How is Burna AI's grading engine built?

The grading engine is twelve specialized agents operating in a cascading constraint pipeline. Each agent's output bounds the valid output space of every downstream agent: a grade cannot be produced that contradicts the system's own prior findings. The pipeline spans extraction, verification, resolution, standardisation, taxonomic matching, differential analysis, temporal tracking, documentation integrity, evidence synthesis, causal attribution, confidence calibration, and regulatory encoding. Citation is structural: the engine cannot produce a grade without naming the specific CTCAE criterion and the source clinical text. Two patents filed.