Skip to main content

Akabot Vision

Akabot Vision is the intelligent document processing (IDP) module that reads, classifies, and extracts structured data from unstructured sources — scanned invoices, contracts, forms, emails, and images. It combines OCR, machine learning, and configurable extraction templates to turn paper-based and digital documents into actionable data for your workflows.

Akabot Vision

Key features​

  • OCR engine — High-accuracy text extraction from scanned PDFs, images (PNG, JPEG, TIFF), and digital documents.
  • Document classification — Automatically route incoming documents to the correct extraction template based on content.
  • Field extraction — Define extraction rules with a visual template editor; no ML expertise required.
  • Table detection — Identify and extract multi-row tables from invoices, purchase orders, and reports.
  • Confidence scores — Every extracted field carries a confidence score so low-confidence items can be routed for human review.
  • Human-in-the-loop — Built-in validation station for operators to review and correct uncertain extractions before data enters downstream systems.

Supported document types​

  • Invoices and purchase orders
  • Identity documents (passports, national IDs)
  • Contracts and legal agreements
  • Bank statements and financial reports
  • Medical records and forms
  • Custom document types via template training

Getting started​

  1. Upload sample documents to the Vision portal.
  2. Create a document type and label fields using the template editor.
  3. Train the model and validate extraction accuracy.
  4. Use the Vision activities in Studio to send documents and retrieve extracted data.

Next steps​