Deterministic · Auditable · Resilient
Enterprise governance layer for BFSI, GCC, BPM, and AI‑enabled operations
The Neelkanth Protocol is not an application – it is a governance operating system. It replaces institutional trust with verifiable, deterministic logic. Built on the #MehendiPrinciple, #LongewalaDoctrine, and #GumnamKuheliPrinciple, this pilot edition transforms compliance from reactive reporting into autonomous Governance‑as‑Code.
| Locked Baseline | Value |
|---|---|
| Total FTEs | 713 (611 operational, 102 governance, 143 reserve) |
| Lines of Business | 31 |
| Governance Ratio | 1:6 |
| Audit Completeness | 99.5% (target 99.9%) |
| Logic Drift Threshold | <1.5% |
| Regulatory Breaches | 0 |
| Cumulative Value Protected | ₹269.6 Cr (simulation + baseline) |
The protocol does not ask for trust – it provides verifiable evidence.
Every decision carries an immutable Trace‑ID, every high‑risk action requires a HAT‑ID, and every control leaves an audit trail that can be reconstructed in under 6 minutes.
| Phase | Milestone | Core Focus |
|---|---|---|
| Phase 1 | Hackathon Genesis – Autonomous Sentinel | Proved deterministic governance vs. AI black box |
| Phase 2 | Neelkanth Protocol V4.4 – HAT‑ID / Trace‑ID | Standardised auditability and identity mapping |
| Phase 3 | Delta Remediation – Audit Hardening | Closed gaps in LGI, Drift, and Cross‑LOB correlation |
| Phase 4 | Shadow Sentinel – Failover Resilience | Defined AI‑free governance for system crashes |
| Phase 5 | Current 48‑hour Sprint – Industry Prototyping | Operationalising evidence‑based reporting |
| # | Function | FTEs | HAT‑ID Range |
|---|---|---|---|
| 1 | ALPHA – Intelligence & Risk Analytics | 15 | HAT-ALPHA-001..015 |
| 2 | BETA – Operations Governance & Orchestration | 15 | HAT-BETA-001..015 |
| 3 | CHARLIE – Governance Engineering | 20 | HAT-CHARLIE-001..020 |
| 4 | DELTA – Audit & Reconciliation | 15 | HAT-DELTA-001..015 |
| 5 | Quality Excellence (QA/QC) | 9 | HAT-QA-001..009 |
| 6 | Talent & Training Academy | 7 | HAT-TT-001..007 |
| 7 | HR Tactical Cell | 6 | HAT-HR-001..006 |
| 8 | PMO | 4 | HAT-PMO-001..004 |
| 9 | WFM – Workforce Management | 4 | HAT-WFM-001..004 |
| 10 | MIS – Management Information Systems | 3 | HAT-MIS-001..003 |
| 11 | IT, AI & Compliance Team | 4 | HAT-IT-001..004 |
| 12 | QRT – Quick Response Team (from reserve) | 8 | HAT-QRT-001..008 |
| 13 | ERT – Emergency Response Team (from reserve) | 5 | HAT-ERT-001..005 |
Total Governance FTEs (1–11): 102 ✅
Resilience reserve (QRT + ERT + general): 143 FTEs – activated on demand.
| Role | Key Performance Indicators |
|---|---|
| TIGER | Strategic drift, enterprise heatmaps, kill‑switch override integrity |
| ALPHA | EWS accuracy >95%, false positives <5%, LGI anomaly detection |
| BETA | Queue depth <50, SLA >99%, reserve utilisation trend (<20% MoM) |
| CHARLIE | Logic drift <1.5%, detection <5 min, Sentinel uptime >99.9% |
| DELTA | Audit completeness >99.5%, HAT coverage 100%, Trace‑ID >99.9% |
| LGI | Name | Target |
|---|---|---|
| LGI‑01 | HAT Re‑validation Rate | <2% |
| LGI‑02 | Kill‑Switch Readiness Score | 100% |
| LGI‑03 | Drift Precursor Anomaly Count | Alert if >5/hour |
| LGI‑04 | Reserve Utilisation Trend | <20% MoM |
| LGI‑05 | Trace‑ID Freshness | <2 hours |
Full LGI library: /governance/leading-indicators.md
| # | Scenario | Governance Domain |
|---|---|---|
| 1 | AML Risk Surge | Compliance |
| 2 | Workforce Governance Failure (Maternity Leave) | HR |
| 3 | SME Attrition Event | Knowledge Preservation |
| 4 | Knowledge Transfer Failure | Institutional Memory |
| 5 | AI Logic Drift | AI Governance |
| 6 | Multi‑LOB Operational Backlog | Operations |
| 7 | Regulatory Shock Event | Regulatory |
| 8 | Enterprise Black Swan Event | Resilience |
Each scenario includes audit templates, escalation matrices, drift thresholds, and success criteria (see /scenarios/).
Continuous Drift Sweep (Scenario 5 – Silent Pulse): The Sentinel runs a background, continuous logic drift detection independent of scenario triggers. This is the #GumnamKuheliPrinciple operationalised at the earliest possible detection point.
As of June 2026, India has access to Anthropic’s Claude Mythos – a specialised AI model for codebase vulnerability scanning. Integrated into the governance stack:
| Function | Responsibility |
|---|---|
| IT, AI & Compliance Team | Weekly 100% codebase scan |
| CHARLIE | Monitors security posture drift |
| DELTA | Logs every remediation with Trace‑ID + HAT |
| ALPHA | Feeds vulnerability trends into EWS |
| QRT/ERT | Activated for P1 incidents not fixed in <4 hours |
KPIs: 100% scan coverage · P1 remediation <4h · zero critical vulnerabilities outstanding.
| Condition | Threshold | Action |
|---|---|---|
| Queue depth | >50 for 2 consecutive cycles (10 min) | Pause intake; notify TIGER |
| Logic drift | >2.0% | Pause LOB; escalate to TIGER |
| Audit completeness | <95% for 2 consecutive hours | Freeze all new transactions |
| Trace‑ID coverage | <99% for any LOB over 1 hour | Pause that LOB |
| HAT compliance | <100% for high‑risk decisions (rolling 24h) | Block further high‑risk actions |
“Kill‑switch not triggered” is not a failure – it is governance discipline.
Full specification:/docs/kill-switch-specification.md
| Trigger | Reserve Deployed |
|---|---|
| Queue depth >50 for 5 minutes | 10 general reserve FTEs |
| Unplanned FTE absence (>20% of critical LOB) | Match headcount from general reserve |
| Kill‑switch activation | QRT + 20 general reserve; ERT on standby |
| Logic drift >1.5% and rising | QRT only |
Full specification: /docs/reserve-activation-specification.md
Grade A+ (DELTA certified) – deterministic, AI‑free fallback that activates when primary Sentinel fails.
- Triggers: Primary Sentinel heartbeat loss >30s, audit completeness <95% for 2 cycles, TIGER manual override.
- Capabilities: Mirrors drift detection, HAT attestation, Trace‑ID generation using hardcoded thresholds.
- HAT‑ID range:
HAT-SHADOW-001…(audit‑distinguishable). - Proven: Correctly fired in Scenario 2 (Workforce Governance Failure).
Read the specification: /docs/shadow-sentinel-spec.md
| Principle | Enterprise Application |
|---|---|
| Management by Objectives (MBO) | KPI/KRA‑driven governance |
| PDCA (Plan‑Do‑Check‑Act) | Continuous improvement cycle |
| Lean Management | Waste reduction and process efficiency |
| Six Sigma | Error reduction and quality excellence |
| Management by Exception | Escalation‑driven governance |
| Risk‑Based Management | Resource prioritisation |
| Systems Thinking | Cross‑LOB dependency management |
| Human‑in‑the‑Loop Governance | HAT‑ID accountability |
| Balanced Scorecard | Strategic KPI alignment |
| Resilience Engineering | Recovery and continuity management |
The Neelkanth Protocol governs behaviour without replacing core operating systems:
- 🏦 Banking & Financial Services – AML, KYC, FCC, Payments, Lending, Treasury, Trade Ops, Retirement
- 🛡️ Insurance – Claims, Underwriting, Fraud, Policy Admin
- 🏠 Mortgage – Origination, Servicing, Collections, QA
- 🌍 GCC / GIC Operations – Finance, HR Shared Services, Procurement, Customer Ops, Transformation
- 📊 BPM / BPO – Workforce Governance, Queue Management, Capacity Planning, Knowledge Management
- 🤖 AI‑Enabled Enterprises – Agent Governance, Model Governance, Logic Drift Monitoring, Explainability
Traditional Models: FTE Billing / Hourly Billing / Transaction Billing (scale by headcount)
Outcome‑Based Governance Model: Measurable, auditable, contractible capabilities.
| Governance Outcome | Target |
|---|---|
| Audit Completeness | >99.5% |
| Logic Drift | <1.5% |
| Regulatory Breaches | Zero |
| HAT Coverage | 100% |
| SLA Compliance | >99% |
Artefacts generated: Trace‑IDs, HAT‑IDs, Audit Trails, Logic Drift Scores, Kill‑Switch Logs, Resilience Activation Logs, Evidence Packages.
| Layer | Technology |
|---|---|
| Operating System | Ubuntu 24.04 LTS (also runs on Windows, macOS, Azure Linux, AWS Ubuntu) |
| Language | Python 3.12 |
| UI | Streamlit |
| API Layer | FastAPI |
| Database | SQLite (portable, file‑based) |
| Version Control | GitHub |
| CI/CD | GitHub Actions |
| Deployment | Streamlit Cloud (Pilot) / Any cloud (Enterprise) |
The pilot is platform‑agnostic – runs locally or on any cloud without external dependencies.
| Capability | Pilot (V4.4) | Enterprise (V4.5+) |
|---|---|---|
| Audit Store | SQLite (file‑based) | PostgreSQL / BigQuery |
| Analytics | Python rules (deterministic) | Advanced analytics (ML‑assisted) |
| Drift Detection | Deterministic thresholds (<1.5%, 2.0%, etc.) | AI‑assisted detection (optional models) |
| Policy Intelligence | Local JSON / markdown rules | Microsoft Foundry IQ / Vertex AI Search |
| Vulnerability Intelligence | Manual / static analysis | Claude Mythos integration (API) |
| Deployment | Local / Streamlit Cloud | Enterprise multi‑cloud (Azure, AWS, GCP) |
| Area | Status |
|---|---|
| Architecture Design | ✅ Completed |
| Enterprise Simulation (3,000 days) | ✅ Completed |
| Governance Validation | ✅ Completed |
| Demonstration Readiness | ✅ Approved (June 8, 2026) |
| Production Readiness | 🟡 Organisation‑specific (requires deployment validation) |
DELTA final certification (June 8, 2026):
“The pilot repository is mathematically complete. Governance Integrity Score = 5.0/5.0.”
Audit report: /audit/delta-final-certification.md
neelkanth-enterprise-pilot/ ├── README.md (this file) ├── LICENSE (MIT) ├── enterprise-baseline/ │ └── baseline-v4.4-locked.json ├── governance/ │ ├── support-functions-kpi-kra-v4.4.md │ ├── hat-register-v4.4.json │ ├── leading-indicators.md │ └── ewi-heatmap.md ├── sentinel/ │ └── sentinel-config-v4.4.yaml ├── scenarios/ │ ├── scenario-1-aml-risk-surge.md │ ├── scenario-2-workforce-failure.md │ ├── scenario-3-sme-attrition.md │ ├── scenario-4-knowledge-transfer.md │ ├── scenario-5-ai-logic-drift.md │ ├── scenario-5-silent-pulse.md │ ├── scenario-6-multi-lob-backlog.md │ ├── scenario-7-regulatory-shock.md │ └── scenario-8-black-swan.md ├── audit/ │ ├── delta-final-certification.md │ ├── court-martial-record.md │ └── pilot-readiness-certification.md ├── dashboard/ │ └── (RBAC role views – TIGER, ALPHA, BETA, CHARLIE, DELTA) ├── engine/ │ └── (scenario engine – 8 interactive scenarios) ├── exports/ │ └── (evidence generator – PDF/CSV reports) ├── docs/ │ ├── architecture-overview.md │ ├── governance-hierarchy.md │ ├── kill-switch-specification.md │ ├── reserve-activation-specification.md │ ├── shadow-sentinel-spec.md │ ├── cross-lob-correlation-model.md │ └── drift-to-resilience-loop.md └── demo/ └── walkthrough.md (15‑min demonstration script)
text
Note: This repository is design‑ and documentation‑ready for pilot execution. The live prototype (hackathon) is at
neelkanth-logic-drift-sentinel.
| Repository | Purpose | Visibility |
|---|---|---|
neelkanth-logic-drift-sentinel |
Hackathon prototype – 20 LOBs, Streamlit demo, proof of concept | Public |
neelkanth-enterprise-pilot (this repo) |
Enterprise V4.4 – 31 LOBs, deterministic governance, pilot baseline | Private (can be opened for industry pilots) |
The pilot edition scales, hardens, and production‑readies the validated concepts from the hackathon.
- #MehendiPrinciple – Human accountability remains indelible. Every high‑risk decision requires
HAT‑IDandTrace‑ID. - #LongewalaDoctrine – Scale without losing command. Maintain 1:6 governance ratio.
- #GumnamKuheliPrinciple – Undetected drift is the highest risk. Real‑time drift detection is continuous.
MIT – see LICENSE file.
Maintained by Dipankar Kundu – Neelkanth Governance Initiative
For inquiries: LinkedIn
#MehendiPrinciple | #LongewalaDoctrine | #GumnamKuheliPrinciple | #NeelkanthProtocol