Emotional AI

EchoDepth Insight

Facial emotion analysis using FACS and VAD scoring. Captures what people feel but don't say.

How it works

Tracks 44 facial Action Units via webcam, computing Valence, Arousal and Dominance scores in real time during research sessions.

Use cases

  • Market research & concept testing
  • Pharmaceutical patient interviews
  • FMCG packaging & advertising
  • Employee culture measurement
Explore Insight →
Deception Detection

EchoDepth Signal

Linguistic deception detection using forensic statement analysis. Captures what people write but isn't true.

How it works

Analyses written statements for linguistic patterns associated with deception, scoring credibility and flagging specific indicators.

Use cases

  • Insurance claims verification
  • Recruitment & candidate screening
  • Workplace investigations
  • Legal evidence analysis
Explore Signal →

Shared Foundation

One thesis. Two detection methods.

EchoDepth is two products built on one thesis: that what people say is a filtered version of what they think, and that the filter is measurable. EchoDepth Insight measures the emotional response itself, scoring 44 facial Action Units in the browser and mapping them to 47 discrete emotional states with Valence-Arousal-Dominance profiles — used with consenting research participants in concept tests, pharma research and product testing, where it captures reaction before a participant has put it into words. EchoDepth Signal works on the words themselves, analysing written statements for more than thirty forensic linguistic indicators and returning a credibility score with flagged passages — used in workplace investigations, claims handling, recruitment screening and legal review. Insight is for research settings; Signal is the workplace product, and it is text-based by design, because inferring emotion from biometric data at work is prohibited under the EU AI Act.

Both products are built on the same insight: people reveal more than they intend through involuntary signals. The face cannot fully mask emotion. Language cannot fully mask deception.

🎭

Beyond self-report

Traditional methods ask people what they think. EchoDepth captures what they cannot consciously control.

🔬

Science-backed

Insight uses FACS (Ekman & Friesen). Signal uses forensic linguistic analysis. Both grounded in peer-reviewed research.

📊

Actionable output

Not raw data. Structured reports with scores, flags, and recommendations your team can act on immediately.

See what EchoDepth reveals in your context.

Whether you're verifying claims, screening candidates, or understanding emotional response, we can show you what traditional methods miss.

Request a Demo About Cavefish