Grably

Grably is a multi-modal human interaction data research lab that captures and models multi-signal human activity and decision-making processes, combining physical motion, physiological signals, gaze/vision, speech, intuition and contextual cues. It serves AI labs and research teams working on multimodal learning, cognitive modeling, human-AI collaboration, and decision-making analysis to enable AI that acts as a real teammate.

Grably defines critical aspects of human activity, attention, and decision-making that today’s AI cannot yet model. The organization emphasizes datasets and methods designed to help AI systems understand human behavior and interaction in real-world contexts.

Grably provides access to datasets described as responsible for advancing model cooperation with users, multimodal alignment, and conversational grounding. It emphasizes delivery timelines and diversified data types like video, text, and other modalities to support various AI research needs.

Grably offers custom dataset design for clients or collaborators with niche research goals, implying flexibility in research collaborations and dataset development to suit specific study aims.

Use cases

• Supports model cooperation with users and multimodal alignment