That´s interesting! I want to know if you are a paper person? Do you mostly like real physical paper (writing and reading)?
This is a little off topic but I’ll answer it because it’s a good question.
Actually, I’m not - I like notebooks but I never use them religiously unless I’m working with a client on-site which does not happen much at the moment.
For the rapid-fire prototyping I’ve been doing I’ve resorted to something ridiculous: screenshots and OCR search as well as a really thorough AI-assisted documentation system where I can keep records and summaries and then grep over hundreds of documents later to revisit what I’ve done.
I love the
idea of slowly methodically keeping a lab notebook but the tooling speed at the moment kind of precludes it. That said, I could see a good argument for doing so in order to slow down and really think through problems.
This is only my process though, I’ve perpetually struggled with PKM systems - I used org-mode in emacs for 4 years when I was managing a research department but even that I didn’t use properly, I basically kept a long running list every day in one giant file and moved up the items that didn’t get done from yesterday to today (or friday to monday, etc.) It worked surprisingly well even without using all the org-mode flair and prioritization stuff. Tasks were enough.
Now I have multiple agents keeping tasks on my behalf and I have some overarching research documentation in markdown files that I revise and iterate on, including append-only logs and retraction logs.
All the tooling is great but it is kind of …a lot. Most of this a Vision Pro would not really help with, but I am still in the market for one eventually if they make it lighter even if mostly for media and novelty.
I absolutely can see a future where ambient computing with extremely good AI assistance can do a lot of what I’m doing, if it can internally track all the complexity and kind of abstract it away from me. But consumer stuff I don’t think will ever get there. Anthropic
briefly allowed you to use full voice interaction with their best models and that was an
incredibly productive time for me because I was able to rubber-duck and brainstorm freeform while walking around and I found that very valuable, to the point where I might eventually code my own version of it if they never reintroduce it. The prosidy they were using (which I’ve since learned was actually farmed-out) was incredible, the voice could even do imitations.
We’re really only on the cusp of what’s possible, and I’m not a blind AI booster either - there are a ton of issues - but there is also a lot of significant potential and a spatial computer / vision pro descendant could absolutely be a part of that world.