A decision engine that
shows its reasoning.

A decision engine is software that evaluates rules against data and returns a decision, usually with the reasons that produced it. The shape is always the same: a request goes in carrying whatever the decision needs, the engine applies logic that lives outside your application code, and an outcome comes back in real time, typically approve, decline, refer, price, or route. What makes it an engine rather than a function is that the logic is a managed asset: versioned, testable, editable by the people who own the policy, and auditable long after the fact. GoRules provides that as visual decision graphs and decision tables running on an open-source evaluation core.

In practice the terms overlap, and vendors use them interchangeably. The distinction people usually mean is one of scope: a rules engine executes a set of if/then rules, while a decision engine or decisioning platform is the system around it, orchestrating several rule sets and data sources with simulation, versioning, environments, approvals, and an audit trail. GoRules is both halves, and they are separable. The MIT-licensed ZEN Engine does the evaluating and embeds on its own, while the BRMS around it handles authoring, testing, version control, and multi-environment releases.

Hardcoded logic is the right answer when the rules are few, stable, and owned by the engineers who wrote them. You have outgrown it when the same rule is copied across more than one service and the copies drift, when business teams file tickets to change thresholds they own, when you cannot explain why an individual case got the outcome it did, or when you cannot show who approved the version that was live on a given date. The clearest tell is cadence: if policy changes weekly and your deploys are quarterly, the logic belongs outside the code. Moving one decision out is usually enough to prove the point.

Model output enters the decision as an input, not as the decision itself. Your model emits a score, a probability, or a classification, and the engine decides what to do about it: where the cutoff sits, which cases are knocked out before the score is consulted, who gets referred to a person, what price or limit applies. That separation is what keeps the outcome explainable, because it traces to readable rules rather than to a score alone, which matters under GDPR Article 22, the EU AI Act's high-risk classifications, and internal model risk review. GoRules supplies the traceability; the compliance obligations remain yours.

Yes. Run the BRMS self-hosted with Docker or Kubernetes on AWS, Azure, or Google Cloud, and put the Agent in front of it for a high-performance REST endpoint with hot rule reloading. If you want evaluation in-process with no network hop at all, embed the open-source ZEN Engine directly through the Node.js, Python, Java, Go, C#, Rust, Kotlin, Swift, or WebAssembly SDKs, so decision data never crosses a vendor boundary. SSO and role-based access control govern who can read or change rules in either setup, and GoRules Cloud is there when managed hosting is the simpler answer.