Wise Help CenterAI Session Health is a paid add-on, billed on usage. It uses AI, not a human reviewer, to automatically score every ended live session, flag the ones that deserve a look, build per-tutor reports, and link every flagged moment straight to its timestamp in the recording.
For the full walkthrough of every tab, screen, and setting, see the detailed admin guide: Session Quality: Admin Guide. For a product overview, see wise.live/tutoring-session-quality-monitoring. Try a live demo of the dashboard: session-quality.integrations.wise.live/demo. A structured summary follows below.
- How scoring works
- Every ended session is scored automatically, usually within about 30 minutes.
- Two-stage AI review, with no human reviewer involved: an AI Gatekeeper estimates every session quickly; an AI Judge reviews selected sessions in depth, producing scores, reasons, highlights, and flags.
- How many sessions get the in-depth Judge review is a monthly budget you set (50–20,000 reviews/month), with guaranteed minimum coverage per tutor and per batch — see Configuration below.
- A session can also come back Estimated only (Gatekeeper pass only, no band, not counted in averages) or Not analysed (no attendees, fewer than 15 transcript lines, a regional-language class, or no transcript/recording yet).
- Rubric: 20 measures across 5 groups
- Relationship and conduct — Rapport & opening, Interaction health
- Engagement and participation — Engagement, Name usage, Students spoke, Students punctuality
- Instruction — Clarity, Model → guide → practice, Checks for understanding, Personalisation
- Structure and objectives — Clear objective, Structure & pacing
- Hygiene and readiness — Classroom management, Technical readiness, Teacher punctuality, Camera on, Virtual background
- Demo-session-only measures (scored only on sessions recognized by class name as a demo, consultation, or parent-teacher meeting) — Demo: early hook, Demo: parent engagement, Demo: child showcase, Demo: next-step close
- Severity bands
- Healthy — no signals; no review needed.
- Quick monitoring — one or two minor signals; worth a glance.
- Review recommended — three or more signals, or any signal carrying a review floor.
- Critical — at least one severe signal; look here first.
- Signal thresholds are recalibrated automatically every week, per institute, so bands stay meaningful as volume and norms change.
- Overall score and verdict
- Overall score: 0–100, a weighted average across every scored measure (weights you set, renormalised when a measure is switched off, zero-weighted, or not applicable to that session).
- Strong session: 85+ · Solid session: 70–84 · Needs coaching: 55–69 · Escalate: under 55.
- Evidence and recording links
- Flagged measures show a timestamped quote; selecting it jumps the recording player straight to that moment.
- A Highlights tab lists good and needs-review moments, each with its own play button.
- "Clip a moment" pins an exact start/end stretch of a recording to a measure, for coaching later.
- Reports
- Overview — band counts, a sessions-by-day chart, the overall-score trend, which measures are trending weak, and which tutors need attention.
- Sessions — the full session list, filterable by band, flag, tutor, and reviewed/bookmarked status.
- Teacher reports — a per-tutor comparison table, plus an individual tutor report (strongest/weakest measures, trend by period, flags, full session list) that prints straight to PDF for a coaching conversation.
- Configuration
- Review coverage — review every session in depth, or set a planned monthly sample with guaranteed minimum coverage per tutor and per batch.
- Session scoring — switch any measure on or off and reweight it to match your own rubric; changing a weight rescores all history, so trend lines don't jump from a settings change.
- Who sees what
- Owners/admins: full access to every tab, every tutor's data, all review tools, and Configuration.
- Tutors: see only their own sessions, read-only, with no review tools or settings access.