What file formats does Document AI support?

PDF, Word, Excel, PowerPoint, images (JPG, PNG, TIFF, BMP), and many other formats. Supports both scanned and digital files.

How accurate is OCR for Vietnamese?

99.5% for printed text, 95%+ for handwriting. The model is specially trained for Vietnamese with diverse regional data.

How many pages can be processed at once?

Supports batch processing up to 1000 pages per batch. Average processing time is <2s/page on GPU.

Is document data secure?

End-to-end encryption, data deleted after processing (configurable), GDPR and ISO 27001 compliant. Self-hosted on-premise support available.

Does it support table extraction?

Yes, the system recognizes and extracts data from complex tables, preserving row/column structure and supporting merged cells.

How does the document Q&A feature work?

Uses RAG (Retrieval-Augmented Generation) — the system finds the most relevant text passages from the document repository, then LLM synthesizes answers with accurate source citations.

Home

Document AI API

Experience Document AI firsthand and see how to integrate into your system

99.5%

OCR Accuracy

<2s

Per page processing

15+

Supported formats

95%+

Handwriting

Drag & drop files here or click to select

Supports PDF, DOCX, JPG, PNG, TIFF and many other formats

Model

Processing Capabilities

Vietnamese OCR: 99.5% accuracy
Handwriting: 95%+ accuracy
Tables: Preserves structure
Multi-page: 1000 pages/batch
Processing: <2s per page

Upload ảnh ở khung bên trái để bắt đầu.

Comprehensive solution suite

From handwriting recognition to AI Q&A on document repositories — a single platform handling the entire document lifecycle

Handwriting Recognition

Specialized OCR for Vietnamese handwriting — meeting minutes, surveys, handwritten forms with 95%+ accuracy

Vietnamese & Latin handwriting recognition
Complex handwritten form processing
Mixed (printed + handwritten) support
Custom handwriting model training

OCR & Smart Extraction

Recognize and extract key-value pairs from any document type — invoices, ID cards, contracts, tables

99.5% OCR for Vietnamese
Table extraction preserving structure
Automatic key-value extraction
15+ file format support

Document Search

AI semantic search — not just keyword matching but understanding query meaning to find the right documents

Semantic search with vector embedding
Cross-language search
Similarity scoring & ranking
Full-text + semantic hybrid search

Storage & Management

Structured document storage system — automatic indexing, versioning, metadata management for millions of documents

Integrated vector database
Auto-indexing on upload
Versioning & audit trail
Automatic metadata tagging

Document Q&A

Chat directly with your document repository — AI reads, understands and answers questions with accurate source citations

RAG (Retrieval-Augmented Generation)
Answers with source citations
Multi-document reasoning
Multi-turn conversation support

Classification & Routing

AI automatically classifies document types and routes to the right department — invoices, contracts, complaints, CVs...

50+ document types recognized
Custom taxonomy training
Auto-routing by workflow
Confidence score per classification

Document AI Pipeline

Upload

PDF, images, scans, handwriting

OCR & Extract

Recognition & extraction

Classify

Auto classification

Index & Store

Storage & indexing

Semantic search

Q&A

AI-powered Q&A

Detailed features

Each component in the pipeline is optimized specifically for Vietnamese documents

Handwriting Recognition

Specialized OCR model trained on millions of Vietnamese handwriting samples — recognizes meeting minutes, surveys, applications, medical records with 95%+ accuracy.

Handwriting recognition demo results

Full name: [handwriting]

Nguyen Van Minh97.2%

Address: [handwriting]

45 Le Loi, D.1, HCMC94.8%

Notes: [handwriting]

Payment due by March 1593.5%

99.5%

Printed text

95%+

Handwriting

96%+

Mixed content

92%+

Signature recognition

Search & AI Q&A

Semantic search that understands meaning instead of just matching keywords. Chat directly with your document repository — AI reads, understands and answers with source citations.

Semantic Search vs Keyword Search

Query: "compensation terms for early contract termination"

SemanticFound Article 12.3 — Liability for damages upon unilateral termination

KeywordNot found (no exact keyword match)

Document Q&A (RAG)

Question

When is the payment deadline for phase 2 of contract ABC?

AI Answer

According to Article 5.2 of Contract No. 2024/HD-DV/001, the phase 2 payment deadline is June 15, 2024, valued at 30% of the total contract.

Source: HD-DV-001.pdf, page 5, Article 5.2

Smart Storage

Integrated vector database for millions of documents

10M+

Documents / collection

<500ms

Index speed

<50ms

Search latency

Auto Classification

AI recognizes 50+ document types with confidence score

50+

Document types

97.8%

Accuracy

<1s

Processing

Summary & Analysis

Generate summaries, extract key insights from long documents

90% quality

Summary

Yes

Multi-doc

PDF/JSON

Export

Technical specifications

OCR Engine

99.5% accuracy for Vietnamese
95%+ handwriting
50+ languages supported
Table extraction
Layout analysis

AI/LLM

Integrated RAG pipeline
Multi-doc reasoning
Summarization
Entity extraction
Custom fine-tuning

Infrastructure

Self-hosted support
<2s/page processing
500+ req/min
Auto-scaling
GPU acceleration

Security

End-to-end encryption
GDPR compliant
ISO 27001
Data auto-delete
On-premise option

Frequently asked questions

Ready to automate document processing?

From handwriting OCR, semantic search to AI Q&A on document repositories — deploy in days

Contact for consultation