Decision Assistant and the Discovery of Decision Services
Creating a decision service typically requires early decisions about business data modeling, evaluation flow, and rules, often before policy intent is fully understood. This can introduce unnecessary complexity during the initial analysis phase. The Decision Assistant is designed specifically for discovery: interpreting policy and regulatory text, identifying decision points, and proposing an initial organization of decision logic. It prepares decision artifacts that can be refined and implemented appropriate target environment (ODM/DI). After discovery, teams proceed along one of two paths:
-
DI path: modeling continues in Decision Designer, with build and execution in Decision Intelligence.
-
ODM path: Decision Assistant prepares ODM-aligned artifacts that are imported into IBM Operational Decision Manager, where business users and IT collaborate to model, test, govern, and deploy the decision service.
/odm Mode — Targeting ODM During Discovery
Using /odm at the beginning of a prompt indicates that IBM Operational Decision Manager (ODM) is the intended target environment for implementation after discovery.
In this mode, Decision Assistant structures discovery outputs using ODM concepts, including rule packages, ruleflows, and decision operations, as well as execution semantics such as task orchestration and rule sequencing. ODM is treated as the collaborative platform where decision services are modeled, governed, tested, and deployed after discovery.
Start with “/odm Approve a loan”
Figure 1 – Using “/odm” for the initial prompt
In response, the assistant will interpret your prompt in the context of ODM and generate:
Why this matters: ODM’s execution discipline is orchestrated by Ruleflows (start/end nodes, rule tasks, transitions, and ruleset parameters) and exposed to clients via Decision Operations (the callable interface of your ruleset). Starting with /odm makes these targets explicit from the outset.
Workspace Structure and ODM Alignment
When operating in ODM mode, Decision Assistant organizes discovery results using ODM-specific semantics across three tabs: Policy, Data Model, and Task Model.
-
The Task Model directly informs the ODM Ruleflow, which controls how, when, and in what order rules execute.
The Task Model in Decision Assistant
The Task Model is the central discovery artifact when targeting ODM. It serves two purposes: organizing rule logic into packages and describing the intended evaluation flow of the decision.
First, the Task Model defines the logical structure of the rule library. Each task represents a coherent group of rules, corresponding to rule packages in ODM.
Second, the Task Model defines the intended orchestration of decision evaluation, represented in ODM as a ruleflow. This includes task sequencing, rule selection, and conditional transitions between tasks.
Visual Representations - Two View Modes in the Task Model Tab
To balance explainability and governance-ready output, the Ruleflow View presents tasks, transitions, and branching to support review of execution intent during discovery. The Artifacts View provides a read-only summary of rules by package, focusing on policy intent rather than execution details.
Ruleflow View (graphical)
Figure 2 – Ruleflow View
The Artifacts View
-
A compact, read‑only list that groups items by package and shows only two fields for each artifact:
- At the top of each package block, the view displays the package name and a count of Artifacts: N. No additional metadata (parameters, filters, signatures) is shown here; those remain visible in the Ruleflow View or in the exported specification.
Figure 3 – Artifacts View
Finalize and export the specification
After reviewing the Policy, Data Model, and Task Model, Decision Assistant generates a JSON specification intended for importing into ODM. The specification includes rule package structure, ruleflow definitions, and decision operation metadata.
What happens in this step
Figure 4 – Finalize & export the specification
Apply the specification in ODM Decision Center (Swagger UI)
The specification can be applied using ODM’s REST API (for example, via Swagger UI). Successful application indicates that discovery artifacts are ready for further modeling and governance.
Figure 5 – Apply specification using IBM ODM Decision Center API via Swagger
Once imported, decision artifacts are collaboratively evolved in ODM. This includes refinement, testing, simulation, governance, and preparation for deployment.
With /odm, Decision Assistant supports the discovery phase of decision services by helping teams move from policy analysis to structured decision artifacts. When targeting ODM, these artifacts provide a starting point for collaborative modeling, governance, and deployment within IBM Operational Decision Manager.