VNPT AI Ecosystem

mic

Workflow Automation
Solution

A unified platform to read - understand - extract - act on documents
enabling digitization & data extraction (OCR/KIE/DocSum) to create structured data
and automating simulated user processes for cross-checking,
data entry, and coordination across existing systems.

Contact us
Outstanding Features
Character Recognition (OCR) & OCR Plus
Basic/Standard/Advanced: From printed text recognition and layout analysis (paragraphs/tables/images) to enhanced recognition of handwriting, signatures, seals, and LaTeX formulas.
KOCR Plus: Reconstruct editable DOCX/XLSX files while preserving paragraph/table formatting for convenient post-checking and storage.
Recommended Input Quality: ≥150 dpi (PDF/scan) or FHD images; supports asynchronous multi-page processing.
ocr
Key Information Extraction (KIE)
Standard: Template/layout-based extraction, featuring a library of common document types (administrative texts, invoices, business registrations, land use rights, etc.).
Advanced (VLM): Template-independent extraction, achieving ~99% accuracy for 1-page documents and ~95% for multi-page documents (data-dependent).
nlp
Document Summarization (DocSum)
Summarize PDF/image/text documents up to ~20 pages or ~8,000 words, defaulting to about 30% of original content length (customizable).
genAI
Robotic Process Automation (RPA)
Automatically log in, read profiles, check rules, enter data, generate documents/vouchers, and compile reports; suitable for attended/unattended automation, weekly deployments, and excellent component reusability.
sematic
Flexible Deployment
Supports on-cloud deployment on VNPT infrastructure or on-premise on customer infrastructure, serving diverse user groups while fully adhering to security and safety requirements.
sematic
The values we deliver
Productivity & Speed

24/7 batch processing; real-world references show ~10–30s/page with standard configs, and CCU clustering scales out.

Cost & Error Reduction

Eliminate repetitive data entry tasks; with average ROI references under 6 months, process costs can significantly decrease (depending on context).

Fast Deployment – Zero Core Modification

Operates via the UI, ideal when interacting with multiple packaged applications and systems.

Security & Compliance

On-premise deployment, temporarily storing input data; no transaction logs stored by default; license mechanism controls duration/hardware/quota.

Typical Use Cases

Industry Solutions

Extract ID numbers, abstracts, dates, classification, and routing.

Finance – Accounting

Invoices, vouchers; auto reconciliation & recording; exporting DOCX/XLSX from OCR Plus for storage.

HR – Records

Extract data from forms, update HRM systems, compile periodic reports.

Security Architecture & Performance References

Microservices on‑prem
  • Processing backend, message queues, AI inference, object storage, DOCX/XLSX rendering, licensing services.
icon-quanly-1
Data
  • Temporarily store original files on Object Storage (e.g. MinIO compatible); existing DB/Object Storage infrastructure can be reused.
icon-quanly-2
License
  • Control by duration/hardware/quota; monitor processing states.
icon-quanly-3
Performance Reference
  • Response time
    ~10–30s/page, CCU/node
    ~5–15 (depending on GPU/version); linear scaling by nodes.
  • Recommended A-class GPU (VRAM 24–80GB) for advanced Recognition/KIE/DocSum; blueprint configs available for clusters serving ~40 CCUs.
icon-quanly-4
Outstanding Business Applications

Public Administration

Receive, verify document components, cross-check conditions against templates.
Automatically notify results/deficiencies.
Integrate OCR-RPA in classifying components of electronic notarization public administrative procedures.
Automate workflows to assist administrative procedures; ~90% automation rate, saving ~450 working hours/year on pilot scale.
Integrate OCR-Screening for public health administrative procedures: Medical/Pharmaceutical/Cosmetic practice licensing.

Hành chính công

Read/cross-check vouchers (invoices, deposit slips).
Extract information based on templates.
Automatic accounting data entry.
Reconcile receivable/payable accounts.
Compile reports.

Services (Customer Service & Enterprise Ops)

CS/Back‑office: Extract data from emails/forms, update CRM/DWH
Periodic reporting synthesis
SLA alerts.

HR - Procurement – Supply Chain

Extract data from contracts/orders/vouchers
Input into destination systems
Track progress & quality
reader.rpa_price_title

Document AI Block (OCR/KIE/DocSum)

Licenses/processing capabilities managed by duration, hardware, and quota; costs depend on document templates, KIE fields count, and page volumes.
Refer to the OCR pricing/information table at: https://vnptai.io/smartreader/en https://vnptai.io/smartreader/vi

Khối AI tài liệu (OCR/KIE/DocSum)

Determined by process scale (attended/unattended), number of robots, developers, and orchestrator infrastructure; requires detailed process surveys to optimize cost and ROI.
Please contact the consulting team to receive appropriate sizing proposals, implementation roadmaps, and quotes based on actual document volumes, number of processes, and integration levels.