Decisions

Turn business rules into decisions you can test and explain

Combine customer history, risk and commercial constraints to determine eligibility, limits and terms. Test your policy before using it, and see which rule led to each result.

See Ormuz in action Start building

Decision

Offer payment terms

Buyer & order

Customer for: 3 years
Country: France
Risk: Moderate
On-time payments: 98%
Order value: €15,000
Available credit: €12,000

Payment terms policy

Matching rule

Payment terms policy
Customer for	Market	Risk	On-time payments	Credit / order	Terms
≥ 1 year	EU	Low	≥ 98%	≥ 100%	Net 60
≥ 2 years	EU	Moderate	≥ 95%	≥ 80%	Net 30
Any	Any	High	Any	Any	Prepay

Payment terms offered

Deferred amount: €12,000
Upfront payment: 20%
Payment term: 30 days

Illustrative policy · Selected rules and a fictional buyer

Business policies

One policy. The right decision for every situation.

Customer history, risk and available credit combine to determine what you can offer. The same policy can approve longer terms, require a deposit or ask for payment upfront.

DecisionOffer payment terms

Established buyer

Risk: Low
On-time payments: 99%
Credit / order: 100%

60days

100% deferred

Growing buyer

Risk: Moderate
On-time payments: 98%
Credit / order: 80%

30days

20% upfront

High-risk buyer

Risk: High
On-time payments: 92%
Credit / order: 100%

100%upfront

No deferred payment

Illustrative profiles · One policy, three sets of terms

Test before use

Treat policy changes as product changes

A Decision can be evaluated against representative inputs before a process depends on it. This creates a natural place to exercise thresholds, edge cases and routing outputs without replaying an entire end-to-end journey.

Implication

What this boundary changes in the design

When a rule is genuinely simple and deterministic, keeping it here is also cheaper and more predictable than delegating the same rule to an AI model.

Boundary

Not every judgement is a deterministic rule

Ambiguous document interpretation, contextual classification or language-heavy analysis may justify an Agent activity instead. Human review may still be required when the impact or applicable policy calls for it.

The important design choice is explicit: deterministic policy in a Decision, bounded AI work in an Agent, operational sequencing in a Process.

Read the documentation for this capability

Decisions

See Ormuz in action, then start building.

Explore a real journey, then open the console to build your own processes.

See Ormuz in action Start building