Where AIOOJ fits
The legal AI market has powerful tools for research, drafting, and document review. What it lacks is a model that converts all of that into a single, stress-testable answer about one specific case. That is the gap AIOOJ fills.
The legal AI stack in 2026
Legal AI has expanded rapidly. The global legal AI market reached $3 billion in 2025 and is projected to grow at 28% annually to $7.1 billion by 2032. Corporate legal adoption more than doubled in a single year — from 23% to 52% between 2024 and 2025. By early 2026, Thomson Reuters CoCounsel had reached one million professional users.
But the market has grown in one direction: broader, faster, and more automated at the general layer. What has not been built is a layer that takes the output of all of that work and converts it into a structured, calibrated, stress-testable prediction for a single specific case. That is the structural gap. It is architectural — not a product gap incumbents will close next quarter.
What existing services cannot do
Stanford HAI research found error rates of 17% for Lexis+ AI and 34% for Westlaw AI-Assisted Research on legal queries. Over 700 court cases worldwide have now involved AI hallucinations, with sanctions ranging to $31,100 per incident. The error rates are structural: asking general platforms to produce a specific, calibrated, defensible probability estimate for a single case is not what they are designed to do.
General legal AI — useful but unanchored
Ask Lexis+ AI or CoCounsel "how strong is this case?" and you receive a well-researched, generally accurate analysis of the legal landscape. What you do not receive is a probability. You do not receive a settlement floor. You do not receive a sensitivity analysis showing which inputs matter most. You receive information — not a decision-support instrument.
AIOOJ — anchored, calibrated, stress-testable
AIOOJ takes the legal analysis that platforms like CoCounsel and Lexis+ AI help produce, anchors it to the specific facts of the matter, converts it into named probability inputs (P1–P9), and produces a defensible probability distribution with a settlement floor, a stress-tested downside, a driver ranking, and a rational settlement corridor — all updating in real time.
| Capability | General legal AI | Document platforms | AIOOJ |
|---|---|---|---|
| Legal research & case law | ✓ Strong | ✗ Not designed for this | → Consumed as input |
| Document review & e-discovery | ∼ Partial | ✓ Strong | → Consumed as input |
| Drafting & document generation | ✓ Strong | ∼ Partial | ✗ Not in scope |
| Named probability estimate for a specific outcome | ✗ No | ✗ No | ✓ Core function |
| Sensitivity analysis — which inputs matter most | ✗ No | ✗ No | ✓ Shapley + P2×P4 heat map |
| Stress-tested downside scenarios | ✗ No | ✗ No | ✓ Scenario A, B + P2×P4 |
| Settlement band distribution | ✗ No | ✗ No | ✓ Five bands, live probability |
| Calibration to empirical case law anchors | ∼ General benchmark | ✗ No | ✓ 14 real anchor cases, 60/40 blend |
| Rational settlement corridor with PV | ✗ No | ✗ No | ✓ PV floor to costs-adj ceiling |
| Real-time recalculation on input change | ∼ Query-response only | ✗ No | ✓ Instant on every slider move |
What makes AIOOJ different
Named assumptions
Every probability is named (P1 through P9), described, and traceable to a specific legal issue. General platforms produce outputs whose reasoning is often opaque.
Anti-overfitting discipline
Eight deliberate score reductions were applied during model development. Hard caps prevent artificial inflation. A 60/40 empirical calibration blend anchors each input to real case data.
Stress testing architecture
The P2×P4 heat map, the Stress Floor scenarios, the Shapley decomposition, and the sensitivity analysis all exist to find where the case breaks under attack — and how badly. This adversarial discipline is entirely absent from general legal AI platforms.
Empirical calibration
14 verified real anchor cases from official HCA/NSWCA sources. 100-case dataset with weighted calibration. Bootstrap confidence intervals from real cases only. NSW court statistics. FCA empirical data on time-to-trial.
Settlement-ready outputs
The model produces numbers counsel and Corrs can use directly in mediation: the P10 settlement floor, the rational corridor, the costs-adjusted total exposure, the branch-weighted PV.
Conduct-adjusted modelling
The defendant's pre-litigation conduct — illegal NDA conditions, systematic data withholding, story-shifting, corrective payment after denials — is explicitly incorporated as named variables.
Who should use AIOOJ
AIOOJ is not a replacement for Lexis+ AI or CoCounsel. It is the layer above them — the tool you use after you have done the research and analysis, to convert that work into a defensible probability estimate for settlement negotiations, counsel briefings, and litigation funding discussions.
- Litigants in person and self-represented parties facing complex commercial disputes where legal costs make comprehensive external advice prohibitive. This is Harrison v Aegon: the model was built precisely for this use case.
- Instructed solicitors (Corrs Chambers Westgarth and equivalent firms) who need a defensible, documented basis for settlement positioning and litigation sequencing advice.
- Briefed counsel preparing for mediation or settlement conferences. The P2×P4 stress table, the Shapley driver ranking, and the stress floor scenarios are specifically designed for counsel briefing.
- Litigation funders assessing whether a case merits third-party funding. The model provides the probability distribution, the expected value, the sensitivity analysis, and the downside floor in a format that funding analysts can interrogate directly.
- In-house counsel and corporate legal departments managing complex commercial litigation where settlement decisions need to be documented and defensible at board level.
The market signal AIOOJ is responding to
No major player currently occupies the case-specific outcome intelligence layer. Litigation prediction is listed as a market segment in analyst reports, but the dominant players remain focused on the research and document layers where the volume is higher and the technical requirements are lower. This is AIOOJ's structural opportunity.
What AIOOJ is not
It does not retrieve case law, search databases, or generate legal analysis. It consumes the output of those tools as inputs and converts that analysis into probability estimates.
The probability outputs are modelling estimates — calibrated expert judgements expressed quantitatively. They are not statistical frequencies from a large sample of identical cases. There is only one trial.
The model structures and quantifies legal analysis. It does not replace the analysis itself. The quality of outputs depends entirely on the quality of the legal judgements entered as inputs.
Every number is derived from a named formula, a documented assumption, or an empirical anchor. There is no generative AI component. The model does arithmetic, logic, and statistics — not language generation.
The current model is built specifically for Harrison v Aegon in the NSW Supreme Court Commercial List. The architecture is transferable — the specific inputs, calibration, and legal constructions are not.
"General legal AI tells you what the law says. AIOOJ tells you what is likely to happen — in this case, with these facts, against this defendant, in this court."