Title: AISAY - AI Document Parser

Overview

AISAY is an AI-powered document reader that can provide structured responses, extract, and integrate information from various documents into existing systems. It can automatically detect, scan, and comprehend a variety of documents. This includes handwritten notes, printed articles, multilingual text, and structured/unstructured documents.

Unlike traditional Optical Character Recognition (OCR) systems, AISAY can provide structured responses for documents such as images (JPEG, PNG, TFIF) and PDF files. For documents issued by local entities (businesses, governments, and institutions), AISAY can even fine-tune the documents prior to processing.

You can find out more about AISAY from the info deck below.




Key Features

Ability to detect, scan and comprehend a variety of documents

AISAY can process documents such as handwritten notes, printed articles, multilingual text, and structured/unstructured documents. Unlike traditional Optical Character Recognition (OCR) systems, AISAY can provide structured responses for documents such as images (JPEG, PNG, TFIF) and PDF files. For documents issued by local entities (businesses, governments, and institutions), AISAY can even fine-tune the documents prior to processing.

AISAY supports the following document types, but other document types can be supported on request.

  1. ID Documents such as NRIC/employment pass/S pass/work permit/dependent pass/long-term visit pass/student pass/driver’s license/11B or passports (machine-readable)
  2. Agreements and contracts
  3. Invoices/receipts/delivery orders
  4. Bank statements

For latest features, please refer to this page.




How it Works

AISAY accepts image files (JPEG, PNG, TIFF) or PDF documents and generates structured JSON responses with extracted data.

Using Optical Character Recognition (OCR), Document Question-Answering (DQA) techniques, and a Large-Language Model (LLM), AISAY analyses text position and content within documents to provide context and answer queries accurately. AISAY conducts thorough pre-processing and post-processing of documents to ensure precise results.

AISAY scales effectively with options for synchronous and asynchronous invocation, facilitating the handling of large file sizes. It assigns a confidence score to each task to assess the need for further assistance. Tasks with higher confidence levels can proceed with minimal human intervention, while those with lower confidence may require human review or additional verification.




Code for Getting Started

  • 👆🏽 Click on the "Open Notebook" button below to open the Jupyter Notebook 
API Key can be found on Canvas ("Recharge & Prep" between Topic 5 and 6)
The documentation of the API is available here.



Web App for AISAY

From a GSIB laptop or equivalent, you can also access the Graphical User Interface (GUI) for AISAY from https://launchpad.gov.sg