This dashboard is case-level because the public demo currently has one flagship case. The time-saved estimate includes overnight AI opportunity screening plus downstream document review, routing, human review, and BD/Ops decision support.
Processing path reflected in these KPIs
This run used the large normalized-packet path: AI screened the opportunity, inventoried the package, reviewed page-aware chunks, created a master packet, then routed specialists from that normalized record.
Time per workflow stage
Seconds for AI screening, large-package normalization, extraction, and routing; human review and BD/Ops decision remain the longest stages.
Evidence by department
Distribution of extracted facts routed
facts
- Legal 4
- Security 5
- Finance 2
- Implementation 3
Audit trail activity
Each bar marks workflow audit records created by the system, AI agents, human reviewers, or BD/Ops. Click a bar to see the records for that moment.
- AI intake - Government PDFs extracted and normalized into a real-case packet.
- CodexReviewAgent - 16 findings generated and routed.
- Legal reviewer - Legal returned worth_pursuing_with_conditions.
- Security reviewer - Security returned worth_pursuing_with_conditions.
- Finance reviewer - Finance returned worth_pursuing_with_conditions.
- Implementation reviewer - Implementation returned worth_pursuing_with_conditions.
- AI synthesis - AI recommended pursue_with_conditions.
- BD/Ops Lead - BD/Ops logged final decision: pursue_with_conditions.