Touli Rollout Plan, Risk Register, and Readiness Checklist

Purpose

Define phased rollout planning for AI verification, points, and blockchain modules, including risk controls and implementation kickoff readiness.

Phased Rollout Plan

Phase A: Points Foundation

Scope:

• Deterministic points engine design freeze
• config snapshot model
• ledger invariants and idempotency approach

Exit criteria:

• calculation contract approved
• invariants documented and test strategy defined
• schema delta for points domain approved

Phase B: Verification Foundation

Scope:

• rules-first verification architecture
• confidence routing
• manual review pathway

Exit criteria:

• state machine approved
• escalation SLA and reason code taxonomy approved
• data lifecycle controls specified

Phase C: AI + Labels Integration

Scope:

• vision adapter contract
• feature and label capture architecture
• cost and fallback policies

Exit criteria:

• provider interface approved
• outage fallback and retry policy approved
• labeling strategy approved

Phase D: Blockchain Anchoring

Scope:

• leaf schema and pseudonymization
• batch lifecycle and reconciliation model
• proof retrieval contract

Exit criteria:

• anchoring DLD approved
• replay/reconciliation policy approved
• privacy unlinkability controls approved

Phase E: Operational Hardening

Scope:

• observability SLOs
• runbooks and on-call readiness
• compliance checkpoint validation

Exit criteria:

• SLO dashboard definition approved
• runbooks accepted
• go/no-go checklist green

2-Week Architecture Finalization Timeline

Week 1:

• Day 1-2: HLD alignment workshop and decision log
• Day 3-4: DLD deep dives (verification + points)
• Day 5: schema delta and interface review

Week 2:

• Day 1-2: blockchain + compliance review
• Day 3: risk workshop and mitigation ownership
• Day 4: readiness review and sign-offs
• Day 5: architecture freeze package handoff

Risk Register

Risk ID	Risk	Likelihood	Impact	Mitigation	Owner
R1	Points logic drift between product docs and engine config	Medium	High	snapshot versioning + single source of truth review gate	Product + Architecture
R2	Duplicate credits from retries/concurrency	Medium	High	idempotency keys + unique constraints + replay-safe workers	Backend
R3	Manual review backlog exceeds SLA	High	Medium	queue capacity planning + degrade-to-safe routing + staffing fallback	Trust Ops
R4	AI provider outage/degradation	Medium	Medium	adapter abstraction + retry policy + manual fallback	Platform
R5	Residency/compliance policy mismatch by market	Medium	High	market policy engine + pre-launch legal checklist	Compliance
R6	On-chain anchoring delays or failed confirmations	Medium	Medium	async decoupling + reconciliation + replay attempts	Platform
R7	Erasure request does not fully break chain relinkability	Low	High	rotating pseudonyms + mapping destruction verification	Security + Compliance
R8	Schema expansion increases query latency	Medium	Medium	index strategy + read model projections + performance budget	Data
R9	Contract drift between API and event consumers	Medium	Medium	versioned schemas + consumer compatibility tests	Backend
R10	Scope creep before implementation start	High	Medium	architecture freeze and change-control process	PM + Architecture

Readiness Checklist (Go/No-Go for Dev Start)

A. Architecture Readiness

• [ ] HLD approved by architecture owner.
• [ ] DLDs approved for verification, points, and anchoring.
• [ ] ADR set approved and signed.
• [ ] Open questions resolved or logged with explicit assumptions.

B. Data and Contract Readiness

• [ ] Schema delta spec approved (proposal level).
• [ ] API contracts approved (v1).
• [ ] Event contracts approved with versioning rules.
• [ ] Idempotency and replay semantics documented per flow.

C. Compliance Readiness

• [ ] Age/consent policy matrix approved per launch market.
• [ ] Data retention/deletion control design approved.
• [ ] Human review and contestability pathway approved.
• [ ] Residency and processor boundary controls approved.

D. Operations Readiness

• [ ] SLOs defined for verification, points, and anchoring.
• [ ] Alerting and dashboard metric list approved.
• [ ] Runbooks drafted for queue backlog, AI outage, chain failure.
• [ ] Incident ownership and escalation tree defined.

E. Planning Readiness

• [ ] Phase milestones aligned with engineering capacity.
• [ ] Dependencies and blockers mapped.
• [ ] Risk register ownership accepted.
• [ ] Architecture freeze date committed.

Change Control After Freeze

• Any post-freeze changes require:
  • change request record
  • impact assessment (data, compliance, contracts, delivery)
  • architecture owner approval
  • updated ADR entry