That’s where I think coupling a web clipper with audio transcription, CLIP, & an LLM, can provide a lot of value.
Have everything automatically classified, tagged, semantically searchable, and linked.
This way when I, for example, write down a new idea about LLMs, everything related is automatically surfaced, including those Langchain and LlamaIndex tutorials, and the notes I’d left about them.
Same thing when I look things up.
And I think this "autopilot" aspect will drastically simplify the usage of tools like Obsidian & Roam.
It won’t replace careful curation, but I am convinced it can be better than what most have (ie Notes, bookmarks, sticky notes, etc)
Perhaps that’s what the guys over at Quivr[0] are trying to do, but I haven’t gotten the chance to try it out yet.
Have everything automatically classified, tagged, semantically searchable, and linked.
This way when I, for example, write down a new idea about LLMs, everything related is automatically surfaced, including those Langchain and LlamaIndex tutorials, and the notes I’d left about them.
Same thing when I look things up.
And I think this "autopilot" aspect will drastically simplify the usage of tools like Obsidian & Roam.
It won’t replace careful curation, but I am convinced it can be better than what most have (ie Notes, bookmarks, sticky notes, etc)
Perhaps that’s what the guys over at Quivr[0] are trying to do, but I haven’t gotten the chance to try it out yet.
It’s such an obvious use case.
[0]: https://www.quivr.app/