devwithjev
โ€” reading nowโ€” views
Submit a build

Whiteboard Context for Voice Interviews

Jev answers questions about whiteboard content for a voice interviewer that cannot see the whiteboard. The author says it answers a dozen such questions in under half a second.

View on X timeunder half a second
k0n00@k0nOO๐•
My voice interviewer can't see the whiteboard; GPT-Live takes no images. The box I named "Fastly" is read as a CDN at 1.00, and my note "sync or async transcode?" as a question I still owe. Jev answers dozen of questions like that in under half a second. The voice never hears one. The first try was a compact grammar with IDs and a 3x3 grid for positions. It broke on one stray box far from the rest; the grid squashed everything else into one cell. And a realistic canvas of 14 elements was about 90 tokens anyway, so saving tokens was solving nothing. In the end, it's plain English from a determ
Sep 24, 2026X postsView on X
An initial compact grammar with IDs and a 3ร—3 position grid failed when one box sat far from the others, squashing the remaining elements into one cell. The author says a realistic canvas of 14 elements was about 90 tokens, so token savings were not useful.

Also filed under Tools & apps