RedHub AI
How the roll-up gate works
How the Inferred-Fact & Unstated-Assumption Trace works
This tool grades every fact in each finished deliverable, rolls the deliverable up to its weakest material fact, then the portfolio up to its weakest deliverable. It reads 4 deliverables and rolls them into one verdict — the worst, never the average — then a gate that can only make it worse.
The 4 deliverables
A fact can name its inputs and state its rule and still be an assumption, because something underneath it is assumed. The trace follows every derivation back up to ten levels and demotes the ones standing on nothing, so a fact that reads DERIVABLE on its own row becomes ASSUMED. A deliverable is its weakest material fact and the portfolio is its weakest deliverable — worst-not-average, and the grounding percentage decides nothing.
Why: Contact sheet is GROUNDED and Pricing note is INFERRED — but two deliverables rest on an assumption. In the Vendor risk memo, F-05 names F-04 as its input and applies a sensible rule, yet nothing ever sourced F-04, so F-05 is demoted from DERIVABLE to ASSUMED — a failure one row away that no claim-by-claim review sees. The Renewal summary's F-08 derives from F-20, a fact not in the trace at all. The portfolio is its weakest deliverable, so it reads RESTS ON ASSUMPTION even at 50% grounding.
No single deliverable is the crisis — the clustering is. The roll-up takes the worst, the gate escalates when trouble compounds, and it names the one thing to fix first. Grades the facts you record about a deliverable — never people, and never runs an agent. It does not fetch or verify your sources: GROUNDED means each material fact traces back to a source you named, not that the source says what you think. A working aid, not legal, audit, or compliance advice.