bmad-distillator

You can compress documents using lossless distillation, which preserves every fact, decision, and relationship into a token-efficient distillate. This maximizes signal-per-token for downstream LLM workflows without discarding critical information.

The best way is using semantic splitting to create a root distillate plus section distillates with cross-references. This handles mixed source types and produces a single consumable artifact for downstream pipelines like PRD generation.

Yes, you can distill product briefs and discovery notes into a single consumable artifact for PRD pipelines. The process compresses mixed source types into a token-efficient distillate ready for downstream consumption.

Yes, document distillation supports multi-part outputs with cross-references. It uses semantic splitting to divide content into a root distillate and section distillates, maintaining structural relationships across the entire distillate.

You improve token efficiency by applying lossless compression to large documents, which extracts essential information while preserving facts and constraints. The resulting distillate includes a token estimate to manage downstream LLM context windows.