https://qualiastudios.dev/blog What Qualia is building, and what we learn on the line. en https://qualiastudios.dev/blog/decision-models-for-robotics-data-curation qualia-blog-decision-models-for-robotics-data-curation Wed, 07 Oct 2026 09:00:00 GMT A lot of robot data curation is small multiple-choice calls about text. Does this report back up its verdict? Which slot does this note mean? Is this the episode the curator was looking for? A decision model is built for exactly that. You hand it the text and the options, and it picks one and tells you how sure it is, in well under a second. We've been using Jev from TypeSafe for this. OpenAI just opened its own, the Decision API, so we ran both through every test we have on our robot data, next to the big general models and our own harness. https://qualiastudios.dev/blog/what-does-a-vision-model-miss-when-you-just-hand-it-the-recording qualia-blog-what-does-a-vision-model-miss-when-you-just-hand-it-the-recording Wed, 07 Oct 2026 09:00:00 GMT The quick way to curate robot data is one API call per recording. Send the video to a vision model and ask whether the episode is good. We built that, gave it every advantage we could, and measured it against the Qualia curation harness on the same 106 recordings, three runs each. We call the one-call approach the straight feed.