Aligns caller-labeled dimensions, computes cosine similarity and a normalized L1 difference, and flags drift against a supplied cosine threshold. It does not infer goals from behavior, decide which goal is correct or detect deception; results depend entirely on the supplied vector representation.

Payment Options