Snowflake launches AI agent 'Kortek Agent'. /Courtesy of Snowflake

Snowflake released its AI agent "Cortex Agents" on the 21st, which automates tasks by accessing data. This allows corporations to manage enterprise data more effectively without data specialists.

Cortex Agents is available in public preview and enhances the Cortex Analyst and Cortex Search features to automatically integrate and search data, and analyze complex queries to generate accurate answers. In this process, corporations can achieve accuracy, efficiency, and governance.

Cortex Analyst analyzes structured SQL data and has improved the accuracy of converting text to SQL using Anthropic's Claude 3.5 Sonnet. Users can query data in natural language and extract insights, performing complex reasoning, code generation, and multimodal data analysis with powerful enterprise-level AI capabilities. Cortex Search dramatically enhances the search accuracy of unstructured data such as text, audio, images, and video. It recorded at least 11% higher accuracy compared to OpenAI embedding models and has demonstrated excellent performance in various benchmark tests. Additionally, features for large-scale data indexing and customized vector embedding model selection have been added.

Accurate data analysis, maintaining security, and compliance with governance are essential for AI agents to scale effectively in corporate environments. Cortex Agents are designed to search various data sources, comply with security policies, and provide reliable results. Even after deployment, it continuously monitors and improves performance and operation, helping corporations safely scale AI agents while maintaining security and compliance.

For example, financial analysts often need to combine structured revenue data with unstructured financial reports and market data. Providing information to AI securely through end-to-end governance is necessary, and Cortex Agents can be utilized for this. Cortex Agents simplify the integration, search, and processing of the two data sources, enabling corporations to easily build high-quality agents on a large scale.

Christian Kleinerman, senior vice president of product management at Snowflake, noted, "AI strategies cannot exist without data strategies," adding that "many customers struggle to systematically manage data and build governance to maximize AI value. Snowflake supports customers in managing data more easily and achieving tangible results, even if they are not data specialists, through Cortex Agents."

※ This article has been translated by AI. Share your feedback here.