A long-only global equities desk runs the Global Growth Fund against a hard sector-concentration mandate. A corporate action on a held position pushed technology exposure 60 bps over the cap. The default playbook: force the sale.

G-Nosis surfaced an 18-month-old decision from the Sydney desk. Same fund. Same constraint. Resolved with a 30-day cure window instead of a forced sale. The PM had never seen the prior decision; the analyst who originally drafted it had moved teams.

The cure window was approved. Execution slippage avoided. The new decision now sits in the graph as a precedent of its own.

A high-yield credit team runs an investment committee twice a week. Each agenda item is a credit, and each credit drags six hundred prior decisions behind it: covenant negotiations, restructuring votes, sponsor concessions, triggered baskets.

Pre-G-Nosis, the analyst spent the day before each IC reconstructing precedent from PDFs, deal closing binders, and the memory of whichever partner had been in the room. Six hours, on a good day.

With the schema populated to nineteen edge types — including CITES_PRECEDENT, VIOLATES_LIMIT, and a firm-specific COVENANT_OPENED — IC prep is a single trace query. Twenty-two minutes, by the analyst's own stopwatch.