About

Reliability priced on evidence, not age bands.

Autorelx started with a frustration inside an underwriting team: extended-warranty and mechanical-breakdown cover across APAC was still priced off a car’s age and a broad category — while the actual driver of claims, which make, model, year, fuel and mileage, sat unused in plain sight.

109M+
real MOT inspections analysed (2023–2024)
7
validated public & industry data sources, cross-checked
590+
real models scored across 60+ makes — and growing
100%
of figures tagged measured or modelled, with a confidence band

“A book of warranty business was losing money on a handful of models everyone in the workshop already knew were trouble — while over-charging the durable ones that quietly subsidised them.”

The information to tell them apart existed. It just wasn’t assembled into something you could price a single vehicle with — and nobody was doing it for APAC specifically, where the vehicle mix (JDM imports, SEA-market models, a fast EV transition) looks nothing like the US or UK datasets most tools are built on.

How it works

Built the other way around.

01

Start from the evidence

Millions of real inspection records, regulated claims disclosures and manufacturer recall data — assembled into one place you can price a single vehicle with.

DVSA MOTFCA claimsNHTSAworkshop pricing
02

Model it honestly

Failure frequency shape, an anchored claims level, and per-component severity — credibility-weighted, so thin data leans on the market and never pretends to be certain.

frequency × severitycredibility-weightedconfidence bands
03

Only claim what's supported

A grade, an expected cost and a five-year outlook — every number labelled measured or modelled. Where there's no basis, we withhold rather than guess.

A–E grademeasured vs modelledwithheld when thin
The transparency principle

Every number carries its confidence level and where it came from. When the data is thin or missing, we withhold or disclose rather than guess — an “insufficient data” badge is a feature, not an embarrassment. A skeptical actuary should be able to trace any figure back to the source that produced it, and see exactly which numbers are measured and which are estimated. That is the whole product: not a black box that spits out a score, but an instrument you can audit.

Autorelx is a pilot-stage reliability-intelligence product, not actuarial advice. Partners who contribute their own claims or workshop data receive calibrated outputs exclusive to them. See the methodology for exactly how every figure is built, or try the live console.