Skip to main content

Prism AI Features

Context Mentions​

Use @ to provide specific context to Prism AI.

Prism AI Autocomplete

Types​

MentionWhat It Provides
@project.dataset.tableTable schema, columns, partitioning
@datasetDataset information
@Tab 1SQL and results from tab
@resultsQuery results from active tab

Examples​

Optimize @Tab 1 for cost
Find tables with email data in @analytics
Analyze @results for data quality issues
Compare schemas of @dataset1.table1 and @dataset2.table2

How Context Works​

Context mentions are the primary way to share specific data with Prism AI. Without mentions, the AI can still:

  • List tables in your projects (via list_tables tool)
  • Fetch table schemas (via get_table_schema tool)
  • Read and update your query tabs (via tab tools)

Mentions give you explicit control over what context is included in your conversation.

Query Execution​

Prism AI can run queries and show results directly in the chat.

When you ask the AI to query data, it executes against BigQuery using your credentials and displays results inline. A "Share with AI" toggle controls whether the AI can see the actual result rows — when off, the AI only sees metadata (row count, schema) but not the data itself.

Follow-Up Suggestions​

After each response, Prism AI may show clickable suggestion chips for common next steps. Click any chip to send it as your next message.

Reasoning Display​

When using reasoning-capable models (like Claude Sonnet 4.5 or GPT-o series), Prism AI shows the model's thinking process in a collapsible block before the final answer. Click to expand and see the full reasoning chain.