Evidence for the work
we actually do.

Five references behind our approach to vehicle care. Design guidance, not proof of product performance.

Source review · 5 external references · Product documentation linked separately

References guide the design. The questions below are proposed checks, not completed studies.

One question for each product.

DOCUMENTATION / REFERENCE / EVALUATION

No measured Ultrium result is asserted here. These are questions for evaluation.

The supporting references.

EACH SOURCE HAS A SPECIFIC JOB

01
Vehicle diagnostics guidanceEPA

OBD: Frequently Asked Questions

OBD stores information about detected malfunctions to help a repair technician diagnose and fix the vehicle.

Why it matters: codes inform inspection, not a diagnosis. Historical guidance—not proof of coverage or accuracy.

02
Repair decision guidanceFTC

Auto Repair Basics

The FTC advises comparing repair options, requesting a written estimate, and getting a completed repair order.

Why it matters: compare work before approving it. Consumer guidance—not evidence of product savings or legal compliance.

04
Data provenance referenceW3C

PROV-Overview: An Overview of the PROV Family of Documents

Provenance describes the entities, activities, and people involved in producing data, helping assess its quality and trustworthiness.

Why it matters: trace an answer to its source. Provenance is not proof of correctness or W3C conformance.

05
AI evaluation frameworkNIST

Artificial Intelligence Risk Management Framework (AI RMF 1.0)

The AI RMF organizes risk-management work around Govern, Map, Measure, and Manage.

Why it matters: define tests and accountable reviewers. A voluntary framework—not NIST certification or measured product results.

Product documentation.

First-party descriptions, not independent performance evidence.

Help us test
what matters.

Useful evaluations need a clear task, baseline, method, and limitations.

Discuss a product evaluation