bickov
Joined 11 years ago
Karma
1
Submissions
51
Member since
Mar 6
Designer/PM in Riga. Currently building slimsnap.ai.
View on Hacker NewsRecent submissions
Comment · 2 months ago
The per-app breakage is the real problem with text extraction, not accuracy, like etchalon pointed out. Screenshots dodge that since they look the same no matter what is rendering. Cropping to the window and marking it up before it goes anywhere also skips the full screen recording and the orange icon. https://slimsnap.ai already works that way.
Comment · 3 months ago
[flagged]
Comment · 3 months ago
The tiredness isn't from being in the loop. It's from what the loop hands you. Reading a wall of an agent's prose to check if it understood a screen is exhausting. The same check as a marked element with coordinates takes a second.
It's not a human-attention problem, it's format problem. Models emit prose because it's cheap to generate, not cheap to review.