Detects mental-health spirals before they deepen: each memory is stored in Qdrant as six named vectors (semantic, emotion, prosody and more), a custom transformer reads 14-day emotional sequences, and when trouble looms it retrieves the user's own past recovery as proof they have climbed out before.
The strength here is that retrieval is not the endpoint. When trouble is coming, what surfaces is not generic advice but the user's own record of having climbed out before, which is a genuinely different answer to what a vector database is for. Storing each memory as six named vectors puts a specific Qdrant capability at the centre of the design, and reading fourteen-day sequences follows the same logic, since a spiral is a trend.
Summaries are written by this site. Project pages include demo videos, screenshots and the team's own write-up (videos may autoplay).