It's a quick read at one Python file and a schema. Notes go into Postgres with embeddings so AI tools can search by meaning.
For your family-office archive, it's the wrong layer on its own. It needs an embedding API and lives in a database, so it doesn't meet your 'outlast AI models' bar.
The markdown vault is the store you pass down. This could be the searchable index over it. Obsidian's a solid pick for the vault side.
