Interhuman AI debuts model for reading non-verbal social cues

Danish startup Interhuman AI has launched Inter-2, an AI model that reads non-verbal social signals—the stuff you convey with facial expressions, voice, and body language, not words. Unlike most AI, Inter-2 analyzes real-time video, audio, and text to spot 12 signals like hesitation, engagement, and disagreement. The company claims it runs up to four times faster than previous models, which could make this social detection practical for live interactions.

• Inter-2 interprets facial and vocal cues in real time
• Detects 12 types of social signals, like uncertainty
• Works across video, audio, and text input
• Founders come from pharma, behavioral science, and tech backgrounds
• Claims fourfold speed improvement for real-time use