THAT’S LITERALLY INSANE. I GAVE GPT-6 ASTRA ACCESS TO MY OBSIDIAN VAULT, AND THE REASONING DEPTH GENUINELY SURPRISED ME. Years of notes. Old projects. Half-finished ideas. Questions I kept coming back to. I wanted to see what would happen when the conversation started with all that context. → Here’s the test that makes this interesting. Ask a question that crosses years. “What pattern keeps repeating across my projects?” Now the task involves connecting things you wrote at different times, for different reasons. A launch journal. A project retrospective. A weekly review. Then push on the connection. “Which notes support that?” “Which ones contradict it?” “What are you assuming?” That’s where I’d spend my time. Following the sources. Checking whether the explanation holds up. Finding out whether an apparent pattern survives a harder question. Your old notes become something you can interrogate. The decisions you forgot making. The assumptions you never revisited. The same problem showing up under three different names. The video visualizes that workflow. It’s an illustrative concept, not a recording of my vault or a benchmark. My reaction is personal. Try the same questions with your own notes and judge the answers yourself. I documented the whole thing. Full walkthrough in the article below ↓