NADIR Apex Shadow-mode advisory Premium · highest quality · full forensic depth

Algorithm tier deliverable

Forensic ADAS reliability for OEM programs and insurer-backed fleets

NADIR Apex delivers the full perception intelligence stack — A1–A6 forecast ensemble, changepoint detection, CRLB ellipsoids, factor-graph evidence, and budget-optimized fleet intervention planning.

A1–A6 forecast BOCPD + HSMM spectral GLR CRLB ellipsoids factor-graph knapsack optimizer

Premium forensic lane

What Apex adds above Meridian

Weighted ensemble across BOCPD, spectral GLR, and HSMM virtual detectors — plus mock CRLB ellipsoids, factor-graph evidence theater, and attribution boost on hard multi-modal scenarios.

3-detector ensemble

Weighted vote fusion

Mahalanobis + spectral GLR + BOCPD + HSMM stubs — median-weighted tier vote on every frame.

Attribution boost

Dominant-modality lock-in

Higher attribution hit rate vs Meridian when ensemble agrees on CRITICAL — insurer dispute ready.

CRLB ellipsoid

Identifiability glyph

Mock Fisher-information ellipse from extrinsic drift — 3D Fleet Twin preview in production Apex.

Evidence theater

Factor graph + bundle

Forensic HTML appendix with chain-of-custody preview and geodesic replay stub.

Capability map

What Apex includes

A1–A6 ensemble

14-day stochastic band + survival + ellipsoids + knapsack.

3D Fleet Twin

SE(3) frames, CRLB glyphs, forecast overlay.

Evidence Theater

Factor-graph layout + geodesic replay scrubber.

What-if planner

Counterfactual recalibrate scenarios.

Horizon Map

deck.gl regional risk + route identifiability.

Unlimited evidence

Signed EVB bundles with full chain of custody.

Supported

  • Full stochastic 14-day forecast (A1–A6)
  • BOCPD + HSMM + spectral GLR ensemble
  • CRLB identifiability ellipsoids (3D twin)
  • SE(3) geodesic replay + evidence theater
  • Fleet knapsack intervention planner
  • What-if counterfactual forecasting
  • Unlimited signed evidence exports
  • WebSocket live incident rail + geospatial map

Not included on Apex

  • On-vehicle ECU writes or in-loop control
  • OEM homologation sign-off
  • Insurer underwriting certification
  • Guaranteed prediction accuracy
  • Real-time edge inference on vehicle

Commercial

Apex pricing

Indicative USD pricing · shadow-mode on every tier · binding terms in LOI or MSA.

Apex

$8.50

per vehicle / month

  • Full stochastic 14-day forecast (A1–A6)
  • BOCPD + HSMM + spectral GLR ensemble
  • CRLB identifiability ellipsoids (3D twin)
  • SE(3) geodesic replay + evidence theater
  • Fleet knapsack intervention planner
  • What-if counterfactual forecasting

Min 100 vehicles · enterprise quote above 2,000

Enterprise quote →

ROI estimator

Model annual NADIR cost vs calibration-incident avoidance for buyer conversations.

Claim boundary: NADIR is shadow-mode reliability analytics. Not ISO 26262 / ASIL safety-certified. Not in-loop vehicle control. Forecasts and tiers are advisory operational analytics.

Developer surface

Apex API scopes & endpoints

Org-scoped Bearer API keys · RBAC scopes gate algorithm depth · OpenAPI v1 contract.

Required scopes

horizon:forecast

Full A1–A6 forecast plane

horizon:read

All six HORIZON modules

evidence:read

Unlimited signed bundle export

fleet:read

Full extrinsic history + replay

platform:*

Webhooks, admin, audit

Primary endpoints

GET /v1/horizon/forecast/{id} <4s
GET /v1/horizon/twin/{id} <6s
GET /v1/horizon/fleet-forecast <8s
GET /v1/horizon/factor-graph/{event_id} <5s
POST /v1/horizon/what-if <8s
WS /v1/horizon/stream 1 Hz deltas

Latency profile (p50–max)

Demo instrumentation

Go-to-market

Who buys Apex

OEM ADAS program office

“Field monitoring adjacent to homologation with audit-grade evidence.”

Typical buyer on NADIR Apex.

Insurer fleet program lead

“Chain-of-custody bundles at claim close.”

Typical buyer on NADIR Apex.

Large fleet safety engineering

“Which vehicles go CRITICAL in 14 days — with budget optimizer.”

Typical buyer on NADIR Apex.

MSO enterprise QA

“Caliber/Safelite-scale post-repair forensic replay.”

Typical buyer on NADIR Apex.

Onboarding

Integration timeline

Day 0 — API key + sandbox fleet

12-vehicle sample org · all scopes for integration testing · make dev-api local stack.

Week 1 — Telemetry ingest

Wire POST /v1/telemetry/ingest from telematics or batch replay · validate tier gates.

Week 2–4 — Shadow evaluation

Console or webhook alerts · weekly digest · Exhibit A metrics for LOI conversion.

Conversion — production tier

Per-vehicle billing at Apex rate · optional API add-ons per integration depth.

Mahalanobis drift trace (simulated scenario)

Apex overlay
Ground truth ● Apex detection
96%
Detection score
95%
Attribution accuracy*
22%
Efficiency score
100%
Shadow-mode safety

Compare

Apex vs other tiers

Capability Pulse Meridian Apex
Tier gates (NOMINAL/CAUTION/CRITICAL)
Survival / time-to-CRITICAL✓ full
HORIZON Command Deck
3D Fleet Twin + CRLB
Signed evidence exportsLimitedUnlimited
Typical latency<200ms0.5–2s2–8s
Per-vehicle / mo$3–5$3.75$8.50
Pulse Meridian Apex

SDK reference

Python SDK modules for Apex

Install: pip install -e "NADIR_SDK/python[dev]" · TypeScript client at NADIR_SDK/typescript

Core modules

nadir_sdk.perception.forecasting.propagationA1 stochastic 14-day band
SDK
nadir_sdk.perception.forecasting.prioritizeA4 knapsack fleet prioritizer
SDK
nadir_sdk.perception.forecasting.ellipsoidA3 CRLB renderable glyphs
SDK
nadir_sdk.perception.forecasting.graph_layoutA6 factor-graph SMACOF
SDK

Example: score telemetry payload

from nadir_sdk.perception.forecasting import fleet_forecast, twin_payload

forecast = fleet_forecast(org_id, vehicle_ids, horizon_days=14)
twin = twin_payload(vehicle_id)  # SE(3) frames + CRLB ellipsoids

Telemetry contract

Ingest payload shape

Same JSON schema across tiers — algorithm depth is selected by org plan and ?tier= routing.

{
  "vehicle_id": "van-0142",
  "timestamp": "2026-06-10T14:22:01Z",
  "sensors": {
    "camera_rotation_matrix": [1.0, 0.01, 0.0, -0.01, 1.0, 0.0, 0.0, 0.0, 1.0],
    "radar_range_bias_m": 0.002,
    "radar_azimuth_bias_deg": 0.15,
    "lidar_registration_error_m": 0.001
  },
  "timestamp_quality": { "sync_ok": true, "jitter_ms": 4 }
}

camera_rotation_matrix

9-element row-major rotation vs baseline. Primary signal for extrinsic drift.

radar_range_bias_m

Range systematic error in meters — converted to mm in residual vector.

lidar_registration_error_m

Cross-modal registration residual for fusion disagreement detection.

timestamp_quality

sync_ok and jitter_ms — gates unreliable frames before scoring on Meridian+.

organization_id

Resolved from API key context — strict org isolation on every fleet record.

Deployment patterns

Apex use cases in production conversations

OEM field reliability program

2,000+ vehicle homologation-adjacent monitoring with 3D twin exports.

Insurer-backed fleet program

Signed evidence bundles attached to claims workflow at close.

Amazon-class last-mile

Enterprise LOI — knapsack prioritizer under fixed weekly bay capacity.

Caliber-scale MSO

Forensic replay for post-repair QA disputes with factor-graph theater.

FedEx ISP regional

Safety engineering team — 14-day CRITICAL exceedance probabilities.

OEM telematics design partner

Embed Apex forecast API behind OEM branded portal.

Operations

SLA, quotas, and support

Rate limits (Apex)

Ingest bucketPulse-optimized batch POST
50k/day
Read bucketFleet health + extrinsics
20k/day
Forecast computeIncluded
5k/day
Overage blocks+10k ingest
$50

Support tier

Micro pilotBusiness-hours email · 1 kickoff call
Included
SubscriptionDedicated FDE (Apex)
SLA
Evidence exportsUnlimited
Apex
Console seatsUnlimited

Terminology

Glossary for buyer conversations

Mahalanobis distance

Multi-sensor drift vector fused into one statistical distance — core signal on every tier.

Extrinsic drift

6-DOF misalignment of camera/radar/lidar relative to vehicle frame.

Shadow mode

Advisory analytics only — no ECU writes or in-loop vehicle control.

Tier gate

NOMINAL / CAUTION / CRITICAL classification from residual thresholds.

Chain of custody

Signed evidence bundle metadata for insurer and auditor review.

Survival curve

Time-to-tier exceedance probabilities (Meridian+).

CRLB ellipsoid

Fisher-information uncertainty glyph for 3D twin (Apex).

Idempotency-Key

POST replay protection — cached response on duplicate ingest.

Org isolation

API key from Org A cannot read Org B fleet records.

Exhibit A

LOI attachment with week-1 pilot metrics JSON.

Quick integration

cURL ingest example

curl -sS -X POST "https://api.nadirai.net/v1/telemetry/ingest?tier=apex" \
  -H "Authorization: Bearer $NADIR_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: ingest-van-0142-20260610" \
  -d @telemetry_frame.json

Webhook payload (tier transition)

{
  "event": "tier.changed",
  "vehicle_id": "van-0142",
  "previous_tier": "NOMINAL",
  "drift_tier": "CAUTION",
  "mahalanobis_distance": 3.42,
  "probable_fault_source": "camera_extrinsics",
  "algorithm_tier": "apex",
  "organization_id": "org-acme-regional"
}

Extended FAQ

Integration & commercial

How do API keys map to tiers?
Org-level plan sets default algorithm lane. Pass ?tier=apex on scoring routes when permitted by scope.
What telemetry frequency is required?
Daily batch minimum for pilot; hourly or event-driven for production CRITICAL sensitivity.
Can we self-host?
NADIR cloud is default; edge gateway pattern documented in NADIR_SDK/examples/edge_vehicle_gateway/.
What vehicles are supported?
Any ADAS-equipped van, shuttle, or passenger vehicle with camera/radar metadata in telemetry.
How does billing work?
Per-vehicle per month at $8.50 — prorated on tier changes mid-cycle.
Is there a free tier?
Developer sandbox with 12-vehicle sample fleet — $0 for integration testing.
What about Echo archival?
Batch replay add-on — can replay stored telemetry at Meridian or Apex depth.
How do webhooks sign payloads?
HMAC-SHA256 with org webhook secret — verify utils in nadir_sdk.webhooks.
What's in the weekly digest?
Tier counts, CRITICAL queue delta, top drift vectors, Exhibit A summary metrics.
Who owns the data?
Customer telemetry — NADIR processes under DPA; evidence exports are customer-controlled.

FAQ

Apex questions

When does Apex make sense over Meridian?
OEM field programs, insurer evidence requirements, or fleets needing 3D twin + knapsack prioritizer.
Is Apex safety-certified?
No — shadow-mode advisory analytics on every tier. See claim boundary on all pages.
Can we mix Apex on a CRITICAL subset only?
Yes — per-vehicle tier assignment is supported in org billing.

Tier benchmark · Phase 4

Apex vs Meridian on hard scenarios

Multi-modal, delayed-onset corpus — ensemble vote fusion, attribution boost, and mock CRLB ellipsoids in the tier benchmark lab (not final pipeline_v3).

*Simulation benchmark — illustrative for OEM field programs and insurer evidence demos.

Deploy NADIR Apex in shadow mode

Min 100 vehicles · enterprise quote above 2,000 · same claim boundary on every tier · upgrade path without re-architecture.

Enterprise quote Full pricing matrix Talk to us