Interview System
Depth & Modes
How interview depth and mode (voice/text) work.
Interview Depth
| Depth | Spine Mode | Adaptive Mode |
|---|---|---|
| Light | 4 questions | 4 min competencies, ~8 total questions |
| Standard | 6 questions | 6 min competencies, ~14 total questions |
| Deep | 8 questions | 8 min competencies, ~20 total questions |
| Custom | N questions | N min competencies |
In adaptive mode, depth maps to a question budget (including probes), not just core questions. The AI manages within this budget.
Target durations (configurable):
- Light: ~15 minutes
- Standard: ~25 minutes
- Deep: ~40 minutes
Interview Modes
Voice
- Retell AI processes speech in real-time via GPT-4.1
- Requires desktop + camera + microphone
- Integrity monitoring includes face detection + gaze tracking
- Idle auto-end after configurable silence period
- Per-minute cost to Retell
Text
- Retell Chat API with same GPT-4.1 model
- Works on any device including mobile
- Integrity monitoring is browser-only (tab switching, etc.)
- No idle auto-end (no per-minute cost)
- Same scoring pipeline as voice
Both modes use the same AI model and scoring. The difference is input method only.