Semantic View Quick Start

Precog automatically generates a Snowflake semantic view when you create an AI Assistant, enabling natural language queries through Snowflake Cortex Analyst.

Quick Start

  1. Create an AI Assistant — The setup wizard walks you through connecting your data, describing your business context, and configuring your assistant. The semantic view is generated automatically as part of this flow.

  2. Review your Knowledge Base — Approve or refine the business facts (KPIs, definitions, rules) that were discovered during setup. These drive the quality of your semantic model.

  3. Explore the Semantic Layer — View datasets, fields, relationships, and metrics. Download the model as JSON or regenerate it after knowledge base changes.

  4. Connect your AI tools — Get your MCP URL and per-client setup instructions for Claude, ChatGPT, GitHub Copilot, Cursor, and more.

Key Concepts

Use case descriptions matter. The quality of your semantic view depends on how well you describe your business context during setup. Include specific metrics, entities, and the types of questions you want to answer. See Writing Effective Use Case Descriptions for examples.

Multi-source models. When your connection includes multiple data sources (e.g., HubSpot + Xero), Precog generates a unified semantic view that spans all of them — enabling cross-source questions without manual SQL joins.

Snowflake Cortex Analyst. After generation, the semantic view is available in Snowflake for direct use with Cortex Analyst, in addition to being used by your AI Assistant via MCP.

Related Resources