Introduction to Snowflake CoCo

1. Introduction to Snowflake CoCo

Here's something that should bother you. Data professionals spend the great majority of their time just preparing data, not analyzing it, not building insight, cleaning it, reshaping it, debugging it, and doing it all over again when the source changes upstream. It's not just the volume of work, it's the context problem. Your pipeline touches dozens of tables with names that made sense to someone five years ago. Your SQL has to survive null values you didn't know existed until 2 AM on a Tuesday. Your stakeholder wants an answer in natural language, but your answer lives buried in a schema diagram and three conflicting transformation layers. AI is changing this, but great models alone won't get you there. Faster code generation doesn't necessarily generate high-quality data products and reliable pipelines. Generic coding agents may accelerate input, but may compound the real costs of maintenance, debugging, and governance on data engineering work. You end up with more code to manage, not less work to do. That's why CoCo operates directly within your local environment to help you build end-to-end solutions. CoCo brings context-aware assistance and purpose-built skills for Snowflake data engineering features and operates within your security perimeter. It provides access to the latest frontier and open weight models. Over the next four modules, you'll learn how to use Snowflake CoCo Desktop, an AI coding agent built natively for the data ecosystem to automate the work that's eating your week. You'll set it up, connect it to Snowflake, use and build skills, package them into plugins your whole team can share and wire it into your DevOps pipeline. By the end, you won't just be faster, you'll be working differently. Let's get started. In this module, we will start with an introduction to Snowflake CoCo and explain how it differs from generic coding agents. Next, we will talk about how to set up and configure your environment, including describing the AGENTS.md, CoCo's configuration file. Finally, after this introduction, you will walk through a hands-on exercise where you set up a Snowflake trial account and install CoCo Desktop, connect it to your trial account, and do an easy lab. By the way, this course was created by Brett Svenkeson, technical curriculum developer at Snowflake, with contributions from Jeremiah Hansen, Jenna Donlon, and Francisco Kattan, also from Snowflake, and is narrated by an automated voice. Snowflake CoCo is an AI-powered coding agent built natively into Snowflake. It leverages large language models to help data professionals write, understand, and debug SQL and Python code directly within their Snowflake environment. CoCo accelerates query authoring, generates code from natural language descriptions, explains existing logic, suggests optimizations, and automates documentation all without leaving the Snowflake platform. CoCo also handles more varied tasks, such as building data pipelines, training ML models, and deploying applications. Designed for analysts, data scientists, and data engineers, CoCo empowers you to accelerate your workflows and eliminate manual bottlenecks in Snowflake. CoCo uses an agentic workflow. It doesn't just respond to a single prompt, it can plan and execute multi-step tasks. That's a big deal for anything more complex than a one-liner query. CoCo harnesses your Snowflake connection, SQL and other code execution, and specialized skills into a single agent runtime, so the AI can plan, write, execute, and verify data pipelines without you switching tools. It holds the context and brokers every tool call so multi-step tasks complete end-to-end in one session. CoCo has three different form factors or flavors, CoCo in Snowsight, CoCo CLI, and CoCo Desktop. Therefore, you have different ways to work, each with very similar capabilities. We will compare the three soon, and in this course, we will focus on the desktop version. Compared to other coding agents, CoCo offers the following advantages for Snowflake data engineers. CoCo keeps your data within Snowflake security perimeter, governs your agent's access and work, integrates natively with Snowflake where it can understand your schemas, metadata, and data lineage. It ships with a large library of built-in Snowflake skills, can orchestrate the work of your agents, and gives you a choice of frontier or open-weight models. CoCo in Snowsight is the browser-based version of CoCo, accessible directly within the Snowsight web interface. It provides an integrated AI chat panel alongside your workspace, allowing you to ask questions, generate SQL, explain query logic, and troubleshoot errors without switching tools. Because it runs entirely in the browser, no installation or configuration is required. This makes it the most accessible entry point for users new to CoCo. It is ideal for analysts and data professionals who primarily work within the Snowsight UI and want AI-powered assistants seamlessly embedded in their everyday workflow. CoCo CLI is a command-line interface version of Snowflake CoCo designed for developers and engineers who prefer terminal-based workflows. It provides the same AI-powered code generation, SQL authoring, and debugging capabilities as CoCo in Snowsight, but is accessible directly from a terminal and can be integrated into automated scripts and CI/CD pipelines. The advantage over CoCo in Snowsight is flexibility. CLI users can incorporate CoCo into their local development environments, use it alongside code editors like VS Code, and automate repetitive tasks as part of larger engineering workflows, all without opening a browser. CoCo Desktop is a native desktop application that brings CoCo to your local machine and is a coding IDE workspace. It combines the convenience of a standalone app with full access to CoCo's AI capabilities, including SQL generation, code explanation, and debugging support. CoCo Desktop is ideal for users who prefer writing and interacting with code in a desktop IDE and need seamless integration with their local files and development environment. As a fork of the popular VS Code IDE, users can quickly transition to it and can easily write code with CoCo as their coding agent. Let's talk briefly about the installation and configuration of CoCo Desktop. This course contains hands-on labs that you will complete using a free Snowflake trial account. You will need to set up a Snowflake trial account, install CoCo Desktop, and connect it to the account. Before we have you do this, I will talk about CoCo's configuration file, AGENTS.md. After the end of this module video, I will give you the links needed and instructions. As we integrate AI deeper into our workflows, a new standard has emerged, the AGENTS.md file. You are probably familiar with a standard README file, which provides quick starts and project descriptions for human developers. But what about the AI? AGENTS.md is exactly that, a dedicated, predictable place to provide the specific context and instructions that AI coding agents need to successfully work on your project without cluttering the documentation meant for your project. AGENTS.md is the same documentation meant for humans. Because it is written in standard markdown, there are no proprietary formats to learn, and there are no required sections. I listed some popular ones like project overview, testing instructions, et cetera. You simply add an AGENTS.md file to the root of your repository and list out what matters, setup commands, code style preferences, and testing instructions. Plus, you can have multiple files. You can use the nearest file to the code being edited. However, explicit user prompts always override AGENTS.md. By providing this precise, agent-focused guidance, you ensure that CoCo, your AI coding agent, no matter which model you choose, instantly understands the rules of your project and writes code exactly the way you want it. We've now seen what CoCo is and what makes it different from general coding agents. Next, it's time to put it into practice. In the first lab, you'll create your Snowflake trial account, install CoCo Desktop, connect it to Snowflake, and confirm that your environment is ready for the rest of the course.

2. Let's practice!

Create Your Free Account

By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.