Trust · Responsible AI
AI That Recommends. People Who Decide.
The principles that shape every AI feature in the Prakura ecosystem.
Principles
Eight principles
Human oversight
People make consequential decisions; AI recommends.
Transparency
AI-generated content and suggestions are labelled.
Privacy
Minimum data, clear purpose, user control.
Explainability
Recommendations show the factors behind them.
Fairness
Outputs are reviewed for unintended bias.
Safety
Guardrails against harmful or off-topic output.
Evaluation
Features are tested against evaluation sets before release.
Accountability
Decisions are logged with who made them.
AI decision UX
Recommendation → Review → Decision
Step 1
AI Recommendation
Suggestion with reasons and confidence
Step 2
Human Review
A person checks, edits or rejects
Step 3
Final Decision
Made and owned by a person, and logged
Recruitment
AI: Summarises candidate evidence and suggests interview focus areas.
Person: Recruiter decides who advances; no automatic rejection.
Placement
AI: Estimates readiness and recommends practice.
Person: Placement officer reviews readiness before sharing with recruiters.
Employee performance
AI: Drafts check-in summaries from goals and tasks.
Person: Manager edits and owns every review.
Candidate ranking
AI: Orders candidates by explained, job-relevant factors.
Person: Recruiter can reorder, and ranking is never final on its own.