One supervisor over thousands of plaintiff sub-agents.

Mass tort breaks the case-by-case tools that run on personal injury. JustineAI™ will answer it with the supervisor pattern at inventory scale: one Justine, holding thousands of unbranded plaintiff sub-agents inside a single 10M-token context on the Eve-Legal F5/reasoner core. The supervisor is designed to advance the whole pool through the MDL workflow while every plaintiff sub-agent keeps its own facts, posture, and provenance. The conversational surface will query the supervisor; the inventory-wide reasoning is the substance.

What MT is designed to do.

1. 01

   The supervisor will own inventory-level workflow state and dispatch one scoped plaintiff sub-agent per claimant, each carrying that plaintiff’s exposure proof, injury cohort, jurisdiction, and lien posture.

2. 02

   Because all sub-agents resolve inside one 10M-token context, Justine is designed to reason across the whole pool at once rather than re-loading each file as an isolated matter.

3. 03

   Cross-plaintiff consistency will run continuously: a causation theory argued for one cohort that contradicts a fact-sheet allegation elsewhere surfaces for the attorney, not after distribution.

4. 04

   Common findings reasoned once — corporate notice, general causation, a CMO term — are designed to propagate to every plaintiff sub-agent without being re-derived claimant by claimant.

5. 05

   Justine remains the sole named Digital Employee; the plaintiff sub-agents are unbranded facets, and the audit trail will record supervisor sequencing and sub-agent actions as distinct, attorney-attestable events.

The AI reasons; the attorney decides.