# HDP AI Governance, Model Risk & Responsible Automation

Snapshot hash: `1877f61e540b272467bf5b7d479a9d9e92c318b717bff328dc7a587d750bf9ed`

## Readiness

- Score: 42/100 — BLOCKED
- AI systems: 4 (3 active or planned; 1 High/Critical)
- Approved use cases: 1
- Approved evaluations: 0
- Approved oversight policies: 1
- Open incidents: 0
- AI-control coverage: 0%

**No independent assurance, certification, or regulatory conformity is claimed.**

## Blockers
- APPROVED_USE_MISSING: HDP Voice Agent Realtime Assistant has no approved use case.
- APPROVED_EVALUATION_MISSING: HDP Voice Agent Realtime Assistant has no approved evaluation run.
- HUMAN_OVERSIGHT_MISSING: HDP Voice Agent Realtime Assistant lacks an approved human-oversight policy.

## Warnings
- APPROVED_EVALUATION_MISSING: Local Deterministic Executive Council has no approved evaluation run.
- APPROVED_USE_MISSING: HDP Predictive Lead Scoring has no approved use case.
- APPROVED_EVALUATION_MISSING: HDP Predictive Lead Scoring has no approved evaluation run.
- NO_APPROVED_AI_ATTESTATIONS: No independently reviewed AI-control attestation is approved.

## AI System Inventory

### Local Deterministic Executive Council
- Status: ACTIVE_LOCAL
- Risk: MINIMAL (0)
- Provider/model: HDP LOCAL / hdp-governed-council 1.6
- Owner: role_product_lead
- Intended uses: Generate reproducible management perspectives from frozen evidence records; Exercise governance workflows without external model cost
- Prohibited uses: Approve its own synthesis; Authorize spending; Dispatch external actions; Represent output as independent professional advice

### OpenAI-Compatible Executive Council Adapter
- Status: DISABLED
- Risk: LIMITED (20)
- Provider/model: OPENAI_COMPATIBLE / environment-configured unactivated
- Owner: role_product_lead
- Intended uses: Draft competing internal executive recommendations from approved evidence snapshots
- Prohibited uses: Unsupervised production decisions; Autonomous external communications; Self-approval; Processing unapproved sensitive data

### HDP Voice Agent Realtime Assistant
- Status: PILOT_BOUNDARY
- Risk: CRITICAL (90)
- Provider/model: CONFIGURABLE / runtime-configured pilot-boundary
- Owner: role_product_lead
- Intended uses: Approved inbound intake; Qualification; Appointment support; Escalation and logging
- Prohibited uses: Unattended mass outbound dialing; Material promises; Legal, medical, financial, or regulated determinations; Ignoring consent and escalation rules

### HDP Predictive Lead Scoring
- Status: PLANNED
- Risk: LIMITED (20)
- Provider/model: HDP ML DEPARTMENT / lead-readiness-score planned
- Owner: role_product_lead
- Intended uses: Prioritize business outreach using approved business data
- Prohibited uses: Employment, credit, housing, insurance, or protected-class decisions; Automated rejection without human review