Question 1

What kinds of questions can I ask?

Accepted Answer

Health questions ("is orders_daily fresh"), diagnostic questions ("why did row count drop"), inventory questions ("which tables are not monitored"), and recommendation questions ("what should I monitor"). The AI answers with real data, not generic advice, because it can query your account directly.

Question 2

How does Intelligence Q&A avoid hallucinating answers?

Accepted Answer

Every answer is grounded in a tool call. The AI agent decides which query to run, runs it against your metadata, and cites the result in the response. If the agent cannot answer a question from the available data, it says so rather than guessing.

Question 3

Can I ask questions in a language other than English?

Accepted Answer

Yes. The underlying LLM handles major European and Asian languages well. Tool names and schema names stay in English (because your warehouse is), but questions and answers work in Spanish, French, German, Japanese, and others. Quality is highest in English.

Question 4

Where can I ask questions?

Accepted Answer

From the in-app Q&A panel, from any MCP-compatible client (Claude Code, Cursor), from the CLI with `armor ask`, or from the Python SDK with `client.intelligence.ask()`. All four entry points hit the same endpoint and give the same answer.

Question 5

Is Q&A usage metered?

Accepted Answer

Q&A queries are unmetered for Professional and above. Starter includes 100 queries per month. A query is a single user-facing question; the agent can run multiple underlying tool calls to answer it without consuming additional quota.

Intelligence Q&A

Transform Data Questions into Insights

Natural Language Queries

Instant Explanations

Context-Aware Analysis

Ready to ask your data questions?

Frequently Asked Questions