What challenges did Deutsche Bank face before AI implementation?

Digitizing high-volume document workflows and extracting or summarizing information from complex, unstructured documents. Reducing time to train custom models while improving accuracy, operational risk, and fraud prevention. Handling long documents and enterprise-specific document formats at scale.

What was the impact of the AI solution?

Deutsche Bank said the workflow can reduce operational risk, increase capacity, and improve customer experience. BBVA said it can extract data from complex, highly dense, non-structured documents and help prevent errors and potential fraud for faster, more accurate, and more secure service. Orby AI said Document AI-powered contract processing has delivered up to 70% cost savings.

Source

GCPExpanded

Document AI Workbench early customer evidence for generative extractors/summarizer (Deutsche Bank, BBVA, Iron Mountain etc.)

Use case typeDocument automationUpdated Jun 13, 2026

Google Cloud introduced Document AI Workbench generative features for custom document extraction and summarization. Early-access customers including Deutsche Bank, BBVA, Iron Mountain, Orby AI, and Contextual AI described using Document AI Workbench to digitize high-volume document workflows, extract structured data from complex documents, and summarize content. The article says the Custom Extractor with generative AI can extract data from free-form and structured documents with no training required to start, then improve accuracy through confirmed predictions and few-shot examples. The Summarizer can generate configurable summaries for long documents up to 250 pages.

Organization: Deutsche Bank
Industry: Finance
Location: United States
Published: September 2023

Reported outcomes

−70%

costCost savings

Strategic outcomes

Risk & complianceReduced operational riskScale & capacityIncreased processing capacityCustomer experience & trustImproved customer experienceRisk & compliancePrevented errors and fraud

Catalog median for cost savings deployments: −45% across 345 reported metrics. Compare benchmarks →

Primary read

Use case focus

Showing 3 of 4

1Document processing automation
2Intelligent document processing
3Generative AI summarization

Digitizing high-volume document workflows and extracting or summarizing information from complex, unstructured documents.
Reducing time to train custom models while improving accuracy, operational risk, and fraud prevention.
Handling long documents and enterprise-specific document formats at scale.

Google Cloud Document AI Workbench provides a Custom Extractor with generative AI for structured data extraction.
Users define fields, upload sample documents, confirm or correct predictions, and leverage few-shot examples to improve model accuracy.
The platform also offers a Summarizer that can generate configurable summaries in brief, moderate, or comprehensive formats and bullet or paragraph styles.
The workflow exposes API endpoints so customers can call the models directly in production.

Technologies

Google Cloud Document AI Workbench Google Cloud Document AI Vertex AI Generative AI

Deutsche Bank said the workflow can reduce operational risk, increase capacity, and improve customer experience.
BBVA said it can extract data from complex, highly dense, non-structured documents and help prevent errors and potential fraud for faster, more accurate, and more secure service.
Orby AI said Document AI-powered contract processing has delivered up to 70% cost savings.

Architecture

Document AI Workbench lets teams define extraction fields, upload documents, use generative AI-powered custom extractors, confirm or correct predictions to improve accuracy with few-shot examples, and deploy endpoints for production document extraction and summarization. The Summarizer supports configurable length and format for long documents.

Implementation partners1

Automation Anywhere

Sources & evidence1

ExpandedExpanded

The same organization appears in newer AI deployment evidence.