Using the Decision Assistant

The Decision Assistant is a chat-based interactive tool that helps accelerate decision-making. It provides guidance and automation throughout the creation and management if decision services using natural language.

Building and managing decision services

The Decision Assistant acts as a conversational interface for Decision Designer. It helps you build, understand, and manage decision services through an intuitive, multi-turn interaction model.

Because it maintains context across exchanges, the assistant supports complex, step-by-step workflows. Each interaction builds on the previous one, allowing you to progressively refine decision artifacts without restarting your work.

The Decision Assistant provides a comprehensive set of capabilities to support the full lifecycle of decision services:

Explaining decision services
Analyze decision services and provide summaries to help you understand their purpose and behavior.
Validating decision artifacts
Check decision models and data models for errors to ensure consistency and correctness.
Testing decision models
Generate test data sets and run decision models to validate logic and expected outcomes.
Managing dependencies
Create and maintain relationships between decision artifacts to support modular and reusable designs.
Creating decision operations
Define operations that expose decision models for execution.
Building decision services
Create decision services and related artifacts, including decision models, data models, and samples.
Defining the decision logic
Author business rules and decision tables within decision models to represent business policies.

Building decision services from scratch with the Build mode

In addition to its core capabilities, the Decision Assistant offers a dedicated Build mode for interpreting policy descriptions that are written in natural language and translating them into business rules.

With Build mode, you can generate a complete decision service in minutes by simply describing the decision you want to automate. For more precise results, you can also provide detailed policy documents directly in the chat.

The generated decision service contains:

  • A data model: A collection of data types representing the information required for your decision.
  • A decision implementation, which can be one of the following:
    • A decision model: Uses a decision diagram to define logic, rules, and dependencies between inputs and decisions.
    • A task model: Defines decision logic outside of a diagram and organizes it into business rules and ruleflows that control the sequence and conditions for executing tasks.

Starting from basic models, the assistant builds them step-by-step as you interact with it by adding data types, defining the structure of the decision, and authoring the decision logic that drives your decision.

When your decision service is complete, you can continue to refine it in Decision Designer, our fully fledged modeling environment. From Decision Designer, you will also be able to test, build, and deploy your decision service.

Decision services with task models can be as exported JSON specifications and used in IBM Operational Decision Manager. After you export the decision service, you use a REST API to apply the specification and create the corresponding decision service in Decision Center in Operational Decision Manager.