A model-agnostic reasoning engine for defensible legal intelligence.

Traditional contract AI extracts, classifies, and summarizes clauses. POLOXI asks what risks the contract creates for this organization and reasons across interacting provisions.

Unlimited indemnification
        +
Low vendor liability cap
        +
Customer data obligations
        +
Weak security commitment
        ↓
Combined organizational risk

Candidate assessments:
H1 → Standard acceptable agreement
H2 → Material financial exposure
H3 → Significant data/security exposure
H4 → Unbalanced indemnification
H5 → Problematic termination structure

Traditional eDiscovery may return millions of potentially relevant documents. POLOXI investigates which evidence matters to the legal issue.

“Did company executives know about the defect before the product launch?”

2,400,000 documents

H1 → Executives knew
H2 → Engineering knew; executives did not
H3 → Problem known; severity misunderstood
H4 → Defect discovered after launch
H5 → Evidence inconclusive

Investigate:
Email · Collaboration messages · Reports · Minutes
Issue trackers · Engineering tickets
Executive presentations · Launch records

Discovery becomes an evidence investigation rather than document retrieval.

POLOXI reconstructs who did what, when, to whom, using which document, under which obligation, and with what consequence.

Chronology
    ↓
Contract obligations
    ↓
Legal significance
    ↓
Evidence strength
    ↓
Distinguish background events
from potentially dispositive events

Organizations face overlapping federal, state, industry, contractual, licensing, privacy, security, retention, and reporting obligations.

“Customer information may have been sent to the wrong recipient. What obligations could this trigger?”

H1 → Reportable breach
H2 → Non-reportable incident
H3 → Contractual notification only
H4 → Regulatory notification
H5 → Customer notification
H6 → Multiple obligations

Evidence:
Data type · Jurisdiction · Number affected · Encryption
Recipient · Access evidence · Contract · Regulation · Policy

Missing evidence: Whether the unintended recipient accessed or downloaded the data. This is compliance reasoning, not a static checklist.

A data room may contain thousands of contracts, corporate records, litigation matters, licenses, IP records, employment agreements, insurance, regulatory correspondence, privacy documentation, and financial obligations.

“What could materially affect this transaction?”

H1 → Change-of-control restrictions
H2 → Undisclosed litigation exposure
H3 → IP ownership problem
H4 → Regulatory exposure
H5 → Key-contract termination risk
H6 → Employment liability
H7 → Data/privacy exposure

Discovery:
Customer A = 28% of revenue
Contract = Change-of-control termination right
Acquisition = Triggers change of control

Traditional research asks for cases involving an issue. POLOXI asks which authorities support or weaken an argument given the specific facts.

Argument: Contract provision is unenforceable

H1 → Provision enforceable
H2 → Provision unenforceable
H3 → Enforceable but limited
H4 → Factual determination required

For every authority evaluate:
Jurisdiction · Court level · Precedential value
Legal issue · Factual similarity · Distinguishing facts
Subsequent treatment · Current validity

Case C involved negotiated commercial terms between sophisticated parties. The current matter involves a standard-form agreement and materially different bargaining circumstances.

This transforms legal search into argument intelligence.