Back on Agentic AI in Azure!

Spending the day working through a decent agentic ai tutorial. It’s my first time with Azure, having gone a round with AWS and GCP.

Tutorial:

Login: ai.azure.com at azure foundry
Learning Track Start: https://microsoftlearning.github.io/mslearn-ai-studio/Instructions/01-Explore-ai-studio.html
AI Studio Tutorial: https://learn.microsoft.com/en-us/azure/ai-studio/quickstarts/get-started-code?tabs=macos

Tooltip help: Alike the first time on MS Office, the tooltips are familiar and work quite well in terms of navigating the platform.
Lots of documentation: In terms on copilot on azure to churn through their documentation or, learn.azure.com. AI is still moving pretty fast and the documentation usually doesn’t keep up with the API releases. There is little planning when these APIs are released, there is little backward compatibility. Why/How happened to all the API standards and web standards for a reliable backend? It went all out the window in exchange for speed.

Plowed through the first 6 labs pretty quickly. I had some knowledge gaps filled. I got stuck at Azure Foundry, the documentation and UI for the foundry has changed alot.
You have to create a project in Azure Foundry, then deploy a AI model of your choice. You pay more per token (presumedly for OpenAI API calls + Azure’s hosting costs). It makes little sense to do so unless you are fine-tuning, RAG-training or custom training your own LLM, you would connect directly to OpenAI otherwise and skip the middleman. Unless going through Azure provided some convenience that is worth paying for. Or, you are doing a tutorial like me, walk that path to see what it looks like, per se.