Achieved 98% document level accuracy, and data inference accuracy rates of 99.999%+ across tens of millions of documents per customer each year with Cloud Vision API
Removed all instances of personally identifiable information from over 1 million documents in two days with Gemini Pro and Gemini Flash on Vertex AI
Delivered AI-powered document processing and workflow automation services in 300+ languages with Google AI and infrastructure services
Provided services to customers that generate up to 50 million documents per year, per client, with Google AI and infrastructure services
Ensured user data security compliance with Google Workspace console management features
Using a foundation of Gemini Flash on Vertex AI, Cloud Vision API, and Cloud Translation API, Staple AI provides highly accurate AI-powered document processing and workflow automation services in 300+ languages to customers that generate up to 40 million documents per year.
With its automated, AI-driven document processing solutions, Staple AI helps organizations capture and analyze data from any type of document — such as invoices, contracts, ID cards, or data files — regardless of its language, layout, or format. This reduces the workload on customer teams that would otherwise have to invest considerable time and resources to complete this task across up to 50 million documents per year per client. To deliver its solutions, Staple AI uses Cloud Vision API for optical character recognition (OCR), Cloud Translation API for translation, and Gemini Flash on Vertex AI for data extraction and summarization at scale. BigQuery and Google Kubernetes Engine provide the foundation of the Staple AI platform. The organization also uses Google Workspace to strike a balance between the need to meet strict compliance standards, such as SOC 2, GDPR, and HIPAA, and staying fast and collaborative as a lean team.
As a company that processes documents for Fortune 500 companies, if we fail at the OCR stage, we've failed at the first hurdle. Accuracy is our top priority, and with Google Cloud, we get the best.
Dr Josh Kettlewell
Chief Technology Officer and Co-founder, Staple AI
To provide its solutions, Staple AI needed AI products that could meet strict SLAs of 98% document straight through processing accuracy and 99.999% data extraction and OCR accuracy, as well as support multiple languages. The organization chose Cloud Vision API for its superior accuracy and native handwriting and non-Latin character processing, and Cloud Translation API for its reputation and performance in translation. Through testing its accuracy in data extraction, Staple AI confirmed Gemini Models suite was superior to alternative models and the best fit for its business needs. With Google AI, the organization was able to process over 1 million documents in two days, redacting personally identifiable information with near-100% accuracy to ensure that a financial services client met compliance requirements. The organization also uses Gemini with Google Workspace to improve workforce productivity by automating meeting transcription, email summarization, and document creation.
When we measure success, we can't just measure accuracy — we measure time saved for users and the value we deliver to our customers. Google Cloud helps us reach our mission to provide a holistic solution for document processing and workflow automation.
Dr Josh Kettlewell
Chief Technology Officer and Co-founder, Staple AI
Staple AI provides a platform that learns, classifies, and extracts data from documents in 300+ languages, and integrates them into business systems. Its platform can ingest any type of document — such as invoices, contracts, ID cards, or data files — regardless of its language, layout, or format. The AI then automates the manual tasks a person would normally perform, such as categorizing and splitting documents, extracting key data, and routing them through workflows.
Industry: Technology
Location: Singapore
Products: Cloud Translation API, Cloud Vision API, Gemini Flash, Gemini with Google Workspace, Vertex AI, BigQuery, Google Kubernetes Engine