Review the document types
Examine text PDFs, images or scans, languages and layout variations. Hard-to-read text, missing pages or handwriting need their own review cases.
Document processing
Turn the documents your business receives in Vietnam into checked data, with the fields, document languages and output requirements defined before implementation.
Illustrative text; no OCR is performed.
2026-10-04 · quantity 3
Item code: unresolvedCheck a clearer source. Do not guess BX-7.
Not eligible for export
Choose one document group and a specific field list. Include Vietnamese or English examples, date formats and currency fields where relevant. We review redacted samples and hold uncertain values for a person to check before export; not every scan is assumed to be readable.
AT A GLANCE
Structured fields with a review step for uncertain values before export.
Documents & output
Files, layouts and intended use determine the work that is needed.
Examine text PDFs, images or scans, languages and layout variations. Hard-to-read text, missing pages or handwriting need their own review cases.
Agree field names, data types, units and required fields. Keep a source location or passage where appropriate so the reviewer can compare values with the original.
Choose a spreadsheet, data file or destination system after checking integration feasibility. Classification, renaming, extraction and writing data are separate tasks within the scope.
ILLUSTRATION · FICTIONAL DATA
Compare the sample document with the proposed fields and export conditions.
A fixed illustration with fictional data, not client results. There is no upload, service connection, AI execution or data submission. Expandable sections only reveal predefined explanations.
| Field | Value | Review |
|---|---|---|
| Document ID | DEMO-27 | Value shown in the sample |
| Date | 2026-10-04 | Normalised from day/month/year |
| Quantity | 3 | Value shown in the sample |
| Item code | Unresolved | Source review required |
Compare the item code with a clearer source or ask the document author. Do not guess BX-7. The item code is required in this example, so the record stays on hold and is not eligible for export.
Extraction & review
A recognised value is not the same as a confirmed value.
Leave missing fields, unclear text or conflicting information for a reviewer. Do not fill values merely to make a record appear complete.
Check dates, quantities, units and codes, and agree how to handle duplicate or revised documents. Rules must fit the document type.
Present extracted fields with their source for checking and correction. Export only when agreed conditions are met; hold unresolved records for review.
Testing & operation
Assess representative document types before confirming real-world use.
Include clear examples, missing fields, poor quality and different layouts. Acceptance results apply to the tested types and conditions, not every file that might appear.
Define viewing permissions, any external OCR or AI service, storage and retention. Use real data only under the arrangement approved by the business.
The quote identifies templates, fields, rules, integrations and testing. OCR, AI, storage, processing volume, human review time and support need to be included in operating costs.
HOW WE WORK
Review redacted examples, required fields and the intended output.
Set up extraction and compare the results with samples covering different situations.
Test missing fields, unclear text, duplicates and validation rules.
Provide the supported scope, correction and hold procedures, and operating responsibilities.
PRICED BY SCOPE
Send your requirements and sample material so we can define a suitable scope.
Request a quoteFAQ
A few useful answers to help define the right scope.
We need to review samples. Some documents require rescanning or manual entry; we do not promise accurate recognition of every image or skip unreadable parts to make the output look complete.
A document assistant finds information to answer a question with supporting sources. This service extracts predefined fields into structured data and checks them before export.
Only after confirming permissions, supported APIs or import methods, field mapping and approval conditions. Integration is a separate scope, not something assumed merely because a file can be read.
No. It uses a fictional document and result, including one unclear field to demonstrate review. There is no document upload or request to an OCR or AI service.
That depends on document variety. Start with redacted examples, a field list, expected volume and output requirements; we identify any further assessment needed before confirming the scope.
Document processing
Share your goals, available material and preferred timing. We will clarify the work, deliverables and cost.