference
GUIDANCE.
Handover Documentation
This page is the handover reference for Befree internal users working with the ITR Document Automation solution. It documents how document sets move from submission through OCR and classification, into financial data extraction and categorisation, and on to reviewed, structured output.
Use the sections below to understand submission requirements, how extracted data is mapped to Befree's output template, how to review and confirm results before finalisation, and how to identify and report extraction or mapping issues.
Workflow Guidance for Befree Internal Users
This reference covers the complete document automation workflow, from initial submission through to structured output — what the system does at each stage, what requires your review, and how to identify an issue if one appears.
- Supported formats
- Digital PDFs — text-based PDF documents exported directly from a source system.
- Scanned PDFs — PDF files created from a physical scan of a paper document.
- Common scanned-image formats — standard image files captured from a scanner or device camera.
Single and batch submission- A single document can be submitted on its own for processing.
- Multiple documents belonging to the same client can be submitted together as a batch in one submission action.
Document set composition- Group all supporting documents for one ITR under a single document set before submitting.
- A document set represents the complete collection of source documents supplied for one client’s return.
- Keep each document set scoped to one client to preserve accurate traceability at review time.
- Reading digital text
- If a document contains embedded digital text (for example, a native PDF export), the system reads that text directly.
OCR for scanned documents- If a document is a scanned PDF or an image with no embedded text layer, the system applies Optical Character Recognition (OCR) to derive readable text from the page image.
Document classification- Every document is automatically classified against Befree’s agreed document types.
- Classification determines which extraction fields and rules apply to that document going forward.
Monitoring progress- Processing status for each document set can be monitored from the Processing area until reading, OCR, and classification are complete.
- Approved extraction fieldsExtraction methodology
- Extraction is AI/ML-assisted and applies only to the fields approved for each classified document type.
- Accuracy depends on document legibility — clean digital text and clear scans yield the most reliable extraction.
Categorization and mapping- Each extracted field is mapped to a relevant income or deduction category using approved categorisation and mapping rules.
- Mapping output aligns to the structure of the Befree-defined output template so no manual re-entry is required at export.
Source reference- Every extracted field retains a reference back to the source document it came from, carried forward into review.
- Review interface
- Open the Review area for the relevant document set to see extracted values alongside the source document they were taken from.
Verification checklist- Confirm each extracted value matches the source document.
- Confirm the assigned income or deduction category is correct for that field.
- Confirm no required field has been left unextracted for a submitted document.
Confirmation and data locking- Once a user is satisfied the data is correct, the information is confirmed in the review step.
- Confirmed data proceeds to structured output generation; unconfirmed data is not used to generate output.
- Output template structure
- Output follows the Befree-provided template format, with data organised by source document type, extracted field, and assigned category.
Export format- Structured output is generated in the approved exportable format (Excel or equivalent structured template) agreed for the initial scope.
Integration with tax-preparation workflow- The generated output is designed to be used directly within Befree’s existing tax-preparation workflow — it assists preparation and does not independently prepare, approve, or file a return.
Storage and retrieval- Generated output can be located and retrieved from the Outputs area against its originating document set.
- Initial scope referenceDocument typeExtraction fieldsPAYG Payment Summary / Income StatementSalary and wages, tax withheldBank Interest StatementInterest incomeDividend StatementDividend income, franking creditsRental Property Statement / Agent SummaryRental income, related deductionsWork-Related Expense ReceiptsWork-related deductionsPrivate Health Insurance StatementTax offsets
Documents outside this approved list fall outside the current implementation scope and should be flagged rather than submitted for automated extraction. Future expansion of the approved list is managed through Befree’s review and sign-off process.
- Common extraction errors
- A field is missing where the source document clearly contains a value.
- An extracted value does not match the corresponding figure on the source document.
- OCR misreads characters on a low-quality scan, producing an incorrect value.
Misclassification indicators- A document has been classified under the wrong document type, which changes which fields are extracted.
- A document set contains an unrecognised document type outside the approved list.
Mapping inconsistencies- An extracted field has been assigned to the wrong income or deduction category in the output template.
Reporting and escalation- Flag the affected field or document during the review step rather than confirming it as correct.
- Note the document name, the field in question, and the expected value when reporting an issue.
- Escalate recurring or systemic issues so mapping or classification rules can be reviewed.
- Source document tracking
- Each extracted field records the originating document name within its document set.
Field-to-source mapping- The document name, extracted field, and extracted value are held together, so a reviewer can trace any output value back to where it came from.
Verification and cross-reference- Use this traceability during review to cross-check an extracted value directly against the relevant part of the source document.
Audit trail- Traceability is retained end-to-end, from initial extraction through review confirmation to the final structured output.
AI-Driven Tax Workpaper Automation
An internal AI-assisted platform that converts client source documents into verified, structured ITR workpapers.
How It Works
Ingest
Client source documents are uploaded individually or as combined PDF decks, then automatically detected and split into logical document units.
Classify & Extract
AI-assisted classification and data extraction from source documents into verified, structured ITR workpapers, with the appropriate schema applied per document type.
Review & Verify
Extracted data is viewed side-by-side with source documents for verification, taxonomy categorization, and full traceability before sign-off.
Export
Approved cases generate structured workpapers, exportable as Excel, JSON, and CSV with source traceability preserved for every value.
Ready to Process Your Next Case?
This internal staff portal requires an authorized Befree account.
Sign In
No comments yet. Be the first!