I’m not a power-user on this, so apologies if this is a bit basic or a repeat question. I’ve been exploring using ChatGPT (I’m on the Premium plan) tool‑ing with my Obsidian vault, and I’m not sure if there’s a go‑to setup for this kind of integration. I came across “Smart Connect” by Brian Petro and wondered if that’s still considered the best or most reliable solution today.
Has anyone here:
Already plugged Obsidian vaults into ChatGPT Premium LLM?
Used Smart Connect, and does it work smoothly?
Or do you know of an even better or more modern workflow for this?
Would really appreciate any insights or tips (YMMV, links, gotchas, etc.). Thanks in advance
Personally I find Smart Connect to be very resource intensive, much like CoPilot. I found my vault to lag too much (even with high PC specs) that I ended up uninstalling both plugins and opting to use my AI Tools via Custom Frames or on a second monitor. You do lose the “Smart Connect” and “Related Note” feature but it’s not a deal breaker for me. I prefer to write my own notes then use AI Tools to improve my grammer and delivery of those notes.
Example, my original thoughts above vs Improved by Claude 4 Sonnet below
Personally, I find Smart Connect to be very resource-intensive, similar to CoPilot. Despite having high-end PC specs, my vault experienced significant lag with both plugins installed, which ultimately led me to uninstall them. Instead, I now use AI tools through Custom Frames or on a second monitor.
While this approach means losing the “Smart Connect” and “Related Notes” features, it’s not a dealbreaker for me. I prefer writing my initial notes myself, then using AI tools to refine the grammar and improve the delivery of those notes.
Hi Paul, thank you for your reply. Watching your videos it became clear to me that my main purpose is to use LLM to query the whole vault. Not to work with LLMs on specific or within specific notes. I want to upload the vault as the whole knowledge graph and just have ChatGPT premium on top for meta level queries.
Because you were already mentioning the performance issue. How can I achieve this in the easiest possible way?
Claude Desktop + MCP Tools could be an option but you have your Premium account with OpenAI. I didn’t really like waiting for Claude to finish using MCP tools but it’s interesting to experiment and play with.
You could look at a VS Code + Cline + MCP Tool Server setup. This would work well with your Open AI Premium Account. Your conversations with your Vault would be in VS Code though not Obsidian.
