We got tired of fighting compile logs.
Scripta started as an internal tool for a systems lab that couldn't agree on an editor. It's now used by 34 departments and research groups — same small team, much less duct tape.
How we got here
2024
A shared .tex file becomes a problem
Three grad students in a systems lab kept emailing draft revisions back and forth. After the third "final_v3_ACTUAL.tex" attachment, one of them wrote a script to sync edits instead. That script is the reason Scripta exists.
2025
Opened up past the lab
Word got around the department, then the university. By spring, 30-odd research groups were using an internal build held together with cron jobs. We spent the year rebuilding it properly.
2026
AI review, built in from the start
Once the editor itself was solid, we added the thing every advisor already does by hand: reading a draft closely enough to flag what's actually wrong with it. Now it happens before the draft reaches your advisor.
Four things we keep saying no to changing
Compile speed is a feature
A 30-second compile loop kills the habit of checking your work. We treat compile latency as a product metric, not an infrastructure afterthought.
AI reviews, humans decide
Suggestions are just that — inline, dismissible, and never silently applied. Nothing in your document changes without you clicking accept.
No lock-in, ever
Standard .tex/.bib on the way in, standard .tex/.bib on the way out. If we're not earning the subscription, leaving should take five minutes.
Built with the people who'd use it
Every major feature shipped in the last year started as a complaint from an actual research group, not a roadmap guess.