Chooser / Four routes
This index is for situations that are already visible inside a product and engineering system, but still need the right ownership lens.
Choose the operational problem, not a generic service.
Four peer routes into one operational system.
Start with the first boundary that is clearly failing. If another domain sits next to it, follow the seam after the first owner is named.
Four routes
Delivery Recovery
Roadmaps keep moving while releases lose ownership.
The team spends political energy on status instead of moving work to production.
Release-path resets, transition ownership, and lead-time evidence.
Read the Delivery Recovery problem pageAI Engineering Control
AI tools multiply while output stays hard to trust.
Speed theatre replaces a workflow that can be reviewed, measured, and reused.
Specification, evaluation, review, provenance, and workflow control.
Read the AI Engineering Control problem pageAI Product Risk & Safety
A probabilistic feature needs a production boundary before users absorb the risk.
The product can ship uncertainty without a tier, escalation path, or stopping rule.
Risk tiers, release gates, human escalation, and product-safety cues.
Read the AI Product Risk & Safety problem pageResilience & Security
One weak operating layer can turn load, attack, or ambiguity into an outage.
The system depends on hope where failure ownership and layered control should exist.
Attack-path mapping, control layers, migration trade-offs, and monitoring owners.
Read the Resilience & Security problem pageCross-domain seam
Some situations cross a seam. They still belong to one of the four routes.
Operational CTO judgment shows up where product, engineering, and incident ownership touch. That is a lens across the system, not a fifth category.
Representative evidence
Representative evidence across all four domains.
Each cue keeps context, boundary, and role together while Phase 4 claim review is still conservative.
A release path became inspectable only after ownership and cadence were reset together.
- Context
- Delivery Recovery
- Timeframe
- First-hand operating record
- Role
- Operational CTO
- Provenance
- First-party source material with claim and disclosure boundaries retained.
- Confidence
- Conservative wording; a stronger claim requires a separately approved source.
- Disclosure
- Client identity and unsupported metrics are excluded.
A machine-readable system model created a safer base for change than oral history.
- Context
- AI Engineering Control
- Timeframe
- First-hand operating record
- Role
- Operational CTO
- Provenance
- First-party source material with claim and disclosure boundaries retained.
- Confidence
- Conservative wording; a stronger claim requires a separately approved source.
- Disclosure
- Client identity and unsupported metrics are excluded.
A product-risk issue was reproduced and closed before first users carried the failure.
- Context
- AI Product Risk & Safety
- Timeframe
- First-hand operating record
- Role
- Operational CTO
- Provenance
- First-party source material with claim and disclosure boundaries retained.
- Confidence
- Conservative wording; a stronger claim requires a separately approved source.
- Disclosure
- Client identity and unsupported metrics are excluded.
A resilience response held only when layered controls and a named operator boundary were explicit.
- Context
- Resilience & Security
- Timeframe
- First-hand operating record
- Role
- Operational CTO
- Provenance
- First-party source material with claim and disclosure boundaries retained.
- Confidence
- Conservative wording; a stronger claim requires a separately approved source.
- Disclosure
- Client identity and unsupported metrics are excluded.