The lab tradition: experiment records go on paper. Over three years, my notebooks stacked up to a hand-span high. On the day I wrote my thesis, I realized these records might as well not exist—because I couldn't find a single page.
First I tried WorkBuddy
It is an excellent conversationalist, but conversation requires text to enter the chat box. Handwritten notebooks, scans, photos—materials "outside the chat box"—it had no way to collect them. I could only type the content and feed it to it, and typing itself was half a day's work.
Then I used MuchSci
Unified collection of multi-format materials is its core function: photograph handwritten pages and upload them, put scans directly into the library, archive and name them by project and date, and attach annotations alongside—"Reason for this failure: temperature drift," "Next improvement: extend holding time." Now when I need to check a certain experiment, one search locates that page.
By comparison, collection capability is the dividing line: general assistants work within conversations, while vertical platforms work across all your materials. The problem with paper records has never been that they can't be preserved, but that they can't be searched—spend one extra minute filing, save an afternoon when writing your thesis.
Four steps for archiving: photograph and store, archive by project, add annotations, retrieve and use. Three years of paper pages, one afternoon, all turned into archives you can summon at any time.