Exam Center
Book a demo
Platform

Six ways we watch. None of them decide.

Typed detections with confidence scores, weighted into a risk score. Crossing the threshold opens a case for a person — it never issues a verdict.

An illustration of the live monitoring viewA schematic of the monitoring queue: one row per in-progress session with its candidate code, assessment, state and risk band. The highest-risk session is highlighted, with its accumulated violations shown alongside. This is a drawing, not a screen capture.NORTHVALE INSTITUTE OF TECHNOLOGYLIVEDashboardReports & analyticsCandidatesQuestion bankAssessment librariesAssessmentsSessionsMonitoringReviewGroupsSettingsMonitoring · 5 sessions in progressALLHIGH+FLAGGEDCANDIDATEASSESSMENTSTATERISKVIOLATIONSC-10428Data Structures · Sem 3IN PROGRESSLOW—C-10431Data Structures · Sem 3IN PROGRESSMEDIUM2C-10436Data Structures · Sem 3IN PROGRESSCRITICAL5C-10440Data Structures · Sem 3SUBMITTEDLOW1C-10442Data Structures · Sem 3IN PROGRESSHIGH3
C-10436 · why it is CRITICAL
PHONE_DETECTED+9
NO_FACE+4
OFF_SCREEN_GAZE+2
OPENS A REVIEW CASE — A PERSON DECIDES
Interface illustration, not a screen capture. Drawn to show how risk, violations and evidence relate — the same relationships you would see in the console, without a real tenant’s data.
The integrity pipelineA detection produces a violation. The violation contributes weight to a risk score. Past a threshold a review case opens. A person then records the decision — the only step that is not automatic, and one the candidate can appeal.LIVE · ONE SESSIONRISK 62 · HIGH01 · DetectionMODEL OR RULECONFIDENCE 0–100raises02 · ViolationTYPE · SEVERITYRISK WEIGHTweights03 · Risk scoreACCUMULATED WEIGHTTHRESHOLD62 / 50opens04 · Review caseOPENS AT THRESHOLDNO STEP IS AUTOMATIC PAST THE FOURTHto a reviewer05 · A person decidesOne decision, with a reasonAPPROVEWARNINGMANUAL REVIEWINVALIDATEDISQUALIFYappeal reopens
01

What is detected

Identity (NO_FACE, MULTIPLE_FACES), gaze and pose (OFF_SCREEN_GAZE, HEAD_TURNING), objects (PHONE_DETECTED, BOOK_DETECTED), audio (WHISPERING), browser activity (TAB_SWITCH, WINDOW_FOCUS_LOST) and connectivity (MONITORING_GAP, CONNECTIVITY_LOST).

Six detection lanesSix kinds of signal — identity, gaze, objects, audio, browser activity and connectivity — feed one session timeline. Each detection carries a confidence score.NO_FACEOFF_SCREEN_GAZEPHONE_DETECTEDWHISPERINGTAB_SWITCHMONITORING_GAPONESESSIONEACH WITH A CONFIDENCE SCORE
02

Tuned by you, per account

Every detector can be enabled or disabled, with its own threshold seconds, severity and risk weight — set per account and overridable per assessment. The institution decides how sensitive its exams are, not the vendor.

Every detector is yours to tuneEach detector can be enabled or disabled and carries its own threshold in seconds, severity and risk weight — set per account and overridable per assessment.PHONE_DETECTEDThresholdSeverityRisk weightPER ACCOUNT · PER ASSESSMENTYOU SET THE SENSITIVITY
03

Runs on ordinary hardware

Detection runs on CPU with a configurable sampling cadence. Models are swapped by editing a manifest, not by redeploying — so improving a detector is an operational change, not a release.

Detection runs on ordinary hardwareDetection runs on CPU with a configurable sampling cadence. Models are swapped by editing a manifest, not by redeploying.What it costs to runCPU onlySampling cadenceNo GPU neededA MODEL SWAP IS A MANIFEST EDIT, NOT A RELEASE
04

What this is not

It is not a live video wall. Monitoring is a dashboard of sessions, risk and violations that refreshes as they change. There is no automatic disqualification, and no scoring of a candidate’s character.

What this is notMonitoring is a dashboard of sessions, risk and violations. It is not a live video wall, there is no automatic disqualification, and nothing scores a candidate’s character.What people fearA LIVE VIDEO WALLAUTO DISQUALIFYA BEHAVIOUR SCOREWhat it isA RISK DASHBOARDA CASE FOR A PERSONTYPED EVIDENCEASK ANY VENDOR TO DEMO THE LEFT COLUMN
AI flags.Detectors produce typed evidence with a confidence score. Nothing is a verdict.
People decide.A reviewer accepts or dismisses each violation, then records one decision with a reason.
Candidates can appeal.Every decision is appealable, and the appeal is tracked to its own outcome.
01

Detection is not accusation

Each detection carries a confidence score and becomes a violation with a configurable risk weight. Weight accumulates into a band — low, medium, high, critical. A threshold opens a review case. Nothing is automatic past that point.

Bring one paper. We'll run it live.

Thirty minutes, your own exam, no slide deck.