# Transcripts iframe ## Description The **Transcripts tab** lets instructors and admins review **real conversations** between learners and a specific mentorAI—with **context and cost metrics**. See summary KPIs (messages per conversation, average rating, estimated cost), search by topic or user, and open any session to read the **exact learner inputs and mentor replies**. Use these insights to improve instruction and tune your mentor’s **prompts, datasets, and tools**. ## Target Audience **Instructor · Administrator** ## Features #### Session KPIs at a Glance Headline metrics for the selected period: - **Avg messages per conversation** - **Avg conversation cost** - **Avg rating** #### Global Counts Totals for **conversations**, **user queries**, and **assistant responses** in the time window. #### Search & Filter Find transcripts by **topic** (e.g., “mitosis”) or by **user** when you need to check on a particular learner. #### Per-Conversation Details For each transcript, view: - **User name/username** (or *Anonymous*) - **Mentor name** - **LLM model** - **Messages exchanged** - **Estimated cost** - **Timestamp** #### Full Conversation Viewer Read the **exact learner questions** and the **mentor’s responses** to evaluate **clarity, tone, and accuracy**. #### Exports Available Elsewhere Download conversation data from **Data Reports → Chat History** when you need spreadsheets or BI analysis; quick viewing is also available in **Chat History**. ## How to Use (step by step) #### Open Analytics → Transcripts - In the mentor header, click **Analytics**, then select the **Transcripts tab**. #### Review Summary Metrics - Scan **avg messages/conversation**, **avg cost**, and **avg rating** to gauge conversation quality and efficiency. #### Search by Topic or User - Use the **search bar** to locate transcripts about a specific concept or for a specific learner you want to check in on. #### Inspect Global Counts - Note totals for **conversations**, **user queries**, and **assistant responses** to understand overall load and activity. #### Open a Transcript Click any session to view: - **User identity** (or *Anonymous*) and **username** (if login is required) - **Mentor**, **LLM model**, **messages exchanged**, **estimated cost**, and **timestamp** - The **exact Q&A exchange** between learner and mentor #### Decide Follow-Ups - If you see **confusion** or **low ratings**, adjust: - **Prompts** - **Datasets** - **Tools** (e.g., *Web Search*, *Code Interpreter*) - Reach out to specific learners based on what you observe. #### Export if Needed - Go to **Data Reports → Chat History** to download **CSVs** for deeper analysis or archival. ## Pedagogical Use Cases #### Identify Knowledge Gaps Spot recurring misunderstandings and create **targeted reviews, examples, or mini-lessons**. #### Quality & Tone Assurance Ensure responses are **accurate, on-brand, and student-friendly**; refine the **System Prompt** where needed. #### Support at the Right Time Use **user-level transcripts** to reach out with **resources or office-hour invitations**. #### Measure Impact of Changes After updating **prompts**, **datasets**, or **tools**, compare new transcripts’ **ratings**, **message lengths**, and **costs**. #### Assessment & Curriculum Tuning Topics that dominate transcripts may indicate where **lecture materials or assignments** need clarification. Use **Transcripts** to move beyond surface metrics—**read the conversations themselves**, understand learner needs, and continuously improve both your **teaching** and your **mentorAI**.