Research methodology
How GridAPM research is made
GridAPM publishes research to help technical buyers evaluate local-first, human-reviewed agentic AI workflows for transformer APM and condition-based maintenance. Every article follows the same visible pipeline — and you can check our work.
Methodology last reviewed
Editorial pipeline
Four steps, one human gate
The same evidence-in, engineer-approved discipline that shapes the product shapes every article we publish.
-
Source gathering
Primary sources first: standards bodies, government agencies, research institutions, and original research pages — collected before a word is drafted.
-
Drafting with claim boundaries
Claims are written to match implemented scope, carry framing labels (study finding, industry estimate), and never copy standards text.
- Human gate
Engineer review
A qualified reviewer checks technical accuracy, claim scope, and standards references before anything is published.
-
Publication & revision dates
Every article carries published and updated dates; corrections and re-reviews are dated, never silent.
Editorial principles
Useful, careful, and pilot-oriented
Six rules keep GridAPM content valuable for readers while protecting the credibility utilities expect from critical-infrastructure software vendors.
- Primary sources first
- GridAPM articles prioritize official sources from standards bodies, government agencies, research institutions, and original vendor research pages.
- No copied standards text
- Standards are paraphrased for public context. GridAPM does not reproduce copyrighted standard clauses or tables.
- Claims match implemented scope
- Public copy never implies autonomous control, final diagnostic authority, certified compliance, guaranteed failure prevention, or production integrations that do not exist.
- Human-reviewed AI
- Agentic AI is described as evidence organization, draft rationale, gap finding, and work-package preparation — always subject to qualified review.
- Utility buyer relevance
- Articles and tools are written for utilities, TSOs, DSOs, generation, data centers, oil and gas, and industrial power teams evaluating transformer APM pilots.
- Search value with technical value
- Pages target real search intent while adding useful tools, checklists, tables, diagrams, and internal links that help pilot teams act.
Worked example
What a GridAPM citation looks like
Every numeric claim on this site comes from a verified claims sheet and carries its footnote source. Here is one, taken apart.
As published
Large power transformers are typically designed for a service life on the order of 40 years, yet the average age of installed U.S. units is about 40 years — and per a 2014 DOE estimate restated in DOE's 2024 Report to Congress, more than 70 percent were over 25 years old.1
Reference
- Source type
- Primary source — a U.S. government report to Congress, linked to the original PDF.
- Framing
- Study finding, quoted with its original scope: a 2014 DOE estimate, restated in the 2024 report — the vintage is stated, not hidden.
- Verification
- Checked against the source PDF by an independent adversarial pass before entering the approved claims sheet.
What we checked
Standards-referencing policy
Context, not certification
GridAPM articles paraphrase standards for public context and never reproduce copyrighted clauses, tables, or figures. Referencing IEEE C57.104 or IEC 60599 describes the diagnostic conventions our evidence workflows respect — it is never a compliance claim.
The same rule applies to governance frameworks: NIST AI RMF and NERC-relevant posture are described as design context for human-reviewed AI in critical-infrastructure environments, never as an assertion of certification, accreditation, or endorsement.
- IEEE C57.104
- IEC 60599
- IEEE C57.91
- CIGRE TB 642
- NIST AI RMF
- NERC
Every page passes these checks
- Primary-source citations wherever a factual claim depends on external material.
- A clear line between current public positioning, pilot capabilities, and roadmap language.
- No unsupported claims about certification, customers, ratings, ROI, carbon savings, or autonomous control.
- Internal links that help a pilot team act — platform, pilot, security, data handling, standards, tools.
Check our work
The methodology only matters if you can see it applied. Read the research, or see who we are.