AI Building Blocks for .NET: Add intelligence to your C# apps | OD805
Microsoft Developer presents a practical Build 2026 session on adding AI capabilities to C#/.NET apps, including model switching, embeddings, search quality tuning, document ingestion into a vector store, MCP-based tool connectivity, and image generation with Azure AI Foundry.
Overview
A practical, opinionated guide to building intelligent applications in .NET, using a demo “support center” AI app to show common building blocks and patterns.
Resources:
Key segments (from the video chapters)
Demo introduction: support center AI application
The session starts with a demo application that represents a support center scenario, used as the running example for adding AI features to a .NET/C# app.
Switching and testing multiple models
The presenter demonstrates how to switch between and test different models (including GPT and Kimi) to compare behavior and results.
Multilingual embeddings and cloud model advantages
The session covers multilingual embeddings and discusses why cloud-based models can be advantageous for these scenarios.
Search quality: scores and hybrid strategies
The presenter discusses search quality considerations, including:
- How to interpret and use scores
- Hybrid strategies (combining approaches) to improve retrieval quality
Advanced ingestion pipeline: markdown reader to vector store
The session introduces a more advanced data ingestion pipeline, including:
- Reading markdown content
- Writing processed content into a vector store for retrieval
From simple examples to complex document ingestion
The demo transitions from a simple “movie string” example to a more realistic, complex document ingestion flow.
MCP for information access and external tool connection
The presenter shows how MCP can enhance information access and connect the application to external tools.
Agent using an external tool: Microsoft Doc Search
A demonstration of an agent calling an external tool, using a Microsoft documentation search capability.
Image generation with Foundry (text-to-image)
The session closes with image generation, demonstrating a text-to-image tool using Azure AI Foundry.