AI-based features can make almost any software more powerful and productive, and modern tech stacks are evolving rapidly to make this simpler and more reliable. This doesn't just mean chatbots - in fact with semantic search, data extraction, summarization, translation, anomaly detection, and intelligent workflows, it's quickly becoming the baseline for modern software that feels satisfying to use and lets humans skip over the mundane parts of their work.

We'll go from small models that run locally without even needing a GPU, through the current mainstream of large language models (LLMs) in the cloud, up to multi-agent clusters, and consider the tradeoffs for your scenarios. Along the way we'll use .NET-based examples, and see what's coming soon to .NET that will streamline and standardize the usage of language models, vector stores, and text embeddings.