Getting Started with Snowflake Cortex

Create a Snowflake account to explore the Snowflake AI data cloud and cortex features with a 30-day free trial and $400 credit, no credit card required.

Take a quickstart tour of Snowflake Cortex in a private workspace, running a simple select using the Cortex complete function for text generation with an LLM model.

Explore medallion architecture, a three-zone data design in Snowflake—bronze raw data, silver cleaned data, and gold business-ready data for AI and reports.

Explore a bronze to silver to gold workflow in Snowflake Cortex, creating an ecommerce database with bronze.orders, cleaning and standardizing data, and aggregating into gold.orders by order date and country.

Load json data into Cortex UI, create bronze, silver, and gold layers, and ingest and transform across multiple schemas for ecommerce analytics.

Semantic views in Snowflake store meaning as a schema object, delivering authoritative definition and translation layer for business users, enabling AI and BI readiness with Kotec Analyst, metadata-only and shareable.

Create a semantic view named sv-movie-analytics by defining tables, relationships, facts, dimensions, matrix, and comments, enabling cortex analyst natural language queries, SQL, and BI tools access.

Explore semantic views in Snowflake Cortex by building a movie ratings database, creating normal and semantic views, and querying with natural language prompts to generate insights.

Learn to create semantic views from yaml files using Snowflake Cortex, comparing legacy models and the new implementation with RBAC, catalog support, and derived metrics.

Discover how to use Snowflake notebooks, with or without Cortex, to blend SQL, Python, and Markdown in a unified, fully managed interface for data exploration, analysis, visualization, and end-to-end workflows.

Prepare data for notebooks by creating loan_db with a lending schema and four tables—customer, loan, loan_application, and payment—then insert sample data for testing.

Use cortex to create sql and python cells for a loan portfolio summary. Fetch loan details with customer info, run quick portfolio statistics and breakdown by type.

Explore risk analysis with sql and python in cortex by merging loan, customer, and loan application data, highlighting credit score, debt-to-income, and troubled loans through charts.

Learn to use a common table expression to aggregate loan payments for delinquency and cohort analysis, classify risk buckets, and track origination cohorts with running totals.

Learn to use Streamlit in Snowflake to build web apps inside your Snowflake account and connect the Cortex Analyst API to a chat UI that shows SQL, results, and charts.

Learn to create your first streamlet app in snowflake cortex, name app 001, connect to snowflake, and use a slider to update a bar chart of high five by quarter.

Practice streamlit basics with interactive examples in Snowflake Cortex, including hello world, dynamic name display, a simple calculator, pandas data filtering with dropdowns, and slider-driven chart visualization.

Learn how Cortex generates a Python Streamlit app to fetch all customers from the loan db's customers table via Snowpark and display first name, last name, and email id.

Prepare a chatbot-driven flight bookings app by creating a phrase db, skybook_db, a schema, and tables for passengers, airlines, flights, and bookings, then insert sample data.

Create a streamlet app with a two-panel layout, left booking list and right cortex-powered chatbot, using snowflake cortex to view, add, update, and delete bookings in real time.

Explore the model context protocol (MCP) in Snowflake Cortex, create a new domain and tables, set up an MCP server, and leverage an MCP client to complete tasks.

Discover how the model context protocol (MCP) acts as a universal translator, enabling AI to securely connect to databases, APIs, file systems, and tools to interact with external resources.

Explore the MCP problem statement and how standardizing connections between AI applications and external tools eliminates fragmented integrations, reduces custom code, and strengthens security and consistency for robust AI apps.

Explore the MCP architecture as a client-server model with hosts, clients, and servers. See how hosts connect LLM apps, clients manage one-to-one connections, and servers provide context, tools, and prompts.

Create a simple MCP server in Snowflake Cortex by setting up a database, schema, and a products table, then configure cortex search and semantic views for querying.

Explore retrieval augmented generation (rag) by combining vector search with augmented generation, feeding retrieved documents to a language model to produce grounded, up-to-date answers as a source of truth.

Demonstrates building a simple rag app using snowflake cortex, indexing text chunks, and querying via a streamlit chat with hybrid semantic and keyword search for context-aware llm augmented generation.

Explore what an AI agent is in Snowflake Cortex: an autonomous, goal-driven system that perceives context, reasons and plans, acts via APIs and SQL, and reflects to iterate.

Snowflake Cortex agent deploys a managed AI agent inside Snowflake, orchestrating LLMs, SQL-powered structured data, and Cortex Search over unstructured data, with Cortex Analyst converting natural language to SQL.

Learn how to create a first hello world cortex agent in Snowflake Cortex, using a yaml spec, model orchestration, guardrails, and system prompts to run and test queries.

Create a Snowflake Cortex agent connected to real data with a product table and a Python-based tool, register it in an agent spec, and route queries by SQL.

Add Cortex search for unstructured docs by creating a document table of product FAQs, policies, setup guides, and warranty information, and enable the agent to query both data and documents.

Learn to build a Snowflake Cortex support ticket agent that creates tickets, assigns priority, sends email notifications, and looks up existing tickets via stored procedures.

Transcribe audio with Cortex AI transcribe, identify speakers, and extract timestamps; generate JSON transcripts and SQL you can reuse for call center analysis and sentiment insight.

learn to build a fully automated call center analytics pipeline inside snowflake cortex, uploading mp3s, transcribing with speaker timestamps, and classifying topics with llm while scoring agent performance.

Test Hindi transcription in the call center analytics app by running the cortex patch without code changes and verifying functionality.