Develop AI solutions with Azure Database for PostgreSQL: Review

I have a Postgres database on Azure and thought this would be a quick way to bolt on vector search. The path is basically three documentation pages stitched together. The first module is the standard 'how to connect to Postgres on Azure' tutorial. If you need this part, you're probably not ready for the rest of it.

The second and third modules get to the point: enabling `pgvector`, creating a table with a vector column, and running similarity searches with different indexes like HNSW. The code snippets are fine. They work. I copied them into my own environment, swapped out the table names, and got cosine similarity results against my own data. That’s the entire value proposition here.

The 'exercises' are a joke. They give you the complete code block and tell you to run it. There's no challenge, no bug to fix, no thinking required. It's a glorified copy-paste tutorial. I appreciate that it's direct and doesn't waste time on hype, but calling this 'developing an AI solution' is a stretch. This is 'running three SQL queries'. It saved me maybe an hour of reading the raw pgvector GitHub, but it taught me nothing about the trade-offs or how to architect a real RAG pipeline around this. It's just the database part, isolated and sterilized.