Skip to main content

Best Practices for AI Features

AI Features

A guide to getting the most out of Pidima's AI-powered capabilities across all features.

When to Use Each Feature​

Requirements Assist — Autofill​

  • After bulk importing requirements from spreadsheets or Jira
  • When requirements lack domain/type/priority assignments
  • For consistency across large requirement sets

Requirements Assist — Enhance​

  • Vague or ambiguous requirements
  • Requirements lacking testable criteria
  • Non-compliant requirement formats

Requirements Assist — Make Atomic​

  • Requirements with multiple "and" statements
  • Complex requirements covering multiple functions
  • Requirements that are difficult to test completely

Requirements Assist — Enhance with Prompt​

  • Industry-specific formatting needs
  • Custom compliance requirements
  • Specialized technical language requirements

Requirement Intelligence (Specs Wizard)​

  • Creating an initial requirement set from compliance/specification documents
  • Populating a new project level from scratch
  • Deriving requirements from standards (e.g., ISO, IEC, IEEE)

Requirement Analysis​

  • Evaluating requirement quality against writing standards
  • Detecting conflicts between requirements before design phase
  • Finding and eliminating duplicate requirements after imports

Generate Tests​

  • After requirements are reviewed and approved
  • When updating test coverage for changed requirements
  • For regression test planning

Compliance Intelligence​

  • Before milestone reviews to verify regulatory compliance
  • After generating requirements to validate against source regulations
  • When regulations are updated and existing requirements need re-validation

Gap Analysis​

  • After completing a requirements phase to assess coverage
  • Before audits to identify traceability gaps
  • Periodically to monitor project health metrics

Impact Analysis​

  • After editing a requirement's description
  • When applying compliance suggestions that change requirement meaning
  • Before releasing to understand the scope of a change

AI Architecture Generation​

  • After requirements are stable for a level
  • When you need visual documentation of system structure
  • For IEEE 1016 compliance (use module-based generation)

AI Traceability Suggestions​

  • After creating test cases to find missing links
  • When requirements and test cases exist but aren't linked
  • To verify completeness of manual linking

Tips for Better Results​

1. Provide Context​

Upload relevant specification and compliance documents before using AI features. The AI uses these documents via RAG to produce more accurate, domain-specific outputs.

2. Fill Out Templates Thoroughly​

For Requirement Intelligence, the template questionnaire directly influences output quality. More specific answers produce more relevant requirements.

3. Review All AI Outputs​

AI suggestions should always be reviewed by domain experts. Use the "Under Review" status for generated artifacts and establish a review workflow.

4. Work Iteratively​

  • Run multiple passes for complex transformations
  • Generate → Review → Refine → Re-generate if needed
  • Use feedback (thumbs up/down) to signal quality

5. Batch Similar Items​

Process related requirements together for consistency. The sliding window approach in test generation works best when related requirements are in the same batch.

6. Use Custom Prompts​

Leverage "Enhance with Prompt" for organization-specific needs:

  • Specific formatting standards
  • Domain terminology requirements
  • Compliance language patterns

7. Combine Features in Sequence​

The most effective workflow chains multiple AI features:

  1. Requirement Intelligence → Generate initial requirements from documents
  2. Compliance Intelligence → Validate against regulations
  3. Requirement Analysis → Check quality and detect conflicts
  4. Generate Tests → Create test cases for approved requirements
  5. Gap Analysis → Verify coverage and traceability
  6. Impact Analysis → Manage ongoing changes

8. Keep Documents Current​

AI quality depends on your uploaded documents. When regulations or specifications are updated:

  • Upload new document versions
  • Re-run Compliance Intelligence against updated documents
  • Use Impact Analysis to propagate changes

9. Start Small, Scale Up​

When using a new AI feature for the first time:

  • Test with 3-5 requirements to understand the output quality
  • Adjust your approach based on results
  • Then scale to larger batches

10. Monitor Async Jobs​

Most AI operations run asynchronously. Check the jobs panel for:

  • Completion status
  • Partial failures (some items succeeded, others didn't)
  • Error messages that indicate how to fix issues