Live KPIs + forecast triage
Tier counts today and forecast at +14 days, a streaming incident rail with time-to-critical countdowns, and a ranked priority table that tells your team who to dispatch first — and why.
live stream · priority rankingNADIR HORIZON
Mission-control for your fleet's perception health.
A live 3D digital twin of every sensor's calibration, a forecast of which vehicles will go critical and when, and a one-click signed evidence trail — all in shadow mode, wired to your existing telemetry. No new hardware, no control authority, no guesswork.
Investor demo · read-only · no sign-in
Open the full 3D command center against the pilot fleet on api.nadirai.net — live stream, swarm, map replay, and evidence theater.
● NOMINAL
● CAUTION
● CRITICAL
14-day forecast horizon
14 days
Tier-crossing forecast with full survival distributions — never a bare point estimate
2,000
Vehicles rendered in one GPU-instanced 3D swarm at 50–60 fps
6 modules
Deck, 3D Twin, Map, Forecast, Evidence Theater, What-If — one authenticated console
100%
Shadow mode. Advisory analytics only, org-scoped and reproducible against signed evidence
Why HORIZON
Your dashboards tell you a vehicle is critical. HORIZON tells you which ones will be — and what to do first.
Every reactive fleet tool answers yesterday's question. The question a fleet safety director actually buys is: "Which vehicles cross CRITICAL in the next 14 days, with what confidence, and what is the cheapest intervention order?" HORIZON answers it by forecasting each vehicle's coupled sensor-bias state into tier-crossing survival curves — then showing the result as a cinematic, physically honest 3D command surface.
One console, six surfaces
Everything between raw telemetry and a maintenance decision.
True 3D calibration twin
Every sensor's factory baseline vs. current pose rendered in WebGL, with the drift arc between them and uncertainty ellipsoids that honestly stretch where drift is unobservable. Zoom from one vehicle to a 2,000-vehicle swarm.
WebGL · SE(3) frames · uncertaintyGeospatial fleet risk
Your fleet on a live map, colored by tier and sized by urgency. Replay any vehicle's route and see exactly where along the road its calibration health was observable — and where it wasn't.
route replay · regional heatTime-to-critical, with confidence
Per-vehicle survival curves with 10th/50th/90th percentile markers, forecast bands against the CAUTION and CRITICAL thresholds, and per-sensor fault attribution. Uncertainty is always on screen.
survival curves · forecast bandsReplay the proof in 3D
Reconstruct any signed evidence bundle as a 3D factor graph and scrub through the drift history as it actually happened. Chain-of-custody and cryptographic signature included — nothing fabricated, ever.
signed bundles · drift replayPlan the cheapest intervention
Simulate "recalibrate this sensor now vs. in five days" or a scenario shock (post-collision, thermal, vibration) and see the avoided-CRITICAL delta side-by-side. Build a budgeted intervention plan in minutes.
scenario planning · avoided criticalsHow it works
Three planes, zero new hardware.
PLANE 1 · MATH
Forecasting engine
NADIR's perception SDK propagates each vehicle's full sensor-bias state forward in time and converts it into tier-crossing survival distributions — stochastic forecasting, not a trend line.
PLANE 2 · API
Org-scoped service layer
Forecasts, twin payloads, replays, and priority rankings are served from the same authenticated, rate-limited, tenant-isolated API your existing NADIR integration already uses.
PLANE 3 · CONSOLE
Cinematic 3D command center
A fast, beautiful browser console — live streams, GPU-rendered fleet twin, map, and evidence theater. Works against your live fleet or the built-in pilot sample fleet on day one.
Fleet safety & maintenance directors
Last-mile and regional fleets running camera/radar/lidar-equipped vehicles.
- Know which vehicles cross CRITICAL before they do.
- Dispatch recalibration in the cheapest order, with rationale.
- Cut surprise downtime and unscheduled depot visits.
OEM ADAS program offices
Program teams tracking perception reliability across deployed platforms.
- Fleet-wide calibration drift visibility per platform and region.
- Physically honest uncertainty — unobservable axes shown as unobservable.
- Shadow-mode analytics that complement, never replace, homologation.
Collision & repair networks
MSO regional GMs connecting evidence to forecast-driven dispatch.
- Signed, replayable evidence for every calibration event.
- Forecast-to-dispatch: route vehicles to bays before failure.
- Claims-grade chain of custody, exportable in one click.
Our claim boundary
Honest by construction.
- Shadow mode only. HORIZON is advisory reliability analytics. It is not ISO 26262 / ASIL safety-certified, not an OEM homologation replacement, and never holds vehicle control authority.
- Uncertainty always shown. Every forecast renders its confidence bands and percentiles. You will never see a bare point estimate.
- No false confidence. Where drift is mathematically unobservable from your data, the twin shows it — uncertainty ellipsoids stretch instead of pretending to know.
- Real evidence, real signatures. Evidence views reflect the actual signed bundle and chain of custody. Nothing is reconstructed from anything unsigned.
Get started
See your fleet's next 14 days.
HORIZON ships with a full pilot sample fleet, so your team can explore every module the moment access is granted — then connect your live NADIR telemetry when you're ready. Pilots typically go from first login to a forecast-driven intervention plan inside one working session.