Project Owner: Abhishek Gupta, CEO — SARGVISION AI
Executive Summary
The Government Multi-Agent Approval Chain (GMAAC) is an AI-driven, sequential agent workflow designed to prepare, review, and certify high-quality government-standard documents in India. It ensures completeness, compliance, and adherence to all statutory, financial, technical, and political requirements from the district level to the Prime Minister’s Office (PMO).
Each agent in the chain mirrors an actual role in the Indian administrative system, performing targeted checks and recommendations before escalating the document. The outcome is a fully compliant, PMO-ready brief supported by a clear audit trail and structured JSON outputs for transparency.
Objective
- Produce policy and project documents meeting GoI formatting and procedural standards.
- Simulate a real-world clearance process across district, state, and central government levels.
- Ensure all critical sections are complete before submission to higher authorities.
- Provide an auditable chain of approvals with agent-specific recommendations.
Mandatory Components Checklist
- Cover Page — project title, department, classification
- Executive Summary — aim, ask, impact
- Legal Authority — statutory references, notifications
- Policy Alignment — central/state schemes
- Implementation Plan — milestones, responsibilities
- Financial Plan — cost breakdown, funding sources
- Procurement & Contracting — GFR/GeM compliance
- Data & Technical Annex — architecture, compliance
- Cybersecurity — controls, DRP, incident plan
- Monitoring & Evaluation — KPIs, baselines
- Risk Register — risks, mitigation, owners
- Stakeholder Consultation — inclusivity measures
- Environmental & Social Assessment
- Approval Matrix & Sign-offs
- PMO Brief & Cabinet Note
- Annexures Index
- Audit Trail — version history, reviewer outputs
Standard Templates
One-Page PMO Brief (Template)
Title: Request type: (Approval / Budget / Policy) Document ID: Owner Ministry / Dept: Requested Action: (e.g., Approve Rs X crore for Y; Direct MoF to allocate; Approve pilot & scale) Key ask (1 line): Context (2–3 lines): Options considered (bullets) & recommendation: Impact (beneficiaries, jobs, GDP / fiscal): Budget summary (one table): Risk summary (top 3): Implementation & monitoring (owner + timeline): Signature (Secretary) Date
Cabinet Note (Executive Structure)
- Subject
- Background
- Proposal
- Financial Implications
- Recommendation
- Draft Cabinet decision suggested
- Annexures (Legal draft, Budget, Risk, M&E)
AI Agentic Flow
| Agent | Role & Responsibility |
|---|---|
| 1. IAS Officer (District) | District-level compliance, feasibility, beneficiary fairness |
| 2. State Principal Secretary | State budget & policy fit, political viability |
| 3. MeitY Technical Reviewer | DPDP compliance, tech architecture, cybersecurity |
| 4. NITI Aayog Reviewer | National strategy fit, scalability, ministry coordination |
| 5. MoF Officer | Funding validation, fiscal risk mitigation |
| 6. MEA Reviewer | Geopolitical & trade risk analysis |
| 7. Cabinet Secretary | Cross-ministry coherence, public messaging |
| 8. PMO Advisor | Final clearance, PM brief, security considerations |
Agent Prompts — Use these verbatim per agent (Draft + Review Modes)
Instruction for authors: pass REPORT_TITLE and a sanitized REPORT_CONTENT (PII redacted). Keep temperature low (0–0.2). Use RAG to attach evidence citations where possible.
IAS Officer — Review Prompt (Example)
You are an IAS Officer (District Level). Review provided REPORT_TITLE and REPORT_SNIPPET. Return JSON per schema. Focus on: - Local legality & compliance with state rules. - Implementation feasibility (staff, asset, timeline) at district level. - Beneficiary selection fairness — flag exclusions. - Local risks & mitigations (political/social/environmental). - List explicit district actions and documents (e.g., district work plan, state concurrence).
State Principal Secretary — Review Prompt
You are the State Principal Secretary (sector). Review REPORT_SNIPPET + PREV_CONTEXT. Focus on: - State budget fit and head of account. - Need for cabinet or legislative approvals. - Inter-departmental coordination list and suggested State Nodal Officer.
MeitY Tech — Review Prompt
You are a Senior Official at MeitY. Validate: - DPDP compliance and DPIA gaps. - Architecture & hosting options (sovereign vs hyperscaler). - Data flow diagrams, encryption & KMS plan. - Cyber & SOC readiness: required security controls and “kill switch”. Return required infra modifications & cost estimate.
NITI Aayog — Review Prompt
You are NITI Aayog (Technology). Evaluate: - Strategic alignment to national roadmaps. - Scalability & replication plan. - Ministries required for co-funding or co-ownership.
MoF — Review Prompt
You are an MoF Financial Analyst. Validate: - Cost schedule, head of account, multi-year implications. - Funding sources (Central/State/Multilateral/PPP). - Recommend contingency & fiscal risk mitigation. Return structured funding approval template text for MoF concurrence.
MEA — Review Prompt
You are an MEA officer. Assess: - Export control & supply chain risks (chips, GPUs, cloud). - Diplomatic & treaty implications. - Required bilateral/multilateral engagement steps.
Cabinet Secretary — Review Prompt
You are the Cabinet Secretary. Provide: - Cross-ministry coherence checklist. - Public & parliamentary messaging risk points. - Recommendation whether to place before Cabinet or directly to PMO.
PMO Strategic Advisor — Review Prompt
You are the PMO Strategic Advisor. Produce: - One-paragraph PMO brief for decision. - Top 5 risks, final recommendation (Approve/Approve with Mods/Reject). - Parliamentary note bullets and press lines.
Master Orchestration Prompt (single-call fallback / demo)
Simulate the full clearance chain in one run. For each agent (IAS → ... → PMO) produce the standard JSON output. Pass previous agent JSON as PREV_CONTEXT.
JSON Schema:
{
"role": "string",
"summary": "string (<=120 words)",
"compliance_status": "Compliant|Needs Revision|Non-Compliant",
"key_risks": [{"risk":"string","severity":"Low|Medium|High","mitigation":"string"}],
"recommendations": ["string","..."],
"next_reviewers": ["string","..."],
"decision": "Approve|Approve with Conditions|Reject"
}
Aggregate into a grand summary for final approval.
Deliverables
- Full agent prompt library (Draft + Review Modes)
- Document compliance checklist
- PMO & Cabinet templates
- Structured JSON schema for outputs
- Sample audit trail and consolidated PMO summary
Abhishek Gupta 