Platform Docs · Governance

CTO Deployment Review

Formal deployment readiness assessment. Green / Amber / Red scoring across six governance categories. Updated March 2026.

Executive Summary

Entrestate is a live, multi-surface production system with coherent architecture, active release cadence, and a documented one-system model. The platform scores strongly on product deployment and architecture maturity. Operational transparency and SRE readiness are functional but not yet at enterprise audit standard — the primary gaps are the absence of automated uptime monitoring, published SLOs, and a formal on-call policy. Enterprise readiness is strong in documentation and auth but requires SOC 2 public attestation and a DPA to close the compliance gap.

Verdict: CONDITIONAL PASS — deploy-real, operationally functional, enterprise-readiness requires three targeted actions.

3 Green

3 Amber

Category Scores

Product Deployment Status

GREEN8 / 10

Production surface is live, multi-surface, and interconnected. Core modules — AI Copilot, Properties, Developers, Areas, Reports, Docs, Roadmap, Changelog — are all deployed and publicly reachable.

Evidence

7,015+ active UAE projects indexed and queryable
481 canonical developers with reliability scoring
AI Copilot streaming in production with cooldown-based rate limiting
Report generation pipeline persisting to Neon assistant_reports table
Public changelog with entries through Feb 2026

Gaps

No formal SLA publication (uptime target, error budget)
Status page exists but lacks automated service probe data

Action: Connect /status to real health probe endpoints. Publish a formal SLA for enterprise and partner tiers.

Release Discipline

GREEN7 / 10

Active versioned release cadence with documented changelog. Vercel's PR preview + branch deployment model is in place. Recent releases show consistent shipping across Dec 2025–Feb 2026.

Evidence

Vercel deployment pipeline with per-PR preview environments
Post-deploy smoke scripts exist (/scripts/smoke.ts)
Changelog lists monthly feature batches with specifics
Prisma migration history provides schema version trail

Gaps

No formal release notes template or sign-off gate
Rollback procedure not documented in-product (now addressed in /docs/deployment-architecture)
No automated integration test suite observed in CI

Action: Add a pre-merge checklist (lint, type-check, smoke) enforced as a required GitHub Actions status check.

Architecture Maturity

GREEN9 / 10

One-system model is genuinely senior-grade. The 10-phase pipeline → 5-layer evidence stack → 4-stage decision tunnel is architecturally coherent, documented, and implemented — not just described in marketing copy.

Evidence

10-phase pipeline with explicit phase outputs and consumers
L1–L5 evidence hierarchy with adjudication rules
Static Truth Finalization requiring 5 confirmed fields
Decision tunnel with Intent → Evidence → Judgment → Action stages
JSONB evidence payloads persisted on each report artifact

Gaps

No public architecture decision record (ADR) format
Mind-map and documentation exist but not cross-linked to source code

Action: Maintain this standard. Consider publishing an ADR log as decisions accumulate around AI model routing and data source adjudication.

Operational Transparency

AMBER5 / 10

Status page exists and shows live data snapshot. Incident log is maintained manually. However, there is no automated health probe pipeline, no SLO publication, and no public uptime history beyond the two manually recorded incidents.

Evidence

/status page with live market data snapshot (force-dynamic)
Incident log with two resolved entries
Data freshness timestamp surfaced to users

Gaps

No automated health monitoring (no external uptime monitor e.g. Better Uptime, UptimeRobot)
No SLO/SLA publication (target uptime % not declared)
No error budget or burn rate tracking
Incident history is sparse — cannot verify operational continuity

Action: Wire an external uptime monitor to /api/health and /api/market-pulse. Publish monthly uptime data on /status. Define and publish SLOs: 99.5% uptime, <500ms p95 API response.

Deployment Assurance / SRE Readiness

AMBER5 / 10

Vercel's atomic deployment model provides compute rollback in <60 seconds. Neon provides 30-day PITR for database rollback. However, no on-call model, alerting runbook, or observable health probe dashboard exists in the accessible codebase.

Evidence

Vercel atomic swap guarantees zero-downtime deploys
Neon 30-day point-in-time recovery for data rollback
Post-deploy smoke script validates critical paths
Prisma migration system provides schema version control

Gaps

No on-call rotation or escalation policy documented
No alerting for data pipeline failures (silent failure risk)
No p95 latency or error rate dashboards (no Datadog / Grafana observed)
RTO/RPO targets not formally published

Action: Set up Vercel monitoring + Sentry for error tracking. Add a pipeline health webhook that alerts on stale data (>24h without a successful refresh). Document RTO/RPO formally.

Enterprise Readiness

AMBER7 / 10

Documentation is strong. Partner API surface and embed SDK are described in docs. Clerk provides managed auth with enterprise SSO support. However, SOC 2 compliance is claimed but not yet independently verifiable from the public surface.

Evidence

Partner API docs with attribution engine and rate-limit tiers
Embed SDK described for broker and media surfaces
Clerk managed auth supports SAML SSO for enterprise plans
Role-based entitlement system (free/pro/enterprise) implemented
Comprehensive FAQ documentation published

Gaps

SOC 2 report not publicly linked or verifiable
No DPA (Data Processing Agreement) template published
No API versioning strategy published (risk of breaking changes for partners)
Enterprise SLA (dedicated support, incident response time) not defined

Action: Publish SOC 2 attestation letter (or link to trust portal). Publish a DPA template. Define API versioning policy (e.g., /api/v1/ prefix with 6-month deprecation window).

Priority Actions for Full Sign-Off

Connect an external uptime monitor to /api/health

Required for any published SLA — without it, uptime claims are unverifiable.

1 week

Publish formal SLOs on /status page

Customers and partners need a declared uptime target to hold the platform accountable.

1 week

Publish SOC 2 attestation or trust portal link

Enterprise and institutional clients require this before contract signature.

2–4 weeks

Add pipeline health alerting (stale data webhook)

Silent pipeline failures are the highest-risk gap in the data integrity model.

2 weeks

Publish DPA template for partners and enterprise

GDPR / UAE PDPL compliance — required for any B2B data sharing agreement.

2 weeks

Define and publish API versioning policy

Prevents breaking changes from disrupting partner integrations at scale.

1 month

Board Summary

Category	Score	Rating
Product Deployment Status	GREEN	8 / 10
Release Discipline	GREEN	7 / 10
Architecture Maturity	GREEN	9 / 10
Operational Transparency	AMBER	5 / 10
Deployment Assurance / SRE Readiness	AMBER	5 / 10
Enterprise Readiness	AMBER	7 / 10
Overall	CONDITIONAL PASS	7.0 / 10

Reviewed: March 2026 · Next review: June 2026 or after P1 actions are completed.