Google Document AI is a Google Cloud service that lets developers create high-accuracy processors to extract structured and unstructured data from documents, classify and split documents, and automate tedious document tasks. It includes a generative AI-powered custom extractor that can be fine-tuned with as few as 10 documents, a custom splitter for composite documents, a custom classifier, Enterprise Document OCR, a form parser, a layout parser, and a summarizer. Processors can be built through the Google Cloud console UI or API, and workflows can automate document ingestion, processing, and storage. It integrates with BigQuery and Agent Search, and is intended for developers and organizations handling mail rooms, mortgage processing, procurement, and archives. Pricing is pay-as-you-go per processed page, and new customers get $300 in free credit.
How to use Google Document AI
1
Set up the Document AI API
Enable the Document AI API in your Google Cloud project following the official setup documentation.
2
Select a parser in the console
Use the Google Cloud console to choose the parser that fits your project needs, such as Enterprise Document OCR, custom extractor, custom splitter, or custom classifier.
3
Train a custom processor
Fine-tune a generative AI-powered processor with as few as 10 documents, or build one from the ground up with your own documents and custom classes.
4
Process a document
Try processing a document with Document AI to extract text, layout, structures, and entities.
5
Evaluate processor performance
Review precision, recall, F1 score, and other metrics for each parser using the evaluation documentation.
Google Document AI Features
Custom extractor powered by generative AI, fine-tunable with as few as 10 documents
Custom splitter for composite documents with custom classes
Custom classifier with few-shot learning and iterative auto-labeling
Enterprise Document OCR for text and layout extraction
Form parser and layout parser for structured data
Generative AI-powered document summarization
Build processors through the Google Cloud console UI or API
Workflows to automate document ingestion, processing, and storage
BigQuery integration for document analytics
Agent Search integration
Google Document AI Use Cases
Automated document data extraction for mail rooms, shipping yards, mortgage processing, and procurement
Splitting composite documents such as mortgage packages into logical documents
Classifying documents with custom classes
Digitizing archival text for ML model training
Extracting document metadata into BigQuery for analytics
Summarizing large documents
Google Document AI FAQs
It is a Google Cloud service that lets developers create high-accuracy processors to extract structured and unstructured data from documents, classify and split documents, and automate document tasks.
Cost is based on the number of processed pages per month, the relevant quota, and any purchased capacity reservation. Each parser has its own per-1,000-page rate.
New customers get $300 in free credit to try Document AI and other Google Cloud products.
You can achieve higher accuracy by providing as few as 10 documents to fine-tune the large model.
Yes. Custom splitter identifies each logical document and splits composite documents into single-class documents, and can be trained with your own documents and custom classes.
Yes. You can extract metadata from documents directly into a BigQuery objects table and join parsed data with other BigQuery tables.
You can use the Google Cloud console UI or the API to create processors, and enable workflows to automate document ingestion, processing, and storage.
Document AI supports many regions and multi-regions, but functionality varies by region; see the official regions documentation for details.
Cost is based on the number of processed pages per month, the relevant quota, and any purchased capacity reservation. New customers get $300 in free credit.
Enterprise Document OCR
$1.50USD / per 1,000 pages, 1โ5,000,000 pages per month
Pay-as-you-go pricing for digitizing text with the Enterprise Document OCR processor.
Detects and extracts text and layout information from documents
Configurable features to tailor document processing
Billed per 1,000 pages processed
OCR add-ons
$6USD / per 1,000 pages, 1โ5,000,000 pages per month
Additional OCR capabilities billed per 1,000 pages.
Add-on options for OCR processing
Billed per 1,000 pages processed
Custom extractor
$30USD / per 1,000 pages, 1โ1,000,000 pages per month
Extract structured data from documents with a generative AI-powered custom extractor.
Works out of the box across a wide array of documents
Fine-tune with as few as 10 documents
Billed per 1,000 pages processed
Form parser
$30USD / per 1,000 pages, 1โ1,000,000 pages per month
Extract structures and entities from forms, billed per 1,000 pages.
Extracts structures and entities from documents
Billed per 1,000 pages processed
Layout parser
$10USD / per 1,000 pages, 1โ1,000,000 pages per month
Parse document layout, including initial chunking, billed per 1,000 pages.
Includes initial chunking
Billed per 1,000 pages processed
Custom splitter
$5USD / per 1,000 pages, 1โ1,000,000 pages per month
Split composite documents into single-class documents by identifying each logical document.
Usable out of the box or trained with your own documents
Supports custom classes
Billed per 1,000 pages processed
Custom classifier
$5USD / per 1,000 pages, 1โ1,000,000 pages per month
Classify documents with a custom classifier built from your own documents and classes.
Few-shot learning and fine-tuning
Iterative auto-labeling for corrections
Billed per 1,000 pages processed
Summarizer
$25USD / per 1,000 pages, 1โ1,000,000 pages per month
Summarize large documents with a generative AI-powered solution, billed per 1,000 pages.