Tsubasa Kanno

Posted on Jun 11

How Snowflake Copilot Inline Transformed My SQL Analysis Experience

#snowflake #sql #llm #analytics

Introduction

I recently attended Snowflake Summit 2025 in San Francisco for the first time! The Summit featured many notable updates including the Crunchy Data acquisition, the powerful Snowflake Openflow for flexible data ingestion, and significant AI platform advancements. I had meaningful discussions with many customers and partners throughout the event.

Among the developer experience improvements announced, one particularly exciting update is Snowflake Copilot Inline - a new feature in Snowflake Workspaces! This feature makes SQL analysis more intuitive and efficient, delivering what truly feels like having an AI pair programming buddy right in your workspace.

Unlike traditional Snowflake Copilot, which required switching to a dedicated chat interface, Copilot Inline lets you interact with an AI assistant directly within your SQL code. This feels like pair programming with a veteran analyst, where AI supports your code creation - a truly revolutionary feature for SQL analysis.

In this article, I'll share my hands-on experience with Snowflake Copilot Inline, covering its practical value, specific usage patterns, and optimization tips. If you're interested in boosting SQL analysis productivity, this is for you!

Note (2025/6/10): Snowflake Copilot Inline is currently in public preview, so features may undergo significant updates in the future.

Note: This article represents my personal views and not those of Snowflake.

What is Snowflake Copilot Inline?

Snowflake Copilot Inline is an AI assistant feature that integrates directly into your SQL code within Snowflake Workspaces. As an enhanced version of traditional Snowflake Copilot, it offers these key capabilities:

Key Features

Feature	Description
Inline Operations	Call AI assistant with `CMD+I` (Mac) or `Ctrl+I` (Windows) shortcut directly in SQL files
Context Understanding	Automatically grasps current file content and database/schema information
Real-time Suggestions	Provides instant SQL query improvements and fixes based on code context
Diff Display	Shows proposed changes in easy-to-read diff format
One-click Application	Apply suggestions instantly with the Accept button

Supported Regions

Snowflake Copilot Inline is natively supported in:

AWS US West 2 (Oregon)
AWS US East 1 (N. Virginia)

For other regions, cross-region inference functionality can be enabled by setting the CORTEX_ENABLED_CROSS_REGION parameter.

Understanding Snowflake Workspaces

Before diving into Copilot Inline, let's briefly explore Snowflake Workspaces - the foundation that makes this AI integration possible.

Snowflake Workspaces is an integrated development environment for creating, organizing, and managing code across multiple file types, accessible immediately upon logging into Snowsight. Unlike traditional Worksheets, it offers:

Workspaces vs. Traditional Worksheets

Feature	Traditional Worksheets	Workspaces
File Management	Individual worksheet management	File/folder-based hierarchical organization
Development Environment	Single query execution focus	IDE-like integrated development environment
Git Integration	Not supported	Git repository synchronization
Parallel Execution	One query at a time	Two simultaneous queries from same SQL file
Version Control	Basic history only	Full Git-based version control

Note (2025/6/10): Snowflake Workspaces is currently in public preview, so features may undergo significant updates in the future.

Access Control and Permissions

To use Snowflake Copilot Inline, you need these permissions:

Required Permissions

COPILOT_USER database role
- Role within SNOWFLAKE database
- Granted to PUBLIC role by default
CORTEX_USER database role
- Role within SNOWFLAKE database
- Required for Snowflake AISQL functions
- Granted to PUBLIC role by default
claude-3.5-sonnet model access
- Must be included in model allowlist
- No RBAC restrictions applied

Permission Verification

Most users can access the feature with default settings, but if you encounter issues:

-- Check current permissions
SHOW GRANTS TO ROLE PUBLIC;

-- Grant permissions if needed (requires ACCOUNTADMIN)
USE ROLE ACCOUNTADMIN;
GRANT DATABASE ROLE SNOWFLAKE.COPILOT_USER TO ROLE PUBLIC;
GRANT DATABASE ROLE SNOWFLAKE.CORTEX_USER TO ROLE PUBLIC;

Supported Use Cases

Snowflake Copilot Inline excels in these scenarios:

1. Data Exploration

Query AI about unfamiliar datasets to understand structure and characteristics:

"What tables are in this schema?"
"Tell me about the structure of the customer table"

2. SQL Query Generation

Convert natural language requests into SQL code:

"Get monthly sales trends for the past year"
"Show customer purchase frequency ranked by activity"
"Create sales analysis query excluding outliers"

3. Query Optimization

Get improvement suggestions for existing queries:

"Optimize this query for better performance"
"Rewrite this query for better readability"
"Suggest efficiency improvements for this SQL"

4. Syntax Error Fixes

Quickly resolve SQL syntax issues:

"Fix the syntax error in this query"
"Why isn't this query working?"
"Correct the JOIN syntax"

How to Use Copilot Inline

Operating Snowflake Copilot Inline is refreshingly simple:

Basic Workflow

Open Workspaces
- Navigate to Workspaces in Snowsight
- Create new file or open existing SQL file

Call Copilot Inline
- Position cursor anywhere in your code
- Press CMD+I (Mac) or Ctrl+I (Windows)

Submit Request
- Enter natural language request in message dialog
- Click send icon to submit to AI

Review Suggestions
- View AI-generated changes in diff format
- Red: deleted lines, Green: added lines
Apply Changes
- Accept: Apply suggestions to code
- Reject: Decline suggestions
- Close: End session

Execute Suggested Query
- Run the suggested query to see results

Limitations and Considerations

Snowflake Copilot Inline has these current limitations:

Technical Limitations

Limitation	Details
Data Access	Cannot access table data values (metadata only)
Response Time	Complex requests may require several seconds
New Object Recognition	3-4 hours needed to recognize newly created tables/schemas
Search Scope	Considers only top 10 most relevant tables and 10 columns each

Language Support

Supported languages include:

Supported Languages

Business Value and ROI

Snowflake Copilot Inline delivers significant business impact:

Analysis Efficiency Gains

SQL creation time reduction: Significant time savings in analysis and development through natural language generation
Debugging time reduction: AI-powered syntax error detection and fixes
Learning cost reduction: High-quality query creation even for junior developers

Enhanced Data Utilization

Accelerated exploratory analysis: Faster understanding of new datasets
Best practice adoption: AI suggests optimized query patterns
Reduced skill dependency: Less reliance on individual SQL expertise

Cost Reduction Benefits

Optimized development resources: Freedom from routine query creation tasks
Reduced training costs: Shorter onboarding and training time
Quality improvement: Reduced errors and rework through AI assistance

Conclusion

Snowflake Copilot Inline is a game-changing feature that fundamentally transforms SQL analysis experience. It delivers exceptional value through:

Enhanced developer experience: Pair programming-like experience with AI assistance
Accelerated learning: Natural acquisition of best practices through AI suggestions
Massive productivity gains: Focus on creative analysis instead of routine coding tasks

Personally, I often find myself thinking "How do I write this query again?" and this feature has been invaluable for quickly fixing syntax without switching screens.

I highly recommend this feature to everyone involved in SQL analysis and development. Give it a try!