Score any forecaster. Human or AI.
AEQUARA is one company with one thesis — calibrated, verifiable, overridable judgment — and two co-equal sides. Individuals use it before a decision. Institutions license the same engine to measure how calibrated any forecaster is, on the record. Not finance-specific. Domain-general by construction.
The engine behind every verdict, exposed as infrastructure.
A consumer asking “am I about to make a mistake?” and an institution asking “how calibrated is this forecaster?” run the same three steps. That shared spine is the company.
Take a forecast
A verdict, a price, a probability, a model output — anything that makes a claim about the future, from a human or an AI, in any domain.
Resolve against ground truth
Score it on a stated horizon when reality lands. No grading on a curve — we grade on a horizon, and we publish when we were wrong.
Emit a calibration score
A Brier-tracked calibration band with an audit trail and a methodology hash — portable, verifiable, and defensible to a committee.
Same ingest → resolve → Brier engine. One side returns a verdict card. The other returns an API and an index.
Verified judgment, licensed.
Each inherits the same NEXUS substrate, the same anti-issuer-pay firewall, and the same Brier-tracking discipline. Built, not boasted.
Calibration Attestation
Tamper-evident calibration evidence computed over your own logged decisions — a Merkle-chained evidence file, machine-readable attestation, and an SR 26-2-mapped memo your validators can re-verify independently. No SDK, no integration.
NEXUS
The multi-LLM operating substrate — routes across Claude, GPT, DeepSeek, Gemini and Groq with Brier-calibrated dispatch and full override. It is the engine designed to publish NPX-100.
NEXUS BY BRIER
Brier-tracked model outputs with a full audit trail and methodology access — verified judgment your risk committee can defend. Core to Sovereign.
NPX-100
The 100-constituent calibrated index — operational band 97–105, designed for quarterly publication, methodology-hash stable, anti-issuer-pay, Brier-tracked with errors published.
AXISTRUTH
The quantitative substrate as an API — HRP allocation, fractional Kelly, four-factor scoring, Monte-Carlo stress. The same math that underlies NPX-100, usage-metered.
MENSURA
Institutional crypto indices — seven index families, 54 benchmarks, each with a specified methodology and methodology hash. Licensing for funds, ETPs and structured products.
Multimodal Calibration Index
The same Brier-tracked judge that scores text forecasters, extended to image, video and audio generators — one unified rubric, resolved against ground truth, anti-issuer-pay. The engine is built; the public index is accumulating.
Provenance & Safety
A conservative, tri-state scorecard of which generative models carry content credentials (C2PA), durable watermarks and safety filtering — so a compliance team can enforce a provenance policy before a model ships. Unknowns fail closed; nothing is asserted without a public source.
Trustee tier
For custodians, family offices, allocators and methodology stewards who use AI where being wrong has consequences — calibration receipts defensible to a committee, regulator, or journalist.
Built for individuals. Measured in public.
The consumer product
Tell us the decision you’re facing; get a verdict with a calibrated confidence score, the reasoning, cited sources, and a one-click override.
The platform
License the calibration engine: routing, Brier-tracked outputs, the published Model Calibration Index, and indices built on the same discipline.
Measure the judgment you depend on.
Human or AI, in any domain — if it makes a claim about the future, we can score how calibrated it is. On the record.
Talk to a steward