Methodology

Sources

We start with primary research: peer-reviewed papers, government and university laboratory reports, and standards. Preprints are labelled as preprints. Company announcements are treated as reported claims until someone else confirms them. Grants and patents show that work is planned or protected, not that it works.

Every significant factual statement links to a specific source, with the date it was published and the date we last checked it.

Claim types

Each statement on a profile carries one of these labels.

Demonstrated

Shown in a published experiment, test or shipping product.

Reported, unverified

Stated by a company or team but not independently confirmed.

Proposed

A design or plan that has not yet been built or tested.

Disputed

Contested by other credible sources.

Speculative

An idea or possible use. Not evidence of anything.

Superseded

Replaced by newer evidence.

Maturity

We use broad stages because a single readiness number rarely fits a whole field:

• Theory
• Lab demonstration
• Research and prototypes
• Pilot and early commercial
• Commercial
• Not assessed

We use NASA-style Technology Readiness Levels (1 to 9) only when we can justify the level for a specific application, and we publish the reasoning. Otherwise the TRL is shown as “not assessed”.

The radar

Readiness, evidence quality and momentum are rated separately from 1 to 5, each with a dated written rationale. Potential impact is written as a hypothesis. We deliberately do not combine these into one score.

AI assistance and human review

AI agents help find leads, retrieve sources and draft profiles. Every AI-drafted record goes through an automated check against its sources and then human editorial review before publication. The workflow is: discovered, deduplicated, sourced, fact checked, drafted, editorial review, approved, published, monitored. Drafts are never shown on the live site.

Speculation

Possible future uses and product concepts are clearly marked as speculative. They are there to help inventors and readers think, not to predict what will happen. We do not make claims about commercial potential, market size or patent novelty.

Corrections

Each profile keeps its revision history. When we correct something material, we log it on the corrections page.