AI extracts. Rules evaluate.
Responsible AI
AI helps interpret recruitment information. It does not become the legal authority or silently decide whether a citizen qualifies.
One decision screen · three non-interchangeable rolesAI stops before publication. Deterministic code cannot grant consent. A human cannot silently rewrite a captured Passport.
1 · AI may propose
AI stops here- Notification rule structure
- Unfamiliar field terminology
- Document field suggestions
- Plain-language and Hindi explanations
- Grounded answers with page sources
2 · Deterministic rules decide
No model in this path- Age on a cut-off date
- Category relaxation
- Degree and subject match
- Certificate validity
- Readiness and blocking errors
3 · Humans must approve
Mandatory authority- Reviewer publishes proposed rules
- Citizen confirms source choice
- Citizen approves what is shared
- Citizen accepts declaration
- Official body controls real submission
NotificationText + page numbers
→AI extractionProposed rules
→Human reviewApprove, edit, reject
→Rules engineDeterministic result
→Citizen viewReason + source
The official recruitment notification remains authoritative.Only approved structured rules enter eligibility evaluation.A citizen can inspect every rule, value and source.Saved demo responses keep the judge path reliable without an API key.AI failures fall back safely; they never block the citizen demo.