Jon Chery
b054849a99
P4 (Wave 3, docs) — REQ-186, 191, 192, 193, 194, 195, 204, 210, 211, 212, 213 New docs: - docs/METRICS.md — canonical KPI catalog (grounded/derived/deferred) - docs/metrics/*.md — 13 per-KPI definition-of-success docs (D-127) - docs/METRICS_DEFERRED_ROADMAP.md — 8 deferred metrics + hot-path plan + re-eval triggers (REQ-210) - docs/NO_HUMANS_THESIS.md — thesis defensibility brief (REQ-213) New tools: - core/metrics/trust_snapshot.py — 5 trust metrics + chain-integrity verdict + snapshot hash (REQ-211) - scripts/check_north_star_diff.sh — CI check for NORTH_STAR strategic section changes (REQ-204) Modified: - .ciagent/config.json — strategic_direction_file: .ciagent/NORTH_STAR.md (REQ-186) ---ci--- project: acdl phase: 4 milestone: v1.17 status: execute ---/ci---
3.6 KiB
Nova — The No-Humans Infrastructure Platform: Thesis Defensibility Brief
v1.17 — Strategic Direction, Leadership Metrics & Unified Story (REQ-213) Generated: 2026-08-04
The thesis
Nova is the autonomous infrastructure layer that lets product teams ship without engaging an operator, and lets executives trust the AI not because it never fails but because every decision is captured, scored, and accountable.
Autonomy in operations; human at stage gates. The operator is removed from the loop of normal operations. Human attestation remains required at stage gates — QA signs off for production, SRE greenlights based on operational readiness. The absence of an operator is never the absence of a record.
Grounded proof (measurable today)
| Proof | Source | Status |
|---|---|---|
| 18 capabilities verified, 4 honestly skipped (0 broken) | REGRESSION_REPORT.json |
grounded |
| Decision Ledger captures 100% of AI decisions with outcome backfill | metrics/decision_ledger.db |
grounded (this milestone) |
| Attestation Coverage: 100% of prod/dr promotions attested by a human | hitl_gates.py + outbox approver_* |
grounded |
| Confidence-gated policy engine (not an LLM) — 6 weighted inputs, band outcome | confidence_signal.py |
grounded |
| 8-concern attestation matrix with separation-of-duties on prod | attestation_matrix.py + separation_of_duties.py |
grounded |
| Pre-apply cost estimates (Infracost, offline) | infracost_adapter.py |
grounded |
| Test suite passes (~656 tests) | metrics/test-results.xml |
grounded |
Deferred proof (measurable when blocking decisions lift)
| Proof | Blocking Decision | Unblock Requirement |
|---|---|---|
| Touchless Resolution Rate ≥99% across production estates | 0 consumers today | Pilot estate activation |
| Live infrastructure health (ECS, ALB, RPS) | D-096 | Live AWS re-provisioning |
| Onboarding funnel: requested → granted | D-113/D-114/D-119 | Auto-grant implementation |
| Drift auto-reversal rate ≥95% | D-096 + no scheduler | Drift detection scheduler |
| Predictive vs reactive ratio ≥3:1 | future emitter | ML anomaly-forecasting service |
| Tamper-evident ledger checkpoints (S3 Object Lock + JWS) | D-083 | Audit ledger build-out |
Anti-claims (what Nova is NOT)
- Nova's "AI" is NOT an LLM planner. It is a confidence-gated
policy engine (confidence_signal + HITL gate). The Decision Ledger
captures this real decision path — not a fabricated "AI agent" that
doesn't exist yet (D-122). When an LLM planner is added, it will emit
richer
alternatives_consideredwithout schema breakage. - Nova does NOT remove humans from accountability. Only from operations. Every stage-gate promotion (qa/prod/dr) requires a human attestation recorded with approver identity, separation-of-duties check, and the 8-concern evidence matrix (NORTH_STAR Anti-Goal #3).
- Nova is NOT for legacy, untagged, or freeform infrastructure. It requires Terraform-managed, policy-aligned, fully-tagged inputs (NORTH_STAR Anti-Goal #4).
- Nova does NOT fabricate metrics. Every metric is grounded (cites a source file), derived (documented formula), or deferred (cites a blocking decision ID). No fabricated numbers in any deck slide or METRICS.md entry (the "no fabrication" hard constraint).
What "won" looks like
By month 18, Nova is the layer enterprise leadership points to when they say "we don't have an infrastructure ops team anymore, and the audit trail is stronger than it ever was" — and it is the default substrate their AI engineering teams reach for first when an agent needs to deploy.