pipeline-judge-dual

You can evaluate business plan investor readiness by running a dual-LLM judge to merge analytical and strategic evaluations, generating fiscal validation, scoring, and consolidated judge reports for investor-ready artifacts.

Dual-LLM evaluation runs parallel analytical and strategic assessments on business plan artifacts, merging them into a consolidated judge report that provides fiscal validation, completeness scoring, and actionable remediation feedback.

To generate feedback for business plan re-runs, process plan-final.md and vN artifacts through a dual-LLM judge to produce FEEDBACK-FOR-RERUN.json, which provides structured remediation guidance for improvement phases.

Business plan scoring requires markdown inputs such as plan-final.md and vN directory research artifacts, producing JSON reporting outputs including JUDGE-A.json, JUDGE-B.json, JUDGE-REPORT.json, and FEEDBACK-FOR-RERUN.json.

Yes, dual-LLM evaluation works with versioned project artifacts by sourcing inputs directly from the vN directory and research artifacts alongside plan-final.md to assess project versions comprehensively.

Using a dual-LLM judge instead of a single LLM approach provides parallel analytical and strategic evaluations merged into one report, offering broader perspective synthesis for validating investor readiness and business plan consistency.