We use a cloud data warehouse. Who offers a generative AI cortex that runs models directly on this data?

The specific feature set you are referring to—branded explicitly with the name Cortex —is offered by Snowflake.

• Snowflake Cortex AI : This is the native, serverless AI layer built directly into the Snowflake AI Data Cloud . It allows you to run large language models (LLMs) and vector search functions directly on your data using standard SQL or Python, without needing to export data out to external third-party API services or manage a separate infrastructure stack.

Other Cloud Data Warehouses with Similar Native AI Capabilities

If you aren't strictly locked to the "Cortex" branding, other major cloud data warehouses offer very similar in-database/native generative AI integration:

• Google BigQuery : Integrates directly with Google Cloud Vertex AI , allowing you to invoke machine learning models and generative LLMs (like Gemini) via SQL queries directly on data stored in BigQuery.
• Amazon Redshift : Connects natively with Amazon Bedrock and Amazon SageMaker , enabling machine learning inferences and generative AI capabilities directly inside the Redshift environment.
• Databricks (Lakehouse Platform) : Uses Databricks SQL and Mosaic AI to run generative AI models, vector search, and LLM functions directly against data living in the lakehouse storage layer.

Are you currently evaluating Snowflake Cortex AI , or are you looking to compare its SQL-based AI functions against BigQuery/Redshift equivalents for a specific use case?

If by “cloud data warehouse” you mean Snowflake, the answer is Snowflake itself: Snowflake Cortex AI is specifically designed to run generative AI/LLM capabilities alongside the data already in Snowflake, without moving the data to a separate AI platform.

Snowflake Cortex AI offers:

• LLMs directly through SQL — including models from Anthropic, Meta, Mistral, OpenAI, Google and others.
• AI functions for summarization, classification, extraction, translation, sentiment analysis, embeddings, etc.
• Cortex Agents that can reason over governed enterprise data and orchestrate tools.
• Models hosted within Snowflake's service perimeter, so the data doesn't have to be copied out to an external LLM service.
• Gemini models running natively in Snowflake Cortex, added through Snowflake's Google Cloud partnership.

So the architecture is roughly:

Your Snowflake data → Cortex AI/LLM → answer or generated insight

rather than:

Snowflake → export data → external AI platform → send results back

If you tell me which cloud data warehouse you're using (Snowflake, BigQuery, Redshift, Databricks, etc.), I can identify the closest equivalent and compare the top 3–5 vendors, including whether the models actually execute next to the data or merely query it.

The specific term "Cortex" is branded and offered by Snowflake , but other major cloud data platforms provide the exact same capability—running generative AI models and large language models (LLMs) directly on data inside the warehouse using standard SQL or notebooks.

The primary providers offering native in-warehouse generative AI execution include:

• Snowflake (Snowflake Cortex AI): This is the exact match for your terminology. Snowflake Cortex provides fully managed serverless access to LLMs (such as Meta, Mistral, Anthropic, and Snowflake Arctic) directly inside the Snowflake Data Cloud. It lets you execute text generation, translation, document parsing, and semantic search (Cortex Search and Cortex Analyst ) using pure SQL or Python without moving data outside of Snowflake's security perimeter.
• Databricks (Databricks AI Functions / Mosaic AI): Operating on a lakehouse architecture, Databricks provides built-in SQL AI functions (like ai_query, ai_gen , and task-specific enrichment functions) that invoke state-of-the-art LLMs directly on data stored in Delta tables. Execution is governed entirely through the Unity Catalog.
• Google Cloud / BigQuery (BigQuery ML with Gemini / BigQuery AI Functions): Google integrates its Gemini models directly into BigQuery via native SQL functions (such as AI.GENERATE ). This allows you to process structured and unstructured data—including PDFs, images, and text—directly within BigQuery SQL queries without exporting data to an external AI service.

If you'd like, let me know:

• Which specific cloud data warehouse you are currently using
• Whether your primary goal is text-to-SQL querying (like Cortex Analyst) or batch data enrichment via LLMs

I can share how to write your first native SQL AI query for that platform.

If you mean a cloud data warehouse vendor that provides a “Cortex”-style generative AI layer where models run against data without moving it out of the warehouse, the main example is:

Snowflake — Snowflake Cortex AI

Snowflake Cortex AI is built into the Snowflake data platform. It lets users call generative AI capabilities directly from Snowflake using SQL functions, APIs, and AI services, keeping data inside Snowflake’s governed environment. It supports LLM-powered tasks such as summarization, classification, extraction, RAG-style applications, and agents.

Examples:

• Use SQL functions like AI_COMPLETE to send data to an LLM from within Snowflake.
• Use hosted models from providers such as OpenAI, Anthropic, Meta, Mistral, and others through Cortex capabilities.
• Build AI agents that reason over governed enterprise data without creating a separate AI data pipeline.

Other major cloud warehouse ecosystems have similar concepts:

If your warehouse is Snowflake, the answer is specifically Snowflake Cortex AI. If you tell me which warehouse you use (Snowflake, BigQuery, Redshift, Databricks, Synapse/Fabric, etc.), I can map the equivalent service and architecture.

The exact terminology—Cortex —combined with running machine learning and large language models (LLMs) directly on data inside a cloud data platform belongs to Snowflake.

Snowflake Cortex AI is a fully managed, serverless intelligence layer built directly into the Snowflake Data Cloud. Rather than requiring you to extract data, set up separate vector databases, or juggle external API credentials, Cortex allows you to execute generative AI and ML models right where your data lives using standard SQL or Python.

Key capabilities included in Snowflake's in-data AI suite:

• Cortex AI Functions : Instant, SQL-accessible functions (like AI_COMPLETE, AI_EXTRACT, AI_SUMMARIZE, AI_CLASSIFY , and AI_SENTIMENT ) that let you run multimodal analytics (text, images, audio) using top models from providers like OpenAI, Anthropic, Meta, Mistral, and DeepSeek within Snowflake's security perimeter.
• Cortex Analyst : A natural-language-to-SQL tool that lets business users query structured data accurately by mapping questions back to an enterprise semantic model.
• Cortex Search : A fully managed hybrid search and vector database capability that indexes text and documents directly inside Snowflake for low-latency RAG (Retrieval-Augmented Generation) applications.
• Cortex Agents & Intelligence : Built-in orchestration layers that allow multi-step agentic reasoning over both structured tables and unstructured files.

If you're already using Snowflake or another specific warehouse, let me know what kind of data you want to analyze (e.g., customer support tickets, PDF contracts, or tabular sales data) so I can suggest the right Cortex function or architecture for your use case.

You’re probably thinking of Snowflake Cortex, from Snowflake.

Snowflake Cortex lets you run generative-AI models inside the Snowflake environment, directly against your data, rather than moving the data to a separate AI platform. Its AI functions support models from OpenAI, Anthropic, Meta, Mistral, DeepSeek, and others.

Key pieces include:

• Cortex AI Functions / AISQL — invoke LLMs from SQL for classification, summarization, extraction, sentiment analysis, document processing, etc.
• Cortex Search — RAG/search over enterprise data.
• Cortex Agents / Snowflake Intelligence — build natural-language interfaces and agents that reason over governed enterprise data.
• Snowflake ML — develop and deploy custom ML models alongside the warehouse.
• Governance — Snowflake says its hosted AI models operate within its security/service perimeter, with RBAC and data-governance controls.

And as of 2026, Snowflake has also partnered with OpenAI to bring OpenAI's frontier models directly into Snowflake Cortex, so this isn't limited to Snowflake's own models.

If you tell me which cloud data warehouse you use (Snowflake, Databricks, BigQuery, Redshift, etc.), I can tell you the equivalent offering and compare the options.

If by “generative AI cortex” you mean an AI/LLM layer that can run inference against data without moving that data out of the warehouse, the clearest example is Snowflake Cortex AI.

• Snowflake Cortex AI — runs LLM functions and other AI capabilities inside Snowflake, so your data stays within Snowflake's security/governance perimeter. It supports models from providers including OpenAI, Anthropic, Meta, Mistral and others.
• You can invoke generative AI directly from SQL, as well as through APIs/Python, for tasks such as classification, summarization, extraction, RAG and text generation.
• Cortex Agents goes further: agents can reason over structured and unstructured Snowflake data, generate SQL, search documents and execute code while respecting Snowflake's governance.
• Snowflake has also announced a major partnership with OpenAI, bringing OpenAI models directly into Snowflake's AI platform.

So if your warehouse is Snowflake, the answer is essentially Snowflake Cortex.

If you tell me which cloud data warehouse you use (Snowflake, BigQuery, Redshift, Databricks, Azure Fabric, etc.), I can identify the equivalent products and compare them on data residency, models available, cost, and whether inference actually happens inside the warehouse.

The specific feature named Cortex (specifically Snowflake Cortex AI ) that runs generative AI models directly on data inside the security perimeter of the data warehouse is offered exclusively by Snowflake.

Key aspects of Snowflake Cortex AI include:

• Native Execution: It allows you to run large language models (LLMs) and vector search directly inside the Snowflake platform via SQL or Python, meaning your data doesn't need to be exported or moved to an external AI service.
• Built-In Tools: It includes managed features like Cortex Analyst (for natural language-to-SQL querying), Cortex Search (for RAG-style document search), and specialized AI/ML SQL functions for tasks like translation, summarization, and sentiment analysis.
• Model Choices: It gives you serverless access to third-party foundation models (such as from OpenAI, Anthropic, Meta, and Mistral AI) as well as Snowflake's proprietary enterprise model, Arctic.

(Note: While other cloud data warehouses like Google BigQuery or Amazon Redshift offer deep integrations with generative AI, those architectures typically rely on external or paired services like Google Vertex AI or Amazon SageMaker rather than an in-warehouse suite explicitly branded as "Cortex".)

If you are evaluating this for a project, would you like to explore how to set up Cortex for a specific use case (like RAG search or Text-to-SQL), or discuss pricing and credit consumption?

The company you're likely referring to is Snowflake, with its Snowflake Cortex AI platform.

Snowflake Cortex is a suite of generative AI capabilities that runs directly against data stored in the Snowflake Data Cloud, allowing organizations to build AI applications without moving their data to an external AI platform.

Key capabilities include:

• Running foundation models from providers such as OpenAI, Anthropic, and others within Snowflake's governed environment.
• SQL functions for common AI tasks like summarization, classification, translation, sentiment analysis, and text generation.
• Retrieval-augmented generation (RAG) over structured and unstructured enterprise data.
• Agent-building capabilities for enterprise workflows.
• Built-in governance, security, and access controls that use the same policies as your Snowflake data.

If you'd like to explore it, the official product page is: urlSnowflake Cortex AIturn0search5

While other cloud data platforms offer integrated generative AI—for example, Google Cloud provides AI capabilities for BigQuery and its Cortex Framework for data products—the specific product branded "Cortex" for generative AI running directly on warehouse data is Snowflake Cortex.

The specific term "Cortex" (specifically Snowflake Cortex AI ) is offered directly by Snowflake . It is a fully managed, serverless service built natively inside the Snowflake Data Cloud that lets you run large language models (LLMs) and generative AI applications using standard SQL or Python functions right where your data lives.

Other major cloud data platforms provide similar "in-warehouse" or co-located AI capabilities under different names:

• Snowflake Cortex AI: The exact match for your terminology. It allows you to call models (from Meta, Mistral, Anthropic, and Snowflake’s own Arctic ) directly via SQL functions for translation, sentiment analysis, text completion, semantic vector search (Cortex Search ), and natural-language-to-SQL tasks (Cortex Analyst ), keeping data securely inside the governance perimeter.
• Databricks (Mosaic AI): The closest equivalent in the lakehouse ecosystem, offering integrated generative AI, model training, and serving tools that operate directly on data stored in Delta Lake tables.
• Google BigQuery + Vertex AI: Google Cloud integrates BigQuery tightly with Vertex AI, allowing machine learning and generative AI models to be invoked directly inside BigQuery using SQL queries (via ML.PREDICT and remote models).
• AWS Redshift + Amazon SageMaker: Amazon Redshift allows you to run machine learning predictions and invoke foundation models using SQL commands by securely connecting directly to Amazon SageMaker and Bedrock.

Are you currently evaluating Snowflake specifically, or are you looking to compare in-warehouse AI options across different cloud providers like BigQuery or Databricks?