Google Cloud Storage adds AI object enrichments for unstructured data
Google Cloud has introduced object enrichments, a new Cloud Storage feature that uses AI models to extract metadata from images, PDFs, videos and audio files and stores it alongside the objects. The preview service aims to make unstructured data usable by AI agents without custom pipelines.

Key takeaways
- Google Cloud has introduced object enrichments for Cloud Storage in preview.
- The feature extracts AI-generated metadata from images, PDFs, videos and audio and stores it as object context alongside the data.
- Google Cloud claims enriching 10 million images or PDFs costs $700 and offers 75 percent savings versus similar services from other providers.
- Customers include L'Oreal, Hyperscience and Vivint, according to statements in the announcement.
In this article
What Google Cloud has announced

Google Cloud has launched object enrichments, a new capability for Cloud Storage that automatically extracts contextual metadata from unstructured objects such as images, PDFs, videos and audio files. The feature uses AI models to process objects and saves the results in-place as object context, a structured metadata construct stored alongside the original bytes. The service is available in preview, according to the Google Cloud Blog.
How the feature works
Object enrichments lets customers choose from five first-party Cloud Storage models for images and PDFs. These include image annotations such as object detection, face detection, image labeling and explicit content detection, plus a PDF classification model that sorts documents into categories such as Finance, Games, News and Health. Setting up an enrichment job takes three steps: define the buckets and processing location, choose the models, and grant the necessary IAM permissions. The service then handles model inference, response parsing and writing results to object context, with built-in retries and throttling.
Cost and performance claims
Google Cloud says enriching 10 million images or PDFs costs $700 based on published pricing, and that image annotations are available at 75 percent cost savings compared with similar services from other cloud providers. The company also states that customers can use Gemini natively with object enrichments for complex multimodal reasoning across text, PDFs, images, audio and video, enabling summarization, entity extraction, transcription and translation. Access to this upcoming functionality requires contacting Google Cloud at [email protected].
Who is affected and what it means
The feature is aimed at enterprises running AI agents in production that need grounded knowledge from large volumes of unstructured data. Google Cloud says it removes three common challenges: high model costs at scale, operational toil from self-managed pipelines, and fragmented metadata stored separately from the objects. Because object context travels with objects during copy, move and replication operations and is deleted when the source object is removed, customers do not need separate cleanup jobs. Fine-grained IAM controls let administrators manage who can read, add, update or remove object context.
Customer statements
Antoine Castex, principal data and AI architect at L'Oreal, said object enrichments lets the company label its creative asset repository 100 times faster at a fraction of the cost of existing pipelines. Mayur Pillay, VP of partnerships at Hyperscience, said the feature drove a 50 percent plus lift in document classification accuracy while reducing classification inference costs by two to three times. Brandon Bunker, VP of product AI at Vivint, said Cloud Storage object enrichments and BigQuery cut AI development from six months to one month.
Sources
Spotted a mistake? See our corrections policy. Read how News by OnDevice works in our editorial policy.



