{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/aa4f8c399917402bbc8a75789563d0d7\" frameborder=\"0\" width=\"2034\" height=\"1525\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1525,"width":2034,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1525,"thumbnail_width":2034,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/aa4f8c399917402bbc8a75789563d0d7-36adf1d56beafe0d.gif","duration":625.086,"title":"Improve Exercise Selection Accuracy","description":"This Loom explains how to improve exercise selection accuracy when programming workouts using the program feature. It shows that the default generation sends multiple prompts to an LLM and then tries to pick the most applicable exercise outputs, but it may select nothing or the wrong variation. The speaker emphasizes that the biggest improvement comes from keeping your exercise library as complete and specifically named as possible, including adding exact names and abbreviations like DA for double arm. If you do not want to manually add exercises, you can sync from YouTube or Vimeo via the org settings using a channel URL or channel ID, then approve synced exercises, and Vimeo is also supported. If syncing and matching still do not work well, contact support for algorithm tweaks."}