Skip to content

Retrieval-augmented Conversational AI Tutor with modular GROW-aligned coaching routines

Design · All designs
Evidence · no claims cited

Description

The Conversational AI Tutor is a retrieval-grounded conversational module built on GPT-4.1 with a RAG layer that anchors each response in the learner's own multimodal records (transcripts, BARS scores, emotion descriptors, feature summaries) and maintains records across submissions for longitudinal reflection. The tutor implements three modular routines—performance summarization, strength and weakness highlighting, and translation of analytic indicators into learner-facing concepts—with phrasing drawn from a curated library of expert coaching aligned to the BARS rubric and the GROW model (Goals, Reality, Options, Will).

Design Implications

Context

Requirements

  • Verifiable learner records to retrieve; modular routine specifications with fixed instructional goals, required evidence fields, and output schemas to reduce stylistic drift

Constraints

  • The RAG layer restricts generation to retrieved sources from the learner's own records, minimizing unconstrained free-form output

Target Learners

  • adult learners in self-directed online presentation practice

Learning Goals

  • interpreting analytic feedback and planning concrete delivery improvements

Claims

Examples

-

Key Sources

  • Hung-Yue Suen and Kuo-En Hung. (2026). An Interpretable Closed-Loop Intelligent Tutoring System for Multimodal Affective Feedback in Asynchronous Presentation Training. IEEE Transactions on Learning Technologies. https://doi.org/10.1109/TLT.2026.3693864
Page metadata
Field Value
type design
id rag-conversational-ai-tutor-modular-coaching-routines
title Retrieval-augmented Conversational AI Tutor with modular GROW-aligned coaching routines
description The Conversational AI Tutor is a retrieval-grounded conversational module built on GPT-4.1 with a RAG layer that anchors each response in the learner's own multimodal records (transcripts, BARS scores, emotion descrip...
status review
generated by process:settle-candidates on 2026-10-11
trust unverified — no one has confirmed this page's content
sources 1 — hung-yue-suen-and-kuo-en-hung-2026

Frontmatter is the page's machine-readable record — see the Schema & Guide.