Decision Intelligence, vendor-agnostic.
The concepts that define the field — written so you can carry them into any product evaluation, ours included. No gate, no allegiance required.
Good ideas don't belong to a vendor. The distinctions below are the load-bearing ones for the whole category — true whether you buy VYAN, build it yourself, or buy someone else. We'd rather you understand them and choose well than gate them and hope you don't look closely.
Stochastic vs deterministic planning
The architectural choice that separates Decision Intelligence from a generation of single-future supply-chain plans — and the operational consequence of the choice for what the planner sees, decides, and commits.
Deterministic planning treats every driver — demand, lead time, yield, capacity — as a single number. One number for the part's lead time, one number for next-quarter demand, one mean for supplier on-time delivery. The optimizer maximizes against that one synthetic future.
Stochastic planning treats those same drivers as the distributions they actually are. Lead times that range from 9 to 31 days. Demand that surges 18% in one urban market while another goes flat. The planner asks not "what is the best plan for the average" but "what is the best plan across the range of plausible futures."
The math gets harder. The result gets defensible. Best-mean plans optimize an outcome you will never actually see; stochastic plans optimize for the actual P&L distribution your business will live through.
Policy vs plan
A plan answers a single committed question: what should we do, given the forecast we have agreed to act on? A policy answers a broader one: what should the system do under the realistic distribution of futures the enterprise is actually planning against?
A plan is a sequence of decisions calibrated to one specific forecast. When the forecast moves — and it always moves — the plan goes stale. The planner makes a new plan. The new plan abandons the commits the previous plan made. Working capital that was being chased toward last week's posture now reverses course.
A policy is a decision rule that says: given the state of the world right now — observed demand signal, current inventory, supplier Available-to-Promise (ATP), capacity — what's the right action? The policy doesn't go stale when the forecast moves; it ingests the new signal and emits the next-right action consistent with the same posture.
Plans optimize for the world we imagined when we wrote them. Policies optimize for the world that actually shows up. The point of decision intelligence is the latter.
Resilient EVA · dollarized objectives
The objective function that lets a CFO read a planning optimizer's output without translating it first.
Most planning optimizers maximize a service-level number, or a fill-rate, or a forecast accuracy metric. These are not the numbers the business is actually measured on. The CFO is measured on dollars.
Resilient EVA — Resilient Economic Value Add — collapses revenue, cost, working capital charge, expedite, and the tail-risk penalty into a single dollar number. The optimizer maximizes that number. Every decision the policy emits carries a dollar score. Every tradeoff is mediated by the dollar.
The 'resilient' qualifier is the Conditional Value at Risk (CVaR) penalty: a tail-risk charge that prevents the optimizer from picking plans with great expected value but P95 outcomes that would break the business. CFO-readable. Board-defensible. Same number across functions.
Every term these concepts use is grounded in the glossary.