Interhuman

An API and omni-modal model that detects social signals from video, audio, and text to capture how people communicate beyond transcripts. It solves the problem of AI systems missing behavioral cues such as hesitation, confusion, engagement, and frustration. It is for developers and product teams building AI applications such as sales coaching, tutoring, user research, and meeting copilots. It is delivered as an infrastructure API that returns detected signals with evidence-grounded rationales and confidence scores.

Key features

• Detect 12 social signals from video
• Multimodal perception across video audio text
• Temporal alignment of modalities
• Evidence-grounded rationales per signal
• Confidence scores per signal
• Detect engagement and disengagement
• Detect agreement and disagreement
• Detect confusion and uncertainty
• Detect frustration and stress
• Detect hesitation and skepticism
• Detect confidence and interest
• Playground to test API live