Most diligence questions start with a simple prompt: what can be responsibly understood from public material, and what remains uncertain? The answer is only useful if every step from source to conclusion stays traceable.
This brief walks through a sanitised example. The question is whether public information supports a risk concern strongly enough to brief a decision team.
Start with public material
A public record established the baseline fact pattern[1]. Regional media added relevant context but did not, by itself, justify a conclusion[2].
Corroborate before you conclude
A lead is not a finding. Analysts reviewed whether the public material was consistent, whether alternative explanations remained plausible, and what would need to be caveated before briefing[3]. An unresolved point was preserved as an open question rather than promoted to a conclusion[4].
The brief separates confirmed facts, risk context, and open questions. No single source is load-bearing, and limitations are stated alongside the finding.
Where it would have broken
Had we matched on name alone, the finding collapses the moment a defence lawyer asks about coincidence. Had we trusted the AI correlation without analyst review, a plausible-but-wrong link could have entered the briefing unchallenged. The discipline is not the collection — it is the refusal to promote anything to a conclusion until it is sourced, scored, and reviewed.
That is the whole point of a source trail: not to prove how clever the analysis was, but to let someone else retrace it and reach the same conclusion. Defensible intelligence is reproducible intelligence.