What the AI reads

Each AI step receives the recording, text, images, or live speech needed for its task.

Information by step

AI analysis is the work that turns a recording into a transcript, feedback, and observations. Different systems receive different material. You can inspect the resulting words, quoted evidence, and recorded moments in the player.

Follow the recording and text sent for analysis.

The recorded picture and sound

ReceivesThe recording file
ReturnsSpoken words, times, and speaker labels
You seeTranscript beside the video
Select an AI step to see what it receives.

Written feedback models

OpenAI GPT-5.6 Luna, gpt-5.6-luna, writes feedback first. The Studio requests Groq gpt-oss-120b after a failed Luna request. Cloudflare Workers AI gpt-oss-120b follows as needed.

AI feedback model in Course settings can start feedback with gpt-oss-120b instead. The same model order supports grading, the task-fit check, the comment composer, and Draft from my topic. Course owners and admins receive task-fit flags.

Discussion context

Each student's own recording supplies their grade.

The AI may quote only the student's own words as evidence for their grade. The shared recording holds the group card and discussion note.

Other inputs and observations

ToneDistilBERT on Cloudflare Workers AI receives transcript intervals to classify the sentiment of the words.
Screen textTesseract reads shared-screen images on Prepara's own servers.
Expressions, posture, and gesturesThe optional course setting adds three unscored cards for instructors and admins.

Delivery, seen and heard covers Voice, Pace and pauses, Eye contact, Body and gestures, and Presence. The Studio checks replies against the recording's length and transcript. It averages two assessments.

Feedback readers

Instructors and admins can read comments and draft grades. Reviewers see feedback according to assignment settings, including Reviewers see grades. Students read their grade and feedback after posting and release.