Patent Pending · US Provisionals 64/066,231 · 64/075,009 · 64/077,286

The scorer’s seat above AI incident data.

FICO never owned the bureaus’ data. It owned the predictive function above it. Unisapience is built to occupy the same seat for AI risk: an independent scoring instrument plus the open interoperability rail that makes every tool’s output comparable.

Request the data room See the product view
The thesis

Own the function, not the data.

Incident corpora already exist and will multiply. Competing to aggregate them is a race to a commodity. The defensible position is the layer that turns them into a decision — a calibrated, explainable, non-compensatory score with a frozen interface, sold on demonstrated lift.

Scorer, not aggregator

We sit above the corpora. Data sources are inputs; the score is the product. The FICO pattern applied where it has not been applied yet.

Instrument + rail compound

A strong instrument without a standard stays fragmented. A standard without a credible instrument carries no weight. Building both is the wedge and the moat.

Loss-bearer pays

Buyers, boards, insurers, and GRC officers pay for assessment. The published ranking is never for sale — the independence that makes the score worth paying for.

Two surfaces

One institution. Two surfaces that compound.

Instrument

Shadow Simplex Score

Deployment-level risk score with reason codes and decision-grade explanation. Non-compensatory aggregation: any single critical failure collapses the composite. Scores the system, not the paperwork.

Rail

SREL — open standard

Coordinate-based grammar for risk and control exchange. Heterogeneous tools present as one coherent exposure surface. Open spec (CC0), open implementations (Apache 2.0).

Why now

Regulatory and insurance pressure is already binding.

Regulation binds

EU AI Act Article 9 and NIST AI RMF create obligations now. SEC disclosure rules add parallel pressure. Boards need defensible structure.

Insurers need profiles

Underwriters are looking for credible, furnished profiles to price AI exposure. Better measurement is a precondition for a real market.

Format window

No common format for AI risk exists yet. First-mover on the rail compounds — it is the part that closes fastest.

Current state

Built vs. still earning.

We will not tell you the score is validated before it is.

The rail, coordinate space, diagnostic gate, and public index structure are built. What we are still earning is calibrated predictive validation: proxy outcomes first, real-incident calibration as furnishing volume grows. Diligence materials, including the validation worklist and known gaps, are available under NDA.

Built
  • SREL rail · coordinate space · grammar
  • Self-corpus diagnostic gate
  • Public index as participation engine
  • Reason-code spine + explanation structure
Earning
  • Proxy validation → calibrated score
  • Non-incident denominator for calibration
  • Hosted scoring API (preview)
Business model

Two surfaces, three revenue lines.

Adoption surface

Self-run diagnostic

Unsigned, self-run diagnostic gate for DevSecOps pipelines. Free. Creates the adoption path for the rail.

Revenue · metered

Attested signed scores

Metered, signed, attestation-grade scoring against the full corpus. Paid by the loss-bearer (regulator, insurer, counterparty).

Revenue · services

Scoring & assessment

Hosted scoring and enterprise assessment engagements for M&A diligence and governance programs.

IP position

Four tiers of disclosure, by design.

The geometric substrate is published for scrutiny. Calibration weights and the scoring corpus remain trade secret. Provisionals establish priority; they are not issued patents.

  • Filed — SSPLX-001/002/003 provisionals (May 2026)
  • Public — Substrate theory (DOI available)
  • Trade secret — Calibration weights · scoring corpus
  • US 64/066,231 · substrate + composite
  • US 64/075,009 · 9×6 matrix + SREL/AI-1
  • US 64/077,286 · cascade, polarity, controls
What exists today

Built, not slideware.

7
API endpoints documented (preview)
2
Reference SDKs (Python, TypeScript)
2
Emission formats (SARIF + SREL)
CC0
Open spec + Apache 2.0 reference

Diligence-ready, under NDA.

Extended methodology, validation worklist, filing documents, and the roadmap — including what is not solved yet — are available to qualified parties.

Request the data room Contact the founder

This page is informational and does not constitute an offer to sell or a solicitation of an offer to buy securities.