URIM

Decision intelligence for complex disputes

Know what the case depends on.

URIM is being developed to turn a live matter into a reviewable outcome range, the sequence of events that can change it, and a brief built to be challenged.

Request a portfolio briefing Examine an illustrative brief

For litigation partners, in-house disputes teams, claims leaders, and litigation-finance underwriters.

Decision brief · illustrative

Commercial agreement dispute

URM-24-071 v0.8
Decision under review

Proceed through the jurisdiction motion or reopen settlement?

Current assessment 68% Claim succeeds · illustrative

61–75% reviewed range

Largest swing factor
Jurisdiction motion · ±14 pts
Next material event
Motion decision
Review posture
Human review required
Model, data, and sources versioned Not legal advice

Built for the recommendation behind the decision.

  • Litigation partner
  • Chief litigation counsel
  • Claims leader
  • Underwriter

01 · The decision

A second opinion you can interrogate.

A probability without its dependencies is difficult to use. URIM is designed to show what supports the range, what could move it, and what counsel should verify before relying on it.

01

Screen

Decide where deeper diligence, capital, or senior attention is justified.

02

Advise

Make settlement, reserve, and strategy recommendations more explicit.

03

Monitor

Change the assessment when the evidence changes—not when the room gets louder.

02 · The brief

The path is more useful than the point.

Test an illustrative event below. The range changes because one defined dependency changed. A real brief would show the evidence, source, reviewer, and model version behind every movement.

Illustrative scenario

Updated assessment 68%

Current record. The jurisdiction motion remains the largest unresolved swing factor.

Critical path

Commercial agreement dispute · URM-24-071

Illustrative
  1. 01
    Agreement and forum record Comparator set established
    Observed
  2. 02
    Jurisdiction motion Pending · ±14 point sensitivity
    Open
  3. 03
    Expert evidence admitted Required for damages theory
    Conditional
  4. 04
    Adverse precedent distinguished Secondary sensitivity · ±7 points
    Conditional
  5. 05
    Claim path remains viable Reviewed range · 61–75%
    Review
What changed

No event has been resolved. The displayed range remains conditional.

03 · The method

Designed to be reviewed, not believed.

The product should make its reasoning inspectable at the same level as its conclusion. Every layer has a named input, a bounded output, and a human responsibility.

01

Matter record

Posture, decision, known facts, missing evidence, and declared assumptions.

Human confirms scope
02

Comparator set

Relevant decided matters, inclusion logic, exclusions, and source provenance.

Counsel can challenge fit
03

Specialist panel

Bounded analytical models contribute estimates, uncertainty, and dissent.

Agreement is not concealed
04

Calibrated range

A distribution and sensitivity view, not a categorical forecast.

False precision is rejected
05

Critical path

The ordered events that strengthen, weaken, or break the current thesis.

Counsel owns the decision

04 · Reviewability

Every conclusion carries its file.

URIM is intended to make the basis of a recommendation portable: into the partner meeting, client call, claims file, or investment committee.

Sources

Authorities and comparator records linked to the proposition they support.

Assumptions

Material assumptions separated from observed facts and marked for review.

Sensitivity

The events with enough leverage to change the current recommendation.

Dissent

Material disagreement preserved instead of averaged into invisibility.

Version

Model, data, analyst review, and revision reason recorded together.

05 · Evidence standard

No performance theatre.

Public validation will be dated, versioned, scoped to a matter class and jurisdiction, and linked to method. Until that exists, URIM will not publish a corpus count, accuracy claim, win-rate claim, or simulated live feed as evidence.

  • Temporal testingEvaluate on decisions the system could not have seen.
  • Calibration firstTest whether stated confidence corresponds to observed outcomes.
  • Failure inventoryPublish exclusions, abstentions, missing data, and material failure modes.
  • Change recordRecord what changed, when, why, and which briefs may be affected.

06 · Human control

The instrument informs judgment. It never exercises it.

Probabilistic analysis for legal professionals. Not legal advice.

07 · Founding research

Request a portfolio briefing.

We are speaking with senior professionals who make or support repeated, high-value litigation decisions. Share your role and broad portfolio context—not matter facts, party names, documents, or privileged information.

A useful conversation may cover:

  • where uncertainty is most expensive in your workflow;
  • what a decision brief would need to show to be credible; and
  • which validation, governance, and security gates matter first.