TIER 01
ANTAEUS RESEARCH
$450/ MO
- The console: watchlist, dossiers, exhibit builder
- Full read API, all six axes
- Receipts included on every value
- 10,000 calls / month
- 1 seat
- Citation license: cite with attribution
PERSPICAZITY · INSTRUMENT 01 · ANTAEUS
Antaeus is an archive of payer behavior: the difference between what an insurer says and what it does, with the primary document attached. Access arrives as a working console, citation-ready exhibits, and a read API.
Not disclosed in member-facing materials.
0.62 INTERVAL [0.55 – 0.68] · CONFIDENCE 0.91 · GAMING_RESISTANCE A
✓ RECEIPT ON FILE
sha256 9f2c41d8a6e3b07c5d21f4a89e6c0d73b2a15f68c94d07e3a8b6f215c40be41a
“98.6%”, a self-published figure.
INSUFFICIENT_EVIDENCE
Below the confidence threshold, Antaeus returns no number.
SIGNALS ARE CORRELATIONAL. THEY DESCRIBE OBSERVED RECORDS AND DO NOT ESTABLISH CAUSE OR INTENT.
The console, the exhibits, the archive, and the API all read the same engine output. Pick the form the day’s work needs.
firm | plan | claim | market) · period stamp · a receipt object: source document, SHA-256 digest, retrieval date.The shape of a response, with placeholder values. Hover or focus an annotation to trace it into the payload.
GET /v1/payers/PAYER_A/axesSAMPLE RESPONSE · SYNTHETIC DATA
{
"payer": "PAYER_A",
"as_of": "2026-06-30",
"method_version": "1.4.2",
"attribution_wall": "ENFORCED",
"composite_score": null,
"axes": [
{
"axis": "independent_review_overturn_rate",
"attribution_level": "claim",
"value": 0.62,
"ci95": [0.55, 0.68],
"interval_method": "wilson_95",
"gaming_resistance": "A",
"trust_tier": "adjudicated",
"period": "2025-04..2026-03",
"receipts": ["rcp_01hq3f8m"]
},
{
"axis": "mlr_rebate_obligation_ratio",
"attribution_level": "firm",
"value": 0.083,
"ci95": null,
"gaming_resistance": "F",
"trust_tier": "self_reported_filing",
"period": "FY2025",
"receipts": ["rcp_01hq77k2"]
},
{
"axis": "first_pass_claims_accuracy",
"status": "REFUSED",
"reason": "INSUFFICIENT_EVIDENCE",
"min_evidence": {
"independent_streams_required": 2,
"independent_streams_observed": 1,
"records_required": 400,
"records_observed": 112
}
}
],
"receipts": [
{
"evidence_id": "rcp_01hq3f8m",
"source_doc": "state_independent_review_register_2026-03.csv",
"sha256": "9f2c41d8a6e3b07c5d21f4a89e6c0d73b2a15f68c94d07e3a8b6f215c40be41a",
"retrieved_at": "2026-03-14"
},
{
"evidence_id": "rcp_01hq77k2",
"source_doc": "annual_medical_loss_ratio_filing_fy2025.pdf",
"sha256": "3c8a90f217bd4e6a8f02c5d7194ab3e660d1f84b2c97a05e3d6b8f4a1c20de57",
"retrieved_at": "2026-06-02"
}
]
}
The wall, in the schema. Claim-behavior and firm-financial axes are attribution-walled and never composited: composite_score is null by type, not by policy.
A claim-level behavioral axis. Value, 95% CI (Wilson interval), gaming-resistance grade, trust tier. The full record travels together.
A firm-level financial axis. attribution_level "firm" keeps it on its side of the wall; where the source stream supports no interval, none is invented.
The refusal is structured. Below minimum evidence the API returns REFUSED with the shortfall itemized, never an estimate.
The receipts. Source document, SHA-256 digest, retrieval date. Every evidence_id resolves at /v1/receipts/{evidence_id}.
INTERFACE · VERSIONED ENDPOINTSv1
| GET | /v1/payers/{id} | Full record for one payer: all six axes, receipts attached. |
| GET | /v1/payers/{id}/axes/{axis} | One axis, with its interval, grades, and receipt object. |
| GET | /v1/receipts/{evidence_id} | The archived source-document record behind any value. |
| GET | /v1/comparability | Comparability classes: which axes may be read across payers, and under what caveats. |
Fiduciary review of a plan administrator against the observed record rather than the marketed one. Exhibit cards drop into committee minutes and counsel memos with the citation attached.
Vendor review and renewal files that quote a government filing, not a proprietary benchmark with restricted use terms.
Ground routing and prioritization in observed adjudication signals, and cite the primary document, not a vendor’s summary of it.
Replace self-reported payer feeds with signals that carry their own provenance into your product.
Versioned method, archived receipts, reproducible retrieval. The citation chain survives peer review.
Every signal resolves to a primary source your standards desk can independently retrieve and verify.
Every tier reads the same engine output: same axes, same intervals, same refusals. No tier buys a composite, and no tier buys a softer number.
TIER 01
$450/ MO
TIER 02
$1,850/ MO
TIER 03
Custom
REQUESTS ARE REVIEWED MANUALLY. NOTHING IS SELF-SERVE, INCLUDING PAYMENT.
Antaeus is the first instrument built on the Perspicazity engine. The evidence model, its weights, and every version of the method are published on the engine’s record: receipts or silence, correlation only, never composited, refusal below threshold.
THE NAME IS THE METHOD. ANTAEUS: THE GIANT WHOSE STRENGTH HELD WHILE HE KEPT CONTACT WITH THE EARTH. THE INSTRUMENT STAYS GROUNDED IN PRIMARY EVIDENCE.