Google Cloud Storage adds AI labels to 10 million files for $700
Google tells that the new tool adds labels to images and PDF files for 75% less than other cloud providers.
Google released object enrichments in preview. It is a new part of Cloud Storage, the Google tool that keeps files. AI models read images and PDF files and keep labels with each file. Google tells that 10 million files cost $700 at its published prices. The 75% saving is a claim by Google. Agents can then use the labels.
How it works
A customer picks its storage buckets. It chooses one or more of five Google models for images and PDF files. The user writes no code.
The tool runs the models and tries again after an error. It writes the results as key-value pairs with each file. The labels move with the file when a user copies it.
The numbers
All numbers come from Google and its customers. No other group did a test.
- L’Oreal tells that it labels its creative files 100x faster.
- Hyperscience tells that its classification accuracy rose by more than 50%.
- Vivint tells that AI development fell from 6 months to 1 month, with BigQuery.
What is not known
Google does not name the other cloud providers in the 75% claim. The Gemini tool for audio, video and summaries is not open at this time.
Insight: Google puts each layer of agent context in its cloud
In November Google added business definitions, agent records and cloud rules for agents. At this time it adds file labels. All of them stay in Google services. This reads as a race to control the context that agents use, because labels kept with files in Google Cloud are costly to move.
Sources
Posted