Decision Flow

Decision Flow works best for real-time business decisions where multiple steps need to happen in sequence. Typical examples include pricing calculations, approval workflows, eligibility checks, onboarding flows, fraud screening, routing logic, or combining multiple partial decisions into one final result.

If your logic involves first evaluate this, then call another rule, then enrich the data, then produce a final response — Decision Flow is the right fit.

Decision Flow is optimized for fast, real-time execution. It handles most business decision workloads where the evaluation finishes within the allowed execution time.

Integration Flow was built for long-running processes such as large database queries, bulk processing, or slow external integrations that may exceed normal execution limits.

The two complement each other:

Decision Flow handles immediate decisions.

Integration Flow handles large-scale or long-running processing.

Yes. If your Decision Flow grows beyond the execution limits — for example due to large datasets or long-running API calls — you can convert it into an Integration Flow without losing your work.

This allows teams to start simple and scale only when needed.

Decision Flow can orchestrate all major rule types across your workspace, including Decision Tables, Decision Trees, Scripting Rules, and more. Each node in the flow performs one responsibility, while the Flow coordinates the entire execution process.

Yes. Decision Flow supports REST API calls directly inside the flow execution. This allows you to enrich decisions with external systems, retrieve additional data, or trigger downstream services as part of the decision process.

Because Decision Flow is optimized for real-time execution, external calls should be tested to ensure they complete within the execution time limits.

Plan for failure paths from the start. You can add fallback branches, validate required inputs early, and define alternative paths for missing data or failed API calls directly inside the Flow.

But when something still goes wrong, Decision Flow includes Inspect Mode — a visual execution view that shows exactly what happened inside each node during runtime. For every node, you can see:

This makes debugging dramatically easier than tracing hidden backend workflows or digging through application logs.

If execution takes too long, the flow may time out. This usually happens with very large bulk inputs, complex processing chains, slow API responses, or large database operations.

In these cases, the recommended solution is moving the workload into an Integration Flow.

Yes. Decision Flows support the same versioning model as other rules in your workspace. You can save versions, restore previous states, and publish updates without losing historical logic.

Access permissions can be configured at multiple levels including spaces, teams, roles, and accounts. You can allow editing, restrict access to read-only, or manage production workflows through your permission model.

Yes. The Management API allows you to create, update, publish, and manage Decision Flows programmatically. This is useful for automated deployments, or synchronizing rules with external systems.

A good Decision Flow keeps responsibilities separated. Each node should do one clear job — validate data, evaluate a rule, enrich information, or transform outputs.

The Flow itself should focus on orchestration, not hiding massive logic inside one oversized node.

Keep nodes focused and predictable. Each node should ideally perform one clear responsibility — validate data, evaluate a rule, enrich information, transform outputs, or call an external service.

Avoid creating nodes that try to do everything at once. Smaller, specialized steps make the Flow easier to read, easier to debug in Inspect Mode, and much safer to modify as the process evolves over time.