Best practices: Building Enterprise Applications that leverage GenAI

How do you build enterprise AI apps that use your private data without expensive retraining?

#1about 6 minutes

Demonstrating the future of software with natural language

A live demo shows how Semantic Kernel enables a multi-lingual, chat-based interface to control devices like lights, illustrating the concept of "Software V2".

#2about 3 minutes

Building a natural language interface for PowerShell

An application translates plain English queries into PowerShell commands to retrieve system information like running processes and IP addresses.

#3about 2 minutes

Understanding the role of embeddings and vector databases

Embeddings provide a semantic, multi-dimensional representation of text tokens, which are stored and queried in vector databases to find semantic similarity.

#4about 2 minutes

Exploring native vector search in SQL Server 2025

A preview of SQL Server 2025 shows the new native vector data type and distance function, but also highlights potential linear performance scaling issues.

#5about 7 minutes

Using RAG to extend LLM knowledge without retraining

A C# code walkthrough demonstrates how Retrieval-Augmented Generation (RAG) injects new information into a model's context to override its base knowledge.

#6about 4 minutes