NOMOS Decision Hub

Every decision leaves a trail. NOMOS traces each choice to its causal roots, stress-tests assumptions until they break, and rebuilds rankings from the survivors — a deterministic decision engine where nothing is hidden and everything is auditable. Validated by Singapore IMDA AI Verify at 95/100.

A deterministic, auditable decision orchestration layer built on causal reasoning — no probabilistic black boxes. Every decision decomposes into traceable causal topologies, supporting compliance-grade audit trails, scenario stress testing, and human-in-the-loop governance.

Trigger Scenarios

Automatically activates when users ask about:

• Deterministic or auditable decision engine design
• Causal counterfactual analysis and root cause tracing
• AI decision compliance auditing (IMDA/EU standards)
• Decision scenario stress testing and robustness analysis
• Human-in-the-loop governance for high-stakes decisions
• Enterprise decision system deployment

Core Capabilities

• Deterministic Evaluation: Hard-constraint thresholds plus explicit soft-constraint penalties — no hidden score adjustments
• Granular Algorithm Auditing: Every operation generates hash-chained audit events; any modification breaks verification
• Causal Counterfactual Re-selection: When assumptions fail, automatically computes transitive failure closure and re-selects surviving candidates
• Scenario Stress Testing: Supports declarative metric overrides, evidence gaps, and assumption failure simulation
• Reverse Root Cause Tracing: Traces observed deviations back to failed assumption points, outputting root cause hypotheses with causal chains
• Sensitivity & Robustness Analysis: Computes Pareto frontier, identifies fragile criteria, outputs ranking stability scores
• Enterprise Deployment: Docker containerization, PostgreSQL persistence, OIDC authentication

Usage

Basic Decision Analysis

python

from second_perspective import IntelligentDecisionHub
from second_perspective.models import HubAnalysisRequest
request = HubAnalysisRequest.model_validate({
    "decision": decision_payload,  # See examples/market_entry.json
    "scenarios": [
        {"id": "SC1", "name": "Key assumption failure",
         "failed_assumption_ids": ["A1"]},
        {"id": "SC2", "name": "Cost shock",
         "metric_overrides": {"S2": {"capital_required": 6000000}}},
    ],
})
report = IntelligentDecisionHub().analyze(request)
print(report.model_dump_json(indent=2))

Enterprise Deployment

bash

docker build -t nomos-hub .
docker run -p 8000:8000 \
  -e SP_ENV=production \
  -e SP_API_KEY=your-secret-key \
  -e SP_DATABASE_DSN=postgresql://user:pass@db:5432/nomos \
  nomos-hub

Deterministic Contract (Invariants)

• The engine never guesses missing weights, evidence, thresholds, or authorization relationships
• Hard constraints gate eligibility; soft constraints must declare explicit penalties — no silent score changes
• All behavioral policies carry policy_id and version, embedded in results
• Evidence quality is assessed by designated responsible nodes — the engine never fabricates credibility
• Assumption failures propagate along dependency graphs, explicitly naming affected alternatives
• Output is always "the leading candidate under declared inputs" — final authority rests outside the algorithm
• Approver name and authorization_ref must match the anchored decision owner
• All evaluations and approvals are hash-linked new revisions — no silent overrides

License

Personal non-commercial research use only. Government/enterprise commercial use requires written authorization.