Legal AI OS · prediction engine

The master reference

Everything about the Fault-Line Radar in one place — what it forecasts, how it scores, how evidence gets in, and where the pressure is building right now.

Part 1

What it is

The radar forecasts where legal AI is headed by tracking fault lines — the places a new AI capability rubs against an existing legal duty. It does not predict dates. It measures pressure, weighted by who is doing the pushing.

Part A

Forecast core

Will a court, bar, or law move on this?
Graded against what actually happened. This is the part that earns trust by being scored, not by sounding sure.
Part B

Advisory layer

What should a firm build and buy next?
Turns Part A's readouts into build-and-buy advice. Evidence-traced, never framed as a prediction.
Part C

Firm simulation

Can this firm afford to act?
A what-if model of one firm's economics. Its headline: on hourly billing, more AI is a net loss; on fixed fee, it's a net win.
The supply stack — can AI do it, can vendors ship it, can a firm adopt it
L1 capabilitycan the AI do the thing yet
S softwareare legal-AI vendors enabled to build it
E enablementcan a firm stand the fix up with what the market sells
The governance side — will the law move, and will the fix become table stakes
L2 rulingwill a court, bar, or statute move on it
L3 adoptionis the fix becoming table stakes, court or no court
Part 2

The eleven fault lines

Five meters each, 0–10. A fault line "fires" when it crosses 7. Sorted by how much pressure is building.

Fault lineThe fix (control)CapRuleAdoptEnableSoft
verificationDocumented verification + trace logs8.410.08.69.05.6
competenceFirm-wide training + governance program6.69.98.08.45.0
disclosureVerification-by-default with an audit trail5.59.97.77.45.5
convergenceOne operating model, strictest standard2.59.87.74.05.5
insuranceGovernance as an insurable artifact2.58.78.46.94.0
confidentialityData-flow mapping + vendor attestation4.59.97.06.75.5
benchmarkTool-certification benchmark (NERVE)9.09.17.68.97.5
agenticHuman-decides gates + scope-limiting6.99.46.26.58.3
vendor liabilityTool-provenance documentation2.59.12.53.03.8
judge analyticsGuardrailed strategy sim, no actor prediction4.06.43.03.03.0
feesValue-delivered measurement2.55.83.03.53.0
Part 3

How it scores

Authority over volume. Every piece of evidence is stamped with a tier when it enters, and the tier decides the weight — not how many people repeated it.

TierWhat it isWeight
T1Court ruling, statute1.00
T2Bar opinion, regulatory guidance0.80
T3Empirical study, institution0.60
T4Commentary, trade press0.35
T5Vendor marketing0.008

A vendor blog carries about a hundredth of a bar opinion. Ten vendor posts cannot outweigh one ruling. Two things keep the scoring honest: the L2 ruling meter is two-sided (an item can argue against a line and drag it down, not just up), and every capital event is flagged continuation / origination / direction-shift — the momentum signal only predicts continuation, so a first-of-kind move is labeled rather than scored as if it were predictable.

Part 4

How evidence gets in

A fact does not enter the feed because someone found it. It enters because it survived being argued against.

A fact is found on the internet
Split the question, search it from several directions at once — including one angle that looks for where the story is wrong
Three independent reviewers each try to prove it false
Does it survive 2 of 3 skeptics?
Admitted and ranked · or killed — and if a date or amount cannot be verified, the honest answer is "we found nothing," not a guessed number

This is the step that caught real mistakes: a court rule only one source mentioned, a "disqualification" that was actually a reprimand, a claim about two vendors no source could hold up.

Part 5

The streams

Seventy-six dated, cited items flow from four buckets into eleven fault-line predictions. The number beside each prediction is how many streams corroborate it — more streams, more confident call. Full interactive version: the streams page.

Part 6

What to watch next

The open signal

No insurer has yet made a certified AI tool a condition of malpractice coverage, and no bar has adopted a named benchmark. When one does, that is the event that flips the method layer from voluntary to enforceable. That absence — the thing that has not happened yet — is the single most valuable signal in the whole system.