Cold prompt set
The prompts are written as real external questions: who should I use, who is credible, what should I know, who else should I consider?
Method record
How the read is made.
The Readback tests what an external person would find if they asked AI systems about your organisation without being handed your preferred story first.
System sweep
The current read covers ChatGPT, Claude, Gemini, Perplexity and Google search responses, including AI Overview where shown. The systems used are recorded in the report.
Evidence log
Each material output is recorded with the engine, prompt, date, answer returned, competitor named, and source links where the system supplies them.
Human diagnosis
The returned answers are read by Chris against what appears to be true about the organisation. The work is in the judgement, not the prompt running.
Risk order
Findings are ranked by likely commercial risk, not by novelty. A competitor substitution matters more than a harmless wording quirk.
Correction boundary
The report gives a practical order of correction. Any repair work is separate, so the diagnosis is not shaped by the need to sell a fix.
Next step
Request a diagnostic read.
Useful context includes your website, your market, and whether the concern is new business, funding, recruitment, reputation or a forthcoming change.