Automatic Data Agents
AI agent · Data & analytics agent · Data & analytics by Snowflake
Transform static data listings into intelligent conversational experiences.
Automatic Data Agents instantly transform Snowflake data listings into conversational AI experiences for businesses. It allows providers to generate AI agents and semantic views in minutes, enabling non-technical users to query shared data using natural language.
Agent specs
- What it does
- Automatic Data Agents automatically create AI-powered orchestration layers and semantic data representations for Snowflake listings and shares, enabling natural language querying of shared datasets.
- For
- End users who need to query shared data using natural language without technical expertise.
- Actions
- Analyzes listing metadata and data schemas, Automatically constructs semantic views, Constructs Cortex Agents
- Integrations
- Cortex Analyst, Cortex Agent, Provider Studio, Internal Marketplace
- Channels
- Snowflake Marketplace listings, Internal Marketplace listings, Direct shares
- Availability
- Open preview available to all Snowflake accounts.
Category: AI Agents
- Channels
- Snowflake Marketplace, Internal Marketplace, Snowsight UI
- Integrations
- Cortex Analyst, Cortex AI, Provider Studio, Internal Sharing
- Models
- Cortex Agent, Semantic View Autopilot
- Deployment
- Snowflake AI Data Cloud (Cloud)
- Setup
- Up to 10 minutes for generation
- Generation Time
- Up to 10 minutes
- Instantly generates AI-powered agents for data listings and shares
- Automates semantic modeling using Semantic View Autopilot
- Enables natural language querying without SQL expertise
- Generation process completes in approximately 10 minutes
- Integrated testing allows validation before publishing to consumers
Overview
Automatic Data Agents is an AI-powered feature within the Snowflake AI Data Cloud designed for data providers and businesses. It solves the barrier of technical complexity by instantly transforming static data listings and shares into conversational experiences that end users can query using natural language. The system works by analyzing listing metadata and data schemas to automatically construct two critical objects: a business-friendly semantic view and a Cortex Agent orchestration layer. This automation significantly reduces manual engineering time, typically completing the setup in approximately 10 minutes. Once published, these agents allow consumers to explore data without writing SQL, making shared data products accessible to non-technical audiences such as executives and marketers.
Automated Setup and Modeling
- Automated Semantic Modeling: The system analyzes listing metadata and data schemas to automatically construct business-friendly data representations.
- Domain-Aware Orchestration: The agent understands the specific domain and context of the data to provide intelligent assistance.
- Table and View Selection: Providers can choose specific tables and views to include when generating the semantic view to control exposed data.
Consumer Experience
- Natural Language Querying: End users can query shared data using natural language without technical expertise or SQL knowledge.
- Conversational Data Experience: The agent transforms static data into intelligent, conversational experiences for consumers.
- Marketplace Discovery: Consumers can find AI-ready listings more easily when providers add the "Cortex AI ready" category.
Governance and Control
- Integrated Testing: Providers can validate the agent's responses within Snowflake before publishing them to consumers.
- Seamless Publishing: Generated assets are attached directly to existing secure shares, making them instantly available.
- Centralized Management: Providers can configure and manage agents directly within Provider Studio or Internal Sharing pages.
What it does
- Analyzes listing metadata and data schemas
- Constructs semantic views compatible with Cortex Analyst
- Generates Cortex Agent orchestration layers
- Enables natural language querying of shared data
- Attaches AI objects to data listings and shares
Use cases
- Data providers use Automatic Data Agents to make their Marketplace listings "Cortex AI-ready" for non-technical buyers.
- Business teams query shared organizational data using natural language to get insights without writing SQL.
- Data engineers automate the creation of semantic views and orchestration layers for internal data shares.
Specifications
- Channels
- Snowflake Marketplace, Internal Marketplace, Snowsight UI
- Integrations
- Cortex Analyst, Cortex AI, Provider Studio, Internal Sharing
- Data sources
- Snowflake tables, Snowflake views, Data listings, Secure shares
- Human control
- Providers manually select tables and views for inclusion; providers test and validate agent responses before publishing; providers manually attach agents to listings
- Models
- Cortex Agent, Semantic View Autopilot
- Deployment
- Snowflake AI Data Cloud (Cloud)
- Setup
- Up to 10 minutes for generation
- Generation Time
- Up to 10 minutes
- Primary Components
- Cortex Agent, Semantic View
- Compatibility
- Cortex Analyst
- Management Interface
- Provider Studio, Internal Sharing, External Sharing page
- Required Privilege
- CORTEX_USER
- Listing Support
- Snowflake Marketplace, Internal Marketplace, Direct Shares
- Modeling Tool
- Semantic View Autopilot
How it works
- Start Generation: The provider initiates the process within Provider Studio or Internal Sharing, where the system analyzes metadata and schemas.
- Automated Modeling: The system automatically identifies table relationships, metrics, and dimensions to build a semantic view.
- Test and Validate: The provider uses the integrated testing environment to verify the agent's responses and accuracy.
- Publish to Consumers: The generated agent and semantic view are attached to the listing or share, making them available to consumers.
FAQ
What objects are created during the generation process?
Automatic Data Agents analyze your listing metadata and data schemas to automatically construct a semantic view (a business-friendly representation) and a Cortex Agent (an AI orchestration layer).
What privileges are required to use this feature?
To create and manage these agents, you need the CORTEX_USER privilege, along with CREATE SEMANTIC VIEW and CREATE AGENT privileges in the target schema.
Are there any limitations on which shares can use Automatic Data Agents?
You cannot use this feature if the share already contains manually created agents, semantic views, or Cortex Search Services.
Added by AntSpace from public sources (docs.snowflake.com).