governance

Synthetic Enterprise Canvas Demo

Evidence-to-canvas review runs stored source-traceable

Tenant, data, model, and audit boundaries

Governance

Review surface for CLMBS production decisions: customer data separation, anonymized learning, retention, live model use, and access control.

Customer data separation

demo boundary
Scope
SQLite case id and source ids separate demo records.
LLM policy
Live calls require explicit OPENAI_API_KEY; no silent fallback.
Audit
Runs, errors, feedback, prompt overrides, and validation cases are persisted locally.

Anonymized learning

not enabled
Scope
No cross-customer learning pipeline is implemented.
LLM policy
Do not use client data for model improvement without approved anonymization.
Audit
Production needs approval records for any learning dataset.

Archival and retention

handoff needed
Scope
Review packs and manifests are generated on demand.
LLM policy
Generated output status must distinguish deterministic demo and live OpenAI runs.
Audit
Production needs retention, deletion, and export naming rules.

Implemented demo controls

internal demo
  • Synthetic demo data is labeled as internal demo material.
  • Live OpenAI calls fail loudly when OPENAI_API_KEY is absent.
  • Prompt overrides, validation cases, source intake, feedback, and runs are case-scoped in SQLite.
  • Export manifest separates implemented handoff artifacts from planned binary rendering.