Telecommunications companies face unique challenges in managing vast amounts of operational data—from network performance metrics and infrastructure capacity to customer call recordings, support tickets, and satisfaction surveys. To address these challenges effectively, a unified platform is essential—one capable of storing and processing all data types, whether MP3 call recordings, PDF help documents, or structured network telemetry.

Throughout this quickstart, we'll use NovaConnect—a fictitious telecommunications company—as our example organization. NovaConnect operates a nationwide 5G network serving millions of customers, and needs AI-powered tools to analyze network performance, understand customer feedback, and streamline support operations.

This data must be readily accessible for analysis using the latest large language models (LLMs) such as Anthropic, Gemma, LLaMA, or DeepSeek. Ensuring the trustworthiness and security of generated insights is critical, especially when they inform network operations decisions and customer experience strategies.

In addition, developing Agentic AI capabilities allows for natural language question-answering tailored to network engineers, customer service managers, and operations teams who need real-time insights into network health and customer satisfaction.

In this hands-on lab, you'll learn how to build a Telco Operations AI Agent from the ground up using the Snowflake AI Data Cloud.

Learn how you can leverage the latest AI technologies right within the Snowflake platform. When AI is deeply embedded in your trusted data platform, the possibilities are endless. We will be exploring the processing of both Unstructured and Structured data which will then allow the application of a Cortex Agent to help discover insights by leveraging All Data.

This hands-on lab utilizes a comprehensive telecommunications dataset spanning multiple data types to demonstrate real-world AI applications. You'll work with:

Network Operations Data

Customer Experience Data

AI-Powered Analysis Journey

Your AI assistant will synthesize insights from all these sources to answer questions like:

This diverse dataset enables you to experience how modern AI can unify structured network data with unstructured customer interactions, creating comprehensive operational intelligence.

In this quickstart, you'll build a comprehensive AI-powered telecommunications operations platform called NovaConnect Intelligence using Snowflake's Cortex AI capabilities. This end-to-end solution demonstrates how to:

This quickstart offers two deployment paths to accommodate different preferences and skill levels. Choose the option that best fits your workflow:

Best for: Most users, visual learners, those new to Snowflake

Deploy directly from GitHub using Snowflake's built-in Git integration. This approach:

You'll connect Snowflake to GitHub, then run a series of SQL scripts that automatically:

1. Configure your account and create required objects
2. Load sample data from stages
3. Deploy Cortex Search services and Analyst semantic models
4. Create notebooks for data processing

Best for: Developers, automation enthusiasts, CLI power users

Use the Telco Agent Builder skill for guided, conversational deployment. This approach:

By the end of this lab, you'll have deployed:

This quickstart deploys a complete multi-modal AI platform combining:

Data Sources → AI Processing → Structured Data → AI Services → Applications

Data Types:

AI Capabilities:

Latest AISQL Syntax: All examples use 2025 AI_* functions

The NovaConnect AI platform follows a layered architecture that transforms raw unstructured data into actionable intelligence. Data flows from multiple sources (documents, audio, structured tables) through Cortex AI processing functions, into organized tables and search indices, and finally surfaces through Snowflake Intelligence agents and applications.

The architecture diagram below illustrates how each component connects:

The dataflow diagram below shows how data moves through the system to power Snowflake Intelligence. Raw data enters from the left (CSV files, PDF documents, MP3 audio recordings), gets processed through Cortex AI functions in the middle (AI_PARSE_DOCUMENT, AI_TRANSCRIBE, AI_SENTIMENT), and flows into the structured outputs on the right (Cortex Search services and Cortex Analyst semantic models). These services become the "tools" that the Snowflake Intelligence agent uses to answer user questions—combining structured SQL queries with semantic search across unstructured content.

Option A - Free Trial (Recommended):

1. Visit https://signup.snowflake.com/
2. Sign up for a free 30-day trial
3. Choose Enterprise edition
4. Select a cloud region (AWS, Azure, or GCP)
5. Verify your email
6. Log in to Snowsight (https://app.snowflake.com)

Option B - Existing Account:

Deploy directly from GitHub - No downloads or CLI tools needed!

1. In Snowsight, click on Projects
2. Select Workspaces
3. Add new SQL file
4. Copy and paste this script:

4. Click Run (or press Cmd/Ctrl + Enter)
5. Wait for completion (~30 seconds)
6. ✅ Git integration complete! You're now connected to GitHub

Now will create a workspace from the github repository

1. Create a Workspace from Github repository by clicking on My Workspace, then select the option From Git repository

2. When Prompted, use the following URL in the Repository URL field:

3. Press Create

4. Navigate to the assets folder (at the root of the repository):

You should see the following file structure:

5. Navigate to sql/ - This is where the deployment scripts are
6. You'll see SQL files numbered 01-05

✅ You're now ready to deploy!

Snowflake's Git Integration feature allows you to connect directly to GitHub repositories and execute SQL scripts without downloading files locally. This is powerful because:

When you created the Git Repository object in Step 2, Snowflake established a connection to the Telco AI GitHub repository. This connection allows you to:

1. Browse files in the repository through the Snowsight UI
2. Create Workspaces from repository folders for interactive development
3. Fetch updates when the repository changes

The deployment scripts in assets/sql/ are numbered 01-05 and should be executed in order. Each script builds on the previous one, creating the complete NovaConnect AI platform.

1. Navigate: Projects → Git Repositories → TELCO_AI_LAB.GIT_REPOS.TELCO_AI_REPO
2. Browse to: assets/sql/
3. Right-click each file (01-05) → "Open in new worksheet"
4. Execute each script in order

What gets deployed:

1. ✅ Database TELCO_OPERATIONS_AI with schemas
2. ✅ Role TELCO_ANALYST_ROLE with CORTEX_USER privileges
3. ✅ 14+ tables with ~10,000 rows of data
4. ✅ 2 Cortex Search Services
5. ✅ 3 Cortex Analyst Semantic Models
6. ✅ 1 Snowflake Intelligence Agent
7. ✅ 3 Notebooks
8. ✅ Stages with MP3 audio files and PDFs

Deployment time: 15-20 minutes

New! You can also deploy this quickstart using the Cortex Code CLI - Snowflake's AI-powered command-line assistant.

If you have Cortex Code CLI installed, you can use the built-in skills for guided, automated deployment. Skills are structured markdown instructions that guide Cortex Code through complex deployment procedures.

1. Clone the Repository

2. Configure Your Snowflake Connection (if not already configured)

3. Register the Project Skills

The repository includes custom skills in .cortex/skills/, but they need to be registered before use. Run this command once to add the skills directory:

You should see output like:

4. Launch Cortex Code

Start Cortex Code from the project directory:

5. Deploy Using the Skill

Once Cortex Code is running, simply ask it to deploy. Use any of these prompts:

Cortex Code will load the telco-agent-builder skill and guide you through each step interactively.

6. (Optional) Customize Regions Before Deployment

The skill will ask if you want to customize the demo regions. By default, the data uses Malaysian regions (Kuala Lumpur, Selangor, Penang, etc.).

If you prefer different regions, the skill will run a Python script to generate customized CSV files:

This generates new CSV files in assets/data/regional/ which the skill will upload instead of the defaults.

The repository includes two skills in .cortex/skills/:

The telco-agent-builder skill guides you through these steps:

1. ✅ Verify prerequisites - Check Snowflake connection and permissions
2. ✅ Customize regions (optional) - Generate regional data if requested
3. ✅ Configure account - Create role, warehouse, database, schemas, and stages
4. ✅ Upload data - Copy CSV, PDF, and MP3 files to Snowflake stages
5. ✅ Load tables - Create and populate 14+ tables with telco data
6. ✅ Deploy Cortex Search - Create 2 semantic search services
7. ✅ Deploy Cortex Analyst - Upload semantic models and create the Intelligence Agent
8. ✅ Deploy Notebooks - Create 3 Snowflake Notebooks
9. ✅ Verify deployment - Run checks to confirm everything is working

To remove all deployed resources, use the uninstall skill:

This will drop the database, warehouse, role, and all related objects.

Skills not found? If Cortex Code doesn't recognize the skills, verify they're registered:

"No valid skills found" error? Skills require a specific file structure:

To remove a registered skill directory:

After deployment completes, verify in Snowflake UI:

All set? ✅ Continue to the next section!

After Git integration deployment completes, let's verify everything was created successfully.

Open Snowflake UI and navigate to Data → Databases → TELCO_OPERATIONS_AI

You should see:

You should see:

Notebooks:

All verified? ✅ Let's start using the AI features!

Duration: 45

Congratulations! You've successfully deployed the NovaConnect Telco Operations AI platform. The installation phase is complete.

What you've deployed:

What's next: In the following sections, you'll explore and interact with the AI capabilities you just deployed:

Before diving into the notebooks, let's explore the AI & ML Studio - your one-stop shop for trying out AI functions using a user-friendly UI.

Navigate to the AI & ML section in the Snowflake navigation bar.

The AI and ML Studio provides access to:

The Cortex LLM Playground lets you compare text completions across multiple large language models available in Cortex AI.

Before processing unstructured documents at scale, it's valuable to understand how LLMs interpret and extract information from text. The Cortex Playground provides a sandbox to:

This foundational understanding will help you design better document processing workflows when you move to the Notebooks section.

1. Click on Cortex Playground in the AI & ML Studio
2. Select a model (e.g., claude-4-sonnet, llama-3.1-70b, mistral-large2)
3. Try asking a telecommunications question:

Example prompt:

What you'll see: The model will suggest various factors to consider:

Try pasting a sample of unstructured text and asking the model to extract specific information:

Example - Extracting from a support ticket:

What you'll learn:

Key insight: The prompt engineering skills you develop here translate directly to the AI_EXTRACT and AI_PARSE_DOCUMENT functions you'll use in the Notebooks.

The Document Processing Playground is a powerful AI tool that helps you understand how text is extracted from documents. It provides an interactive UI for testing AI_EXTRACT and AI_PARSE_DOCUMENT functions, allowing you to experiment with different extraction strategies before implementing them in production.

1. Click Document Processing Playground in the AI & ML Studio
2. Click Add from stage
3. Select the following:
   • Database: TELCO_OPERATIONS_AI
   • Schema: DEFAULT_SCHEMA
   • Stage: PDF_STAGE
4. Choose 1-2 PDF documents (e.g., NovaConnect help documents)
5. Click Open playground

Once your document is loaded, you'll see three tabs: Extraction, Markdown, and Text.

The Extraction tab is where you can ask questions to pull specific information from the document.

Try creating these key-value question pairs:

After entering each question, click Add Prompt to see the extracted results.

Once you've asked at least one question, the playground automatically generates SQL code:

1. Click Code Snippets in the top right corner
2. Review the generated SQL using AI_EXTRACT and AI_PARSE_DOCUMENT functions
3. Click Workspaces to open the code in a new worksheet

Example generated code:

This SQL code can be used to automate document processing at scale!

Before proceeding to the notebooks, you now understand:

✅ Cortex Playground - How to test LLM capabilities and prompt engineering ✅ AI_PARSE_DOCUMENT - Extract text and structure from PDFs ✅ AI_EXTRACT - Pull specific fields from documents using questions ✅ SQL Code Generation - Automating document processing at scale

Next: Apply these concepts in the notebooks to process help documents and call recordings!

In this section, you'll process unstructured documents using Cortex AI functions through Notebook 1.

What You'll Process:

Cortex AI Functions You'll Use:

Navigate to AI & ML Studio → Notebooks → 1_DATA_PROCESSING

PDF Processing:

Key Tables Created by This Notebook:

After running this notebook, you'll have processed PDFs ready for search and analysis.

The notebook contains detailed instructions and explanations for each step. Simply:

1. Click Start to begin the notebook session
2. Read through each cell's markdown explanations
3. Run All or execute cells one by one
4. Observe the outputs and interactive visualizations

Time: 10-15 minutes to run through all cells.

Process customer call recordings using AI_TRANSCRIBE to convert speech to text, then analyze the content for sentiment and insights.

What You'll Process:

Navigate to AI & ML Studio → Notebooks → 2_ANALYZE_CALL_AUDIO

AI_TRANSCRIBE - Converts MP3 audio to timestamped text:

AI_SENTIMENT - Analyzes emotional tone:

AI_COMPLETE - Summarizes calls:

After running this notebook, you'll have:

💡 Note: The CALL_TRANSCRIPTS table with sample data is pre-loaded during initial deployment for the search services to work immediately.

The notebook includes:

Simply run through the cells to see how audio becomes searchable, analyzable data.

Time: 10-15 minutes to complete.

Notebook 3 provides advanced analytics and visualizations combining all your processed data.

Navigate to AI & ML Studio → Notebooks → 3_INTELLIGENCE_LAB

Cross-Domain Analysis:

Visualizations:

This notebook brings together insights from all data sources for comprehensive operational intelligence.

Time: 10-15 minutes to complete.

The deployment scripts created 2 intelligent search services that make your telecommunications data instantly accessible using semantic search. This is the foundation for RAG (Retrieval Augmented Generation) applications.

Why Search Services Matter:

What's Deployed:

1. CALL_TRANSCRIPT_SEARCH - Search customer call transcripts by conversation content
2. SUPPORT_TICKET_SEARCH - Search support tickets by issue description

1. Navigate to AI & ML Studio → Cortex Search
2. Select CALL_TRANSCRIPT_SEARCH service
3. Try a search: "network connectivity problems"
4. Observe:
   • Semantic search results (not exact match)
   • Ranked by relevance
   • Attributes returned (CALL_ID, SPEAKER_ROLE, SENTIMENT_SCORE)

Cortex Analyst enables natural language querying of structured data using semantic models. Instead of writing SQL, users ask questions in plain English and Cortex Analyst generates the SQL automatically.

What's Deployed:

Purpose: Query network metrics across cell towers and regions

Key Dimensions:

Key Facts:

Example Questions:

1. "Which towers have high latency issues?"
2. "What are the best performing 5G towers by download speed?"
3. "Show me network availability by region"
4. "Which regions have the highest call drop rates?"

Purpose: Analyze bandwidth utilization and capacity planning

Key Dimensions:

Key Facts:

Example Questions:

1. "Which towers are operating at over 80% capacity?"
2. "Which towers need infrastructure upgrades?"
3. "Which towers will run out of capacity in the next 6 months?"
4. "Show me 5G towers with upgrade recommended"

Purpose: Analyze customer sentiment, complaints, and churn

Key Dimensions:

Key Facts:

Example Questions:

1. "Which regions have the most customer complaints?"
2. "What is the trend in customer sentiment over the past month?"
3. "What are the top reasons for customer churn?"
4. "Show me customers at risk of churning"

1. From the navigation bar, click AI & ML → Studio
2. Click on Cortex Analyst
3. The semantic models are available via the stage @TELCO_OPERATIONS_AI.CORTEX_ANALYST.CORTEX_ANALYST

✅ Semantic Models bridge business language and database schemas ✅ Verified Queries provide pre-tested SQL for common questions ✅ Natural Language enables non-technical users to query data ✅ Multiple Models allow domain-specific analysis

Next: These semantic models become tools for the Snowflake Intelligence Agent!

The Telco Operations AI Agent is your AI-powered telecommunications operations assistant that combines multiple data sources, search capabilities, and analytical tools to provide comprehensive insights about network performance, customer experience, and operational efficiency.

What the Agent Can Do:

This agent represents the culmination of all the previous work - it brings together the unstructured data processing, search services, and semantic models into one conversational interface.

1. Navigate to AI & ML Studio → Snowflake Intelligence in your Snowflake account
2. Select the Telco Operations AI Agent
3. The agent will automatically select the right tools based on your question

Location: SNOWFLAKE_INTELLIGENCE.AGENTS."Telco Operations AI Agent"

The Telco Operations AI Agent has access to 5 different tools that it automatically orchestrates based on your questions:

1. network_performance

2. Customer_Feedback

3. Infrastructure_Capacity

4. CALL_TRANSCRIPTS

5. SUPPORT_TICKETS

Try asking:

The agent will:

Behind the scenes: Uses the network_performance Cortex Analyst tool to generate and execute SQL.

Try asking:

The agent will:

Data sources: Infrastructure capacity metrics with utilization percentages and growth projections.

Try asking:

The agent will:

Demonstrates: Semantic search across transcribed call recordings.

Try asking:

The agent will:

Try asking:

The agent will:

Try asking:

The agent will:

Here's an example conversation demonstrating the agent's capabilities:

You: "Which regions have the highest network latency issues?"

Agent: Queries network performance, shows regions ranked by average latency

You: "Show me 5G towers operating above 80% capacity"

Agent: Queries infrastructure capacity, lists overloaded towers with utilization %

You: "Find calls mentioning network connectivity problems"

Agent: Searches call transcripts, returns matching conversations with sentiment

You: "What are the top 3 customer complaints in October 2025?"

Agent: Queries customer feedback, shows complaint categories ranked

Here are the sample questions configured in the agent:

1. "Which regions have the highest network latency issues?"
2. "Show me 5G towers operating above 80% capacity"
3. "Find calls mentioning network connectivity problems"
4. "What are the top 3 customer complaints in October 2025?"
5. "Find support tickets about billing issues"
6. "Which customer segments have the highest churn risk?"
7. "Are network issues correlated with negative customer sentiment?"
8. "Show me regions with both high latency and customer complaints"
9. "What percentage of calls mention competitor names?"

After using the agent, explore its configuration:

1. Navigate to AI & ML Studio → Agents
2. Click on Telco Operations AI Agent
3. Click Edit to view:
   • Sample Questions: The questions you just tried
   • Tools: All 5 tools available to the agent
   • Orchestration Instructions: How the agent decides which tools to use
   • Access Control: Who can use the agent

When configuring the agent, you'll see these key sections:

Agent Description:

Semantic Model YAML Files:

The agent uses 3 semantic models located in the CORTEX_ANALYST stage:

Add Tools with Detailed Descriptions:

All 3 Semantic Model Tools:

Cortex Search Service Configuration:

Test Your Agent:

The Test Agent interface is where you can interact with your AI assistant in real-time before publishing it. Here you can:

Try questions like: "What regions have the highest call drop rates?" or "What are customers saying about 5G coverage?"

Configure Sample Questions:

To customize the appearance of your Intelligence interface:

1. Navigate to AI & ML → Agents → Snowflake Intelligence tab
2. Click Open Settings

Customize the Interface:

Brand Your Intelligence with Logos:

Use the NovaConnect logos to brand your interface:

✅ Good: "Which 5G towers have latency above 25ms?" ❌ Too vague: "Tell me about the network"

The agent maintains context, so you can:

1. Ask initial question
2. Drill deeper on specific findings
3. Request different visualizations
4. Ask for comparisons

The agent automatically generates visualizations based on the orchestration instruction to convert to logs when comparing measures.

The agent is:

This ensures comprehensive, validated answers.

This is a demonstration environment showing telecommunications AI capabilities. Do not make actual business decisions based on this data!

Duration: 5

The sample data uses Malaysian regions by default (Kuala Lumpur, Selangor, Penang, etc.). If you'd like to customize the demo to use regions relevant to your geography, we've included a customization script.

1. Navigate to the SQL script in your Git repository:

2. Or create a new SQL worksheet and copy the script from: assets/sql/99_customize_regions.sql

The script includes several preset configurations you can uncomment:

To create your own mapping, modify the INSERT INTO REGION_MAPPING statements:

The script updates the following tables:

After running the customization, your agent queries will return results using your custom region names!

In this quickstart, you've built a comprehensive AI-powered telecommunications operations platform using Snowflake's Cortex AI capabilities. You started by processing unstructured data—extracting insights from PDF help documents using Cortex Document Processing, transcribing customer call recordings with AI_TRANSCRIBE, and analyzing sentiment across thousands of data points. This transformed raw, unstructured telecommunications content into structured, queryable data that forms the foundation of intelligent applications.

You then created the AI services layer that makes this data accessible and actionable. Cortex Search services enable semantic search across call transcripts and support tickets—finding relevant information based on meaning rather than keywords. Cortex Analyst semantic models allow natural language queries against structured network and customer data, automatically generating SQL from plain English questions. Together, these services power the intelligence layer that connects users to insights without requiring technical expertise.

Finally, you deployed and explored the Snowflake Intelligence Agent—the "Telco Operations AI Agent" that orchestrates all of these capabilities through a conversational interface. The agent seamlessly combines search services and semantic models to answer complex operational questions and generate visualizations. This demonstrates how modern AI platforms can unify structured and unstructured data analysis, enabling faster decision-making and deeper insights across telecommunications operations use cases.

Ready to build AI-powered applications with Snowflake Cortex? Start experimenting with your own data today!

Since the Git repo is in a separate database, you can easily reset:

Then re-run deployment scripts (01-05) from the Git Repositories UI as described in Step 3.