Firasa reads the unseen from the seen.

Firasa runs deeper than discernment. It is the word for someone read as wise and perceptive, the person who sees what others miss. It also happens to describe measurement done well. Everything I build under this name comes back to one move: take something that resists being measured, and infer it from behaviour you can actually observe.

Code quality is one example. The human qualities of a conversation are another. They look like different problems. Underneath, they are the same one, and the same scoring engine sits beneath both. The instruments use language models as the scorers, kept honest by psychometric validation and deterministic rules, so the same input gives the same number every time and every score can be traced back to why it was given.

The discipline for this already exists. Psychology spent a century working out how to measure things you cannot see directly, with reliability, validity, and a healthy distrust of any single number. AI mostly skips that step. Firasa is what the work looks like when you don't.