The Risk Engine reads live readiness signals and history to forecast what actually slips: which test window holds, which issue closes late, and which dependency quietly moves your date.
By the time a slip is visible in the schedule it has already happened. The signals were there weeks earlier, scattered across logs nobody reads together.
The riskiest items are not the loud ones in the morning meeting. They are the ones that stopped talking 42 days ago and are still on the critical path.
Vendor witnesses, test prerequisites and open issues rarely line up the way the lookahead assumes. The window slips on the day.
The lessons of the last five buildings live in people, not systems. The same failure modes repeat with new logos.
Forecasts are computed from the same evidence graph the rest of IQ runs on, so every probability can show its drivers.
Issue cadence, retest rates, evidence flow, vendor confirmations and dependency states, streamed from the live graph.
Patterns are scored against how similar situations resolved before: silent issues, failed-retest chains, witness no-shows.
A probability for each upcoming milestone, decomposed into ranked drivers with the action that most improves the odds.
Not a red-amber-green feeling: probabilities with drivers attached.
A live probability for each test window and milestone, updated as evidence lands, with the trend since last week.
The specific items moving the probability: an untested UPS, a silent issue, an unconfirmed witness, each with its days-at-risk impact.
When a pattern matches past trouble, the engine says so, and shows what actually fixed it last time.
Flags raised weeks before the lookahead notices, routed to the owner of the driver, not broadcast to everyone.