Requirements AI

Sections in this article

Requirements AI is designed to help systems engineers and project teams accelerate the creation and refinement of requirement entities in Innoslate. Leveraging advanced large language model (LLM) technology and reasoning models, it reduces manual drafting effort, contains consistent language, and aligns requirements with industry best practices, primarily driven by INCOSE's Guide To Writing Requirements v4 developed by the INCOSE Requirements Working Group.

This highly requested AI capability is important because it automates the initial generation of requirements, saving significant time in early project phases, while also producing requirement statements that follow structured language patterns to reduce ambiguity. It supports scalability by enabling the rapid creation of requirement sets for large and complex projects, and it maintains traceability by embedding generated requirements within the Innoslate database.

Generating Requirements from New

Requirements AI allows users to create a defined set of new requirement entities from within any Document or Compilation view.

Step 1: Accessing the Generator

To begin, open a Document or Compilation and select "Generate Requirements" from the "More" dropdown menu.

Step 2: Providing Context

The Setup phase prompts you to provide background information that Requirements AI will use to generate draft requirements.

• Target Entity On (Default): When enabled, Requirements AI uses the targeted entity as context for generation. This target function automatically creates a "traces/traced to" relationship to each parent requirement automatically for traceability.

• Target Entity Off: When disabled, users can enter an open-ended prompt to be used as context for their requirements set.

Step 3: Configuring Parent and Child Requirements

After context is set, the Configure phase allows users to define the number of Parent and Child requirements that will be generated. This provides control over the depth and breadth of the resulting requirement set.

Note: The sliders automatically adjust to show the allowable range of parent and child requirements for quality assurance.

Step 4: Verifying Generated Requirements

The Verify phase displays a condensed list of all proposed requirements. Parent requirements are shown in bold, while child requirements appear unbolded beneath their parent. Users should carefully review each entry to confirm accuracy and completeness.

Expanding a requirement shows the generated information in full detail, including:

• Number

• Name

• Requirement Statement

• Rationale

• Label (e.g., Functional, Environmental, Interface, Performance, Reliability, Safety, and Verification Requirement)

Step 5: Finalizing Requirements

Once all requirements are validated, selecting "Create" button embeds them directly into the active document. These new requirements appear alongside manually created entries and can be managed using Innoslate’s full suite of requirement management tools.

 

Requirements Expander — Sibling Requirements

The Requirements Expander feature allows users to generate additional sibling requirements at the same hierarchical level as an existing requirement. This is particularly helpful when you want to broaden coverage of related system capabilities without manually drafting each high or level level requirement.

⚠️ Important: The context for sibling and child expansions is specialized and depends on the selected requirement’s position in the hierarchy. The system intelligently fetches relevant context and adjusts prompts to ensure good quality and non-duplicative generation.

Step 1: Accessing the Expander

To begin, select a requirement in Documents View and choose "Expand Requirements" from the "More" dropdown. This option is only available when a requirement entity is selected.

Step 2: Context Phase

In the Setup phase, Requirements AI automatically gathers context from the selected requirement and its siblings. Existing sibling requirements are included in the context. This information is shared with the agent to generate accurate, relevant, and non-duplicate requirements. 

Step 3: Configuring the Number of Siblings

The Configure phase prompts the user to select the number of sibling requirements to generate. A single slider provides control over how many new siblings will be created.

Step 4: Verification Phase

Proposed sibling requirements are presented in the Verify stage. As always, users are encouraged to review each draft carefully, checking the requirement statement, rationale, and requirement label.

Step 5: Final Output

Once approved, the new sibling requirements are created and displayed alongside the original requirement in the document, maintaining the correct hierarchical structure.

 

Requirements Expander — Child Requirements

The Requirements Expander can also generate child requirements, which sit directly beneath a selected parent requirement. This allows users to break down a high-level statement into more detailed, verifiable sub-requirements.

Step 1: Accessing the Expander

Select a requirement in Documents View and choose "Expand Requirements" from the "More" dropdown. This option appears only when a requirement is selected.

Step 2: Context Phase

In the Setup phase, the expander automatically pulls the selected parent requirement into the context window. This text serves as the foundation for generating meaningful and traceable child requirements.

Step 3: Configuring the Number of Children

Next, the Configure phase allows users to select the number of child requirements to generate. The slider provides control over how granular the decomposition will be in one session.

Step 4: Verification Phase

Generated child requirements are displayed in the Verify stage. Each draft includes a name, requirement statement, rationale, and requirement label designed to provide greater specificity at a lower level, aware of the parent requirement.

Step 5: Final Output

Once validated, the new child requirements are created beneath the selected parent and displayed in the document hierarchy. These children become part of the structured requirement tree, supporting decomposition and traceability.

To continue learning about the SE Lifecycle Agents, Click Here.

(Next Article: Test Case AI)