This feature is subject to the "Pre-GA Offerings Terms" in the General Service Terms section of theService Specific Terms. Pre-GA features are available "as is" and might have limited support. For more information, see thelaunch stage descriptions.

This page describes how to use AI-powered assistance throughGemini to help you better understand your data storage landscapein Cloud Storage. When you use Gemini Cloud Assist, you can enterprompts about how Cloud Storage works in general, and when you enable aStorage Intelligence subscription, you can enter prompts about specificbuckets and objects.

You can then use the information provided by Gemini Cloud Assist todo the following:

• Analyze data usage patterns and trends

• Identify opportunities to save on storage costs

• Optimize your data for security and compliance

• Make informed decisions on managing your data

Gemini doesn't use your prompts or itsresponses as data to train its models without your express permission. For moreinformation about how Google uses your data, seeHow Gemini for Google Cloud uses your data.

This page is intended for developers, data analysts or data engineers, platformadmins, finance operators, and compliance officers. It assumes that you knowhow to use Cloud Storage and querylinked datasets.

What prompts can Gemini answer?

To understand your data storage, you can provide Gemini Cloud Assistwith aprompt, which is a natural language statement or question.

When you use Gemini Cloud Assist alone without aStorage Intelligence subscription, you can ask prompts about howCloud Storage generally works, such as the following:

• "How do I transition objects from one storage class to another?"

• "How do I enable soft delete on a bucket?"

When you enable aStorage Intelligence subscription,you can also ask prompts related to cost saving opportunities, security andcompliance, and data discovery. Gemini Cloud Assist can use themetadata contained inStorage Insights datasets to generateinsights,which are responses to prompts about your bucket and object metadata and usage.You can ask prompts such as the following:

• Prompts related to usage and cost savings:

  • "5 largest buckets without Object Lifecycle Management"

  • "5 largest buckets without Autoclass enabled"

• Prompts related to security and compliance:

  • "5 largest buckets without object versioning enabled"

  • "5 largest objects with a retention expiration date within the next 30 days"

• Prompts related to data discovery:

  • "Buckets with a high volume of small files (each under 1 MB)"

  • "Objects in Standard storage class smaller than 50 MB"

Before you begin

In order to use Gemini Cloud Assist for general prompts related toCloud Storage, you must first set up Gemini Cloud Assist, includinggetting required roles.

If you want to enter prompts related to specific buckets and objects, you mustalso complete the following prerequisite steps:

1. Enable Storage Intelligence, which gives you access to usingStorage Insights datasets.

2. Create a Storage Insights dataset, which Gemini Cloud Assist willanalyze to provide information about specific buckets and objects.

   Alternatively, if there's an existing dataset you want to use, you canget the required IAM roles for accessing the existing dataset.

3. Ensure that the Storage Insights service agent has access to the datasetGemini Cloud Assist will analyze. This enables the dataset to beread and analyzed.

Analyze your data storage by using natural language prompts

As an early-stage technology, Gemini for Google Cloud products can generate output that seems plausible but is factually incorrect. We recommend that you validate all output from Gemini for Google Cloud products before you use it. For more information, seeGemini for Google Cloud and responsible AI.

1. In the Google Cloud console, go to the Cloud StorageStorage Insights page.
   

   Go to Storage Insights

2. In the toolbar, clickspark (Gemini)to open theCloud Assist chat panel.

   TheCloud Assist chat panel appears.

3. In theCloud Assist chat panel, enter a natural language promptabout your data storage. For example, you might enter the following:

   Which is my largest bucket

4. Clickplay_arrow (Generate).

5. If prompted to, enter the name of the dataset that Geminiwill analyze to generate the response, then clickplay_arrow (Generate).

   If successful, Gemini Cloud Assist generates a response similar tothe following:

   Here's what I found by analyzing the data inEXAMPLE_DATASET:
   Bucket name
   Size
   my-bucket
   39.1 TB

   The underlying SQL query that Gemini Cloud Assist uses is alsoreturned. The generated SQL query is similar to the following:

   SELECT bucket_id, bucket_sizeFROM bucketsWHERE project_id = 'example-project'ORDER BY bucket_size DESCLIMIT 1;

Optionally, you can enter suggested prompts:

1. In the Google Cloud console, go to the Cloud StorageStorage Insights page.
   

   Go to Storage Insights

2. In theauto_awesomeSuggested promptssection, select a suggested prompt. For example, a suggested prompt mightsay:Storage size broken down by object content type.

3. If successful, Gemini Cloud Assist generates a response similar tothe following:

   Here's what I found by analyzing the data inEXAMPLE_DATASET:
   Content type
   Size
   MP4
   483.2 GB
   MOV
   239.1 GB
   MP3
   125.8 GB

   The underlying SQL query that Gemini Cloud Assist uses is alsoreturned. The generated SQL query is similar to the following:

   SELECToa.contentType, ROUND(sum(oa.size) / (1024 * 1024 * 1024), 2) AS total_size_gbFROM object_attributes_latest AS oaGROUP BY oa.contentTypeORDER BY sum(oa.size) DESC;

Limitations

• When using prompts to analyze Cloud Storage resources, you canspecify up to five resources in the prompt. For example:5 largest buckets without Autoclass enabled. Even if you include more thanfive resources in the prompt, Gemini Cloud Assist can only returnresults for the top five resources that match the prompt's criteria.

• Gemini Cloud Assist uses the bucket and object metadata contained inStorage Insights datasets to respond to prompts about your data storage.

  The following data is not available in Storage Insights datasets, meaningthat Gemini Cloud Assist lacks the proper context to respond toprompts relating to that data:

  • Specific cost data, for example: "how much does my bucket cost per month".

  • Activity data, for example: "what is the last access time of my object".

  • Data about the configuration of certain features, such assoft delete. For example: "which buckets don't havesoft delete enabled".

  Gemini Cloud Assist also lacks the proper context to respond toprompts related to time series information. For example: "how much did mybucket grow in the last 3 weeks". This is because Gemini Cloud Assistonly reads a dataset's latest snapshot.

What's next

• Learn how towrite better prompts.
• Learn how to use theGemini Cloud Assist panel.
• ReadUse Gemini for AI assistance and development.
• Learnhow Gemini for Google Cloud uses your data.