Loading Runway...

Risk assessment + disruption response: where AI helps and where it freezes — Runway | Runway

MICRO · SUPPLY CHAIN MANAGER

Risk assessment + disruption response: where AI helps and where it freezes

AI runs a thousand Monte Carlo simulations on your supply chain risk model. It flags geopolitical signals from news scrapes. It also delivers exactly the wrong answer when the disruption is a tail event — and that is the disruption that ends the quarter.

14 MIN8 cards5 questions

Module 3 · Card 1 of 8

Where AI helps vs freezes in disruption

AI runs a thousand Monte Carlo simulations on your supply chain risk model in seconds. It flags geopolitical signals from news scrapes. It auto-generates allocation suggestions during shortages. But when the disruption is a tail event — Suez Canal closure, sudden export ban, factory fire at sole-source — the model freezes on patterns it hasn't seen. And that disruption is the one that ends the quarter.

Module 3 · Card 2 of 8

The pattern

Disruption response has three layers — detection (seeing the disruption coming), modeling (quantifying impact across the chain), and execution (calling suppliers, expediting, deciding which customers ship). The agent is excellent at the first two. It is bad at the third — execution requires authority, judgment, and trade-offs no model has the standing to make.

Module 3 · Card 3 of 8

What the evidence shows

MIT CTL's 2026 Q1 Supply Chain Resilience Study analysed 180 major disruption events across 24 industries. AI risk-detection tools surfaced the disruption ahead of impact in 41% of cases — strong on geopolitical and weather signals. But when surveyed on which response decisions actually saved the quarter, supply chain leaders cited human judgment on trade-offs (which customer, which SKU, which port) in 87% of cases. The model saw it coming; the human decided what to do.

Module 3 · Card 4 of 8

What this means for you

Use AI for detection and modeling — let it run a thousand scenarios, surface weak signals, quantify network-level impact. Sit on the trade-off decisions yourself. The SCMs compounding fastest in 2026 are reinvesting saved time into building the playbooks for the trade-offs they will face — which customers get partial shipments, which suppliers we lean on, what's our line on allocating to highest-margin vs strategic accounts.

Module 3 · Card 5 of 8

How to apply this tomorrow

Pick one disruption scenario your AI risk tool surfaces as plausible — sole-source supplier outage, customs delay, single-port concentration. Write a one-page response playbook by hand: who calls whom in the first 4 hours, which customers get the truth first, which suppliers we lean on, where we expect to lose margin and where we don't. The agent gives you the scenarios. You decide the response.

Module 3 · Card 6 of 8

Common failure mode

The trap: defaulting to AI-recommended allocation during live disruption. A 2026 SCMR study of 240 active disruption responses found 56% of AI-recommended allocations damaged at least one strategic customer relationship — because the model optimised on margin and fill-rate metrics without seeing the customer-relationship cost of letting them down at the worst moment. AI saw the math; the human had to live with the customer.