Using AI Features in SQL Server 2025

Store, retrieve, and integrate data for AI systems using SQL Server

Course Summary

Do you want to start working with AI-related features in SQL Server 2025?

As these features are relatively new, the focus of this course is on design and architecture decisions that remain valid, even as individual features, syntax, and external AI services evolve. It does not aim to turn SQL Server into an AI platform, nor does it attempt to teach model training, data science, or vendor-specific AI tooling.

This course is about using SQL Server to store, retrieve, and integrate data for AI systems, in a way that is reliable, understandable, and operationally sound. We assume you already understand SQL Server fundamentals and are comfortable working with T-SQL and database design.

Are you responsible for deciding where AI-related logic should live, and where it should not?
Do you need to support vector search, retrieval, or AI integration without turning the database into an experimental system?
Have you been asked whether SQL Server should be used for embeddings, RAG patterns, or AI orchestration, and wanted a clear answer that you can explain?

If that matches the kind of responsibility you have, you’re in the right place. You’ll get clear explanations, hands-on exercises, and short quizzes to make sure you not only understand the concepts but can confidently apply them in real SQL Server environments.

And while the practical exercises are optional, we strongly encourage completing them. They are designed to be straightforward and focused on realistic scenarios, rather than tooling complexity. Setup is easy. The main requirement is access to a SQL Server 2025 environment (trial or production) or a system you can install SQL Server on.

Enroll now for $195 USD !

Modules and Lessons

Module 0: Getting Started

Who is this course for ?
Who is Greg ?
What will I learn in this course ?
Configuring your lab environment

Module 1: AI in the context of SQL Server

Module introduction
AI features in SQL Server 2025
What is new in SQL Server 2025
What SQL Server is responsible for
What SQL Server is not responsible for
Misconceptions to avoid
Positioning SQL Server correctly
Lab 1
Lab 1 walkthroughs

Exercise 0: Overview
Exercise 1: Installing SQL Server 2025 with AI features
Exercise 2: Applying SQL Server 2025 updates
Exercise 3: Installing SQL Server Management Studio

Quiz 1

Module 2: Vector data and embeddings fundamentals

Module introduction
What is an AI model ?
What embeddings are and why they exist
How vector similarity differs from relational comparison
How SQL Server uses vector similarity
Typical AI scenarios that involve vectors
Why vectors work well for these scenarios
When vectors are inappropriate
How this applies to SQL Server
How to think about vectors going forward
Where embeddings models are hosted
Common embeddings models
Lab 2
Lab 2 video walkthroughs

Exercise 0: Overview
Exercise 1: Installing and configuring Ollama
Exercise 2: Test embeddings

Quiz 2

Module 3: Vector data types in SQL Server

Module introduction
Vector data type basics
Declaring VECTOR columns and variables
How VECTOR values are represented
Embeddings as derived data
Practical table pattern for embeddings
Vectors complement relational data
Parallel tables and schema evolution
Dimensionality is enforced
Dimensionality changes are migration events
Storage characteristics of VECTOR
float16 vectors
VECTOR limitations in SQL Server
Lab 3
Lab 3 solution
Lab 3 video walkthroughs

Exercise 0: Overview
Exercise 1: Create the AIFeatures database
Exercise 2: Work with the vector data type

Quiz 3

Module 4: Querying vector data

Module introduction
What similarity search means in T-SQL
Exact similarity vs approximate search
Writing similarity queries in T-SQL
Using VECTOR_DISTANCE
Combining vector similarity with relational predicates
Using VECTOR_SEARCH
Common query mistakes and inefficiencies
How to reason about vector queries in T-SQL
Using VECTOR_NORM
Using VECTOR_NORMALIZE
Using VECTORPROPERTY
Lab 4
Lab 4 solution
Lab 4 video walkthroughs

Exercise 0: Overview
Exercise 1: Create and populate the Products table
Exercise 2: Work with vector functions

Quiz 4

Module 5: Vector indexing and performance

Module introduction
Why vector indexing exists
Vector index characteristics
Operational constraints of vector indexes
Creating a vector index
Accuracy versus performance trade-offs
CPU and memory impact of vector queries
Monitoring and diagnosing vector query performance
When vector indexes should not be used
Lab 5
Lab 5 solution
Lab 5 video walkthroughs

Exercise 0: Overview
Exercise 1: Add the ProductVector column and populate it
Exercise 2: Work with a vector index

Quiz 5

Module 6: Generating embeddings outside SQL Server

Module introduction
Why SQL Server does not generate embeddings
Common embedding generation patterns
Storing embeddings safely in SQL Server
Refreshing and regenerating embeddings
Handling changes in embedding models
Versioning strategies for embeddings
Re-embedding strategies at scale
Quiz 6

Module 7: REST and HTTP basics

Module introduction
What REST means in practice
Core HTTP concepts
JSON as a data format
Authentication at a conceptual level
Error handling and diagnostics
How REST concepts apply to SQL Server integrations
What SQL Server professionals need to know
Using Caddy as a proxy
Testing the Caddy proxy
Lab 7
Lab 7 solution
Lab 7 video walkthroughs

Exercise 0: Overview
Exercise 1: Start Ollama and configure Caddy as an HTTPS proxy
Exercise 2: Trust the Caddy root certificate
Exercise 3: Enable REST endpoint access for SQL Server
Exercise 4: Check external REST access
Exercise 5: Test that Caddy works

Quiz 7

Module 8: Calling AI services from SQL Server

Module introduction
Synchronous vs asynchronous calls
Patterns for Invoking External AI Services
Security considerations for outbound calls
Cost and throttling implications
Operational risks and failure modes
Designing SQL Server–safe AI integrations
Using sp_invoke_external_rest_endpoint
Calling sp_invoke_external_rest_endpoint
Minimal structure of a REST call in T-SQL
Retrieving text embeddings via SQL Server REST
Using CREATE EXTERNAL MODEL
Using AI_GENERATE_EMBEDDINGS
Handling REST authentication
Supported external model types
Lab 8
Lab 8 solution
Lab 8 video walkthroughs

Exercise 0: Overview
Exercise 1: Retrieve embeddings from Caddy
Exercise 2: Create an external model
Exercise 3: Test the external model
Exercise 4: Reconfigure the products table
Exercise 5: Populate the retrieved embeddings

Quiz 8

Module 9: Retrieval-augmented query patterns

Module introduction
What RAG means in practical terms
Designing SQL tables for retrieval
Executing similarity search as part of a workflow
Passing retrieved data to AI services
Limitations of RAG for SQL Server
Failure modes in SQL Server–based RAG
How to use SQL Server effectively in RAG
Lab 9
Lab 9 solution
Lab 9 video walkthroughs

Exercise 0: Overview
Exercise 1: Create a vector index on the retrieved embeddings
Exercise 2: Retest vector search
Exercise 3: Search with an updated query

Quiz 9

Module 10: Security, governance, and operational concerns

Module introduction
Protecting sensitive data used in AI workflows
Controlling access to vector data
Auditing AI-related queries
Managing secrets and credentials
Operational risks introduced by AI features
Governance boundaries in AI-enabled systems
How to approach AI governance with SQL Server
Quiz 10

Module 11: When NOT to use AI features in SQL Server

Module introduction
Why it matters to say no
Scenarios where SQL Server is the wrong tool
When a separate vector database makes more sense
When Fabric or other platforms are more appropriate
Cost, complexity, and maintenance trade-offs
Decision checklist for architects and DBAs
Quiz 11

Module 12: Next steps

Summary and further steps

Enroll now for $195 USD !