Diagram
A four-step process flow showing unstructured content like documents, images, and audio flowing through AI extraction services (OCR, Speech-to-Text, Generative AI) into BigQuery analytics and machine learning, and finally into business value through automation, prediction, and insights.
How AI Services Extract Meaning from Documents, Images, and Speech
AI extraction services convert unstructured inputs into text and structured outputs that analytics and applications can use. Optical Character Recognition (OCR) extracts text from images and documents, so a scanned invoice becomes searchable fields. Speech-to-Text converts spoken language into written text, so a customer call becomes a transcript that can be searched, summarized, and analyzed. Vertex AI Translation translates text between multiple languages, so the same content can be used across regions and teams.
Generative AI on Vertex AI builds on this foundation by creating new content and answers from the extracted meaning. An application sends documents, transcripts, or images to a generative model, and the model returns summaries, classifications, answers, or drafted content. Vertex AI Workbench supports this work by providing a managed JupyterLab notebook environment where analysts and developers explore data, test models, and build machine learning workflows.
Domain 5 of 7
27 practice questions
Practice this DomainLast updated