Common-benefit discovery, reasoned once.

In an MDL, the evidence that wins is largely common: the same corporate memos, regulatory filings, and adverse-event reports establish notice and general causation for every plaintiff. When the Mass Tort edition opens, Justine will read the common-benefit document depository in a single 10M-token context and surface that shared proof once — the date the defendant first knew, the buried safety signal — then reuse it across thousands of plaintiff sub-agents rather than re-deriving it file by file. The general-causation record becomes a pool asset, not a per-plaintiff cost.

What MT is designed to do.

1. 01

   Justine is designed to ingest the common-benefit depository — internal memos, regulatory submissions, adverse-event reports, deposition transcripts — and reason across it in one context rather than as fragmented per-custodian reviews.

2. 02

   It will surface the documents that establish corporate notice and general causation for the whole pool, and propagate those findings to every plaintiff sub-agent so the proof is shared, not rediscovered.

3. 03

   For Daubert, the sub-agent is designed to assemble the general-causation record — epidemiology, dose-response, mechanism — against each defense expert’s prior testimony and methodological gaps as a pre-read.

4. 04

   Every authority Justine cites will be grounded against a verifiable source via CourtListener; common-issue analysis carries provenance back to the depository page that supports it, never invented citations.

5. 05

   The common-issue work product is attorney-supervised: Justine reasons the shared record and shows its sourcing, and common-benefit counsel decides what is argued, attests it, and owns it under privilege.

The AI reasons; the attorney decides.