AI Assistant

The AI Assistant is the Home page of CloudQuery Platform. It lets you ask questions about your cloud infrastructure in plain language and get answers — without writing SQL or knowing which tables to query.

Unlike the AI Query Writer in the SQL Console, which is scoped to generating SQL, the AI Assistant is a full conversational agent. It can explore your schema, reason across multiple tables, and — depending on your organization’s settings — execute queries and return results directly in the chat.

Data access modes

The AI Assistant operates in one of two modes, configured per organization:

Mode	What the assistant can do
Data Privacy (default)	Reads table names and column definitions only. Returns SQL queries for you to run manually or open in the SQL Console.
Full Access	Executes queries on your behalf and returns results directly in the chat.

Data Privacy mode is the default. Every organization starts with it enabled, and it cannot be bypassed by individual users.

Enabling Full Access

Full Access mode is controlled by organization administrators:

Go to Organization Settings → Platform Settings
Find the AI assistant data access section
Toggle Enable AI agent data access on
The setting applies to all users in the organization immediately

Full Access mode allows the AI Assistant to execute SQL queries against your CloudQuery database and include results in its responses. Enable it only if your organization’s policies permit AI systems to read live infrastructure data.

How to use it

Open the Platform Home page and type your question in the chat input. The assistant searches your schema, figures out which tables are relevant, and responds with either a direct answer (Full Access mode) or an executable SQL query (Data Privacy mode).

Example prompts:

Show count of virtual machines across all cloud environments
Which S3 buckets are publicly accessible?
List IAM users that have not rotated their access keys in the last 90 days
What EC2 instances are running in production?

You can follow up in the same conversation to refine results, narrow scope, or ask clarifying questions. The assistant retains context across messages within a session.

If you ask for a SQL query directly, the assistant will generate one. You can open it in the SQL Console, save it as a policy, or run it as-is.

Rate limits

The following limits apply per organization:

Limit	Value
Requests per minute	10
Tokens per request	500,000
Tokens per day	5,000,000

These limits are subject to change during the initial launch period.

What the AI Assistant cannot do

No web access. The assistant cannot fetch external URLs, load documentation, or analyze websites.
No cross-product features. Generating reports, reading policy violation details, and analyzing Insights are out of scope for the current release. These capabilities will be added over time.
No off-topic conversations. The assistant has guardrails that keep it focused on cloud infrastructure and CloudQuery-related topics. Unrelated requests will be declined.
No persistent memory. Each new conversation starts fresh. Use Custom Context to carry organization-specific knowledge into every session automatically.

Data privacy

In the default Data Privacy mode, the assistant never reads or transmits row data from your CloudQuery database. It only sees table names and column definitions.

The assistant can see	The assistant cannot see
Table names in your CloudQuery database	Row data of any kind
Column names and their data types	Values stored in any column or field
SQL syntax validation results	Credentials, connection strings, or API keys

In Full Access mode, the assistant executes SQL queries and reads result sets. Only the rows returned by those queries are processed — the assistant does not have unrestricted access to your CloudQuery database.

The assistant is powered by Claude (Anthropic) running on AWS Bedrock. AWS Bedrock does not use customer data to train or improve foundation models. Your queries, schema information, and query results are not retained for model training purposes.

FAQ

Which AI model does this use?

Claude Sonnet 4.5 and Haiku 4.5 (Anthropic), running on AWS Bedrock. Sonnet 4.5 handles the main agent loop; Haiku 4.5 handles a lightweight scope-check that runs before every request.

How is this different from the AI Query Writer in the SQL Console?

The AI Query Writer is a focused tool inside the SQL Console that always operates in schema-only mode and always returns a SQL query. The AI Assistant is a broader conversational agent that lives on the Platform Home page, supports both Data Privacy and Full Access modes, and can answer questions directly without requiring you to run SQL yourself.

Can I use both?

Yes. They serve different workflows. Use the AI Assistant for exploration and questions; use the SQL Console and AI Query Writer when you want to author, save, and reuse queries.

Is my data used to train the model?

No. AWS Bedrock does not use customer data to train or improve foundation models.

Who can change the data access mode?

Only organization administrators. The setting is at Organization Settings → Platform Settings → AI Assistant Data Access and applies to all users in the organization.

Next steps

Custom Context — Give the assistant knowledge about your specific environment
SQL Console — Save and run the queries the assistant generates
AI Query Writer — The SQL-focused assistant inside the SQL Console
MCP Server — Connect AI tools like Claude Desktop or Cursor directly to your CloudQuery database

Was this page helpful?

Last updated on April 7, 2026