Introduction
This transparency note covers two AI features that offer live suggestions during oral activities: the Interaction Coach, which supports students, and the Conversation Assistant, which supports teachers and assessors. Both are available in Oral Assessment, only the Interaction Coach is available in Oral Reflection and Oral Practice (when used between peers).
Both features are optional and make suggestions only — they never speak in the conversation, never assess, and never grade.
Functionality
- Interaction Coach (student-facing) — shows the student on-screen suggestions during their session, for example talking points that help them cover a topic more fully. It is enabled per topic by the teacher.
- Conversation Assistant (teacher-facing) — suggests questions and follow-up questions to assessors during a live session, based on how the conversation is developing. Suggestions are visible to assessors only; students never see them.
Both features work from the live transcript of the session (see the Transparency Note: Transcription in Oral Assessment & Practice). Suggestions are generated by Claude Opus 5, accessed through Amazon Bedrock with regional hosting. Inputs are processed entirely within AWS infrastructure and are not sent to Anthropic. Beyond limited automated abuse detection performed by AWS, inputs are not retained after processing, and they are never used to train, retrain, or fine-tune any model.
User Impact
For students, the Interaction Coach acts as a gentle prompt to go deeper or address something they may have missed — the student decides what to do with a suggestion. For assessors, the Conversation Assistant reduces the effort of formulating probing questions on the spot, while leaving the direction of the conversation entirely in their hands.
Limitations and Risks
Suggestions are generated from the live conversation and may occasionally be off-target, repetitive, or less relevant to your assessment goals. They are scaffolds, not requirements: assessors and students should treat them as optional input and rely on their own judgment.
Transparency, Privacy, and Data Governance
Suggestions are generated in real time and are based on the session’s transcript. Recordings and transcripts are governed as described in the Transparency Note: Transcription in Oral Assessment & Practice.
Human Agency and Oversight
The conversation always belongs to the people in it. Assessors choose which suggested questions to ask, students choose whether to act on coaching suggestions, and all feedback and grading decisions are made by humans.