Operational Glossary: Volume 3
Operational vocabulary for governance, engineering leadership, and multi-agent team practices. External industry frameworks (PMBOK, DORA, RACI, ATAM) are explicitly noted; concepts original to this series (Agentic MVP, Agent Master, Context Watchdog) are defined with actionable operational boundaries.
| Term | Operational Definition | Chapter |
|---|---|---|
| Production Alignment | Maintaining system behavior strictly within declared ethical and statutory boundaries once deployed in live production, rather than solely during pre-deployment evaluation. | 1 |
| Right to Contestability | Legal and architectural guarantee that any individual impacted by an automated decision can demand human review and obtain an interpretable explanation. | 1 |
| Orchestrator Engineer | Senior engineering role focused on specifying, formally verifying, and coordinating autonomous agents rather than executing low-level tasks directly. Evolution of the AI Engineer from Vol. 1. | 2 |
| Agentic Maturity Model (AMM) | Capability matrix classifying an individual or team’s practical competency across agentic systems; the highest tier demands engineering proprietary probabilistic evaluation suites. | 8 |
| Swarm Manager | Leadership role managing hybrid human-agent teams; tasked with establishing strategic intent and acceptance criteria rather than dispatching tasks. | 9 |
| Feedback Latency | Elapsed time between an execution action and the arrival of telemetry that permits correction. Dictates the upper bound of organizational learning velocity. | Vol. 2, Ch. 26; Ch. 7 |
| Just-in-Time Pattern | Anticipating and provisioning reactive capacity prior to business demand rather than scrambling after requirements emerge. | 7 |
| DORA Metrics | Four canonical software delivery indicators (deployment frequency, lead time for changes, change failure rate, time to restore service). Adapted in this volume to hybrid agentic delivery teams. | 6, 7 |
| Swarm Quarterly Closing | Quarterly governance milestone auditing delivered business value synthesized by the agent swarm rather than tracking superficial task activity. | 6 |
| Agentic MVP | Minimal viable increment engineered to validate a specific business hypothesis under bounded blast radius, rather than a shrunken mockup of a full product. | 22 |
| RACE Framework | Architectural methodology for constructing agentic MVPs across graduated risk envelopes. | 22 |
| Fatigued Programmer Symptom (applied to swarms) | Degradation of analytical judgment induced by context window saturation: the agent executes instructions mechanically while losing critical evaluation capacity. | 13 |
| Context Watchdog | Automated infrastructure monitor that tracks context saturation, triggering deterministic compaction, memory offloading, or controlled session restarts. | 13 |
| External Refutation | Countermeasure introducing an independent evaluator agent to prevent an executing agent from circularly validating its own work. | 5, 13 |
| Agentic Kata | Bounded engineering exercise with explicit operational constraints and automated success criteria, designed for deliberate practitioner training. | 14 |
| Agentic RACI | Matrix delineating Responsible, Accountable, Consulted, and Informed across hybrid teams; Accountability remains non-delegable and strictly human. | 15 |
| DevSecOps for Agents | Shifting security, permission boundaries, and automated audits into the agent’s continuous deployment lifecycle rather than conducting post-hoc reviews. | 20 |
| Quantifiable Autonomy Index | Composite metric evaluating the operational freedom granted to an autonomous agent based on its environment, tool access, and blast radius. | 20 |
| Taxonomy of Agentic Liability | Legal and operational framework classifying culpability for software outcomes based on the degree of human intervention during generation. | 21 |
| Synthetic Code | Code generated by a foundation model; necessitates re-architecting ownership, verification harnesses, and ultimate statutory liability. | 21 |
| Architecture Review Board Defense | Formal governance gate evaluating architectural viability, unit economics, and risk; Chapter 4 provides battle-tested strategies for defending multi-agent designs. | 4 |
| ATAM (Architecture Tradeoff Analysis Method) | Method for stress-testing architectural quality attributes and trade-offs prior to presenting before executive risk committees. | 4 |
| ADR (Architecture Decision Record) | Structured, immutable log capturing architectural context, evaluated alternatives, decisions made, and operational consequences. | 18 |
| Defensible Moat | Sustainable competitive advantage independent of foundation model capabilities; the central evaluation criterion of the monetization chapter. | 19 |