Interactive Demo
Try the Recall Engine
See how 9-factor ranking works on sample memories. This is a simulated demo using the same scoring algorithm as the real engine — running entirely in your browser, no server required.
Scoring Formula
0.45×similarity + 0.08×recency + 0.05×frequency + 0.10×typeBoost + 0.08×scopeBoost + 0.07×confidence + 0.07×reinforcement + 0.05×tagAffinity + 0.05×graphProximitySuperseded memories receive a ×0.3 penaltyActive conflicts receive a ×0.8 penaltyRecency uses 7-day exponential half-life
Loading playground...
!
Note: Similarity scoring here uses keyword overlap, not real cosine similarity on 768-dimensional vectors. In the actual engine, semantic similarity is computed via nomic-embed-text embeddings and Memgraph vector search.