Glossary

How to use this glossary

Terms are listed alphabetically. Terms defined by a volume of the series are used here as that volume defines them; terms of this dossier (such as established, illustrative, ledger and open thread) are defined by Chapter 1 and by DISPOSITION.md.

Terms

The terms below are those of this dossier and the series terms it uses. Definitions are short and follow the volumes; where a term is defined by an instrument, the instrument governs.

Across the usage modes. A subsection in the chapters that shows, usually as a table, how a subject differs between Mode 0 and Mode 5.

ADR (Architecture Decision Record). The series' record of an architecture decision, kept beside the AI Decision Record for decisions that concern AI.

AI Decision Record (AIDR). The series' basic unit of decision: it states what was decided, by whom, between which options and with what review date. Each volume from AI Red Team onwards allots a number range to each organisation (Chapter 22).

Algorithmic management. Directing, evaluating, rewarding and disciplining people by means of AI; the subject of AI Boss.

Answerability principle. The principle of AI Leader that AI may inform, draft, prepare and, within a mandate, execute, but may not answer; a named leader passes the tests Name, Explain and Reach.

Appendix A (ledger of identifiers). The generated list of every identifier the volumes attribute to Nordvik Region, with the volume and chapter that state it.

Autonomy level. The series' degree of delegation per function and action class, from Level 0 Advisory to Level 4 Autonomous. The volumes set ceilings for Level 4 in each domain.

Baseline. In Landscape, the line under a mark that shows evidence: solid when automatic, dashed when attested.

Benefit record (BEN). A record in AI Environment of the benefit that a use claims, per functional unit, measured against a comparison with a stated hold-out, with an evidence age of 180 days; one per footprint record, owned outside the sponsor line (BEN-NR-0001).

Benefits triage. The Region's purpose-built Mode 4 system that ranks applications for needs-based financial assistance; the caseworker decides every case. Run by a bought product, TriageSoft, in the profile.

Blue team. The people and functions that defend an organisation; the subject of AI Blue. Kept distinct from red team and purple team.

Break-glass account. An emergency account with elevated access, used under a recorded procedure; AI Developer sets a test of such accounts as an open item.

Builder. In Vibe to Dev, a person who is not a developer and builds software with an AI assistant.

Cap. The third test of the method of AI Environment: an agent or a use that can start work has a budget set by someone else, with a stop path outside the system it stops (Count, Compare, Cap).

Claim record (CLM). A record in AI Environment of a public environmental statement with its boundary, period, baseline, evidence, expiry, independent check, reviewer, approver and withdrawal rule; CLM-NR-0001 is the Region's figure of 1.90 g per resolved enquiry, withdrawn on 2032-12-13.

Companion. A book of the series' site that has no number and refers to the series: Landscape, Tools and the two dossiers.

Compare. The second test of the method of AI Environment: the benefit of a use is set against its measured footprint, and against the footprint of a smaller or simpler alternative that meets the requirement.

Contestation record (CNT). The record of a person's contest of a management decision in AI Boss; CNT-NR-0001 is the first at Nordvik Region.

Contour. The visual notation of Landscape: one mark per AI use with seven dimensions, drawn from a machine-readable model.

Count. The first test of the method of AI Environment: a use is counted, with its energy, carbon and water, a boundary and a date, before any claim or decision rests on it (the Count rule of AIDR-1300).

Credibility check. Part V of this dossier: a comparison of the fiction with real organisations of the same kind, from public sources, with a verdict per dimension.

Decision seat. In Landscape, the tier (T1 to T4) of the person or body that holds a decision about a unit.

Delegation profile (DLG). A record that sets the autonomy level per function or action class for a domain, for example DLG-NR-ROS-01 for rostering and DLG-NR-DEV-01 for development.

Deployer. The AI Act's term for an organisation that uses an AI system under its own authority; the Region is deployer of most of its estate.

Energy and carbon budget (EBG). A monthly budget in kWh for one use, set by someone other than the use's owner, with a ladder of alerts (0.80, 1.00 and 1.10 of the quota) and a drilled stop; without a drill it is proposed and not active (EBG-NR-0001, 4,450 kWh a month).

Environmental functions (E1 to E5). The five functions of AI Environment: E1 Measure, E2 Justify, E3 Design, E4 Operate and E5 Account.

Environmental register. The Region's register of its AI uses with owner, footprint record, benefit record, budget and evidence age (register/environment-nr.yaml, adopted by AIDR-1300).

Established. A fact that a volume states as a fact of the profile, as against an illustrative one.

Evidence age. The time since a result was obtained, set against the clock for its kind of claim. A claim is verified only while its result is within the clock (Chapter 23).

Evidence clock. The period within which a result supports a claim. The clocks of six volumes coexist; the Region has no single evidence policy.

Exit test. A test that the Region can leave a supplier, for example the exit test of the code model on 2031-02-03; the series adopts the measure that the Region can show it can leave.

Extract. A working note in data/extracts/, one per volume, from which the chapters are written; section 8 lists the inconsistencies.

Footprint record (FPR). A record in AI Environment of the energy, carbon and water of one use per functional unit, with a boundary, a method, a range and an age of 90 days (FPR-NR-0001, revisions 1 and 2).

FRIA. Fundamental rights impact assessment, for example FRIA-2026-002 for benefits triage.

Function (blue, management, operations and others). A named set of work in a volume: B1 to B5 in AI Blue, M1 to M5 in AI Boss, O1 to O5 in AI Operations, L1 to L5 in AI Leader, D1 to D5 in AI Developer.

Gate. In Vibe to Dev, the point at which a step is finished only when an artefact is shown.

General-purpose AI model. The AI Act's term for a model that serves many purposes; used here as the AI Act uses it.

Hansa Mutual. The sibling fictional organisation of the series, a mutual insurance group headquartered in Hamburg of about 9,500 employees, covered by its own dossier.

High-risk AI system. The AI Act's term for a system in a listed use; the Region's benefits triage is treated as one.

Illustrative. A figure or event that a volume marks as an illustration. Most events of 2027 to 2032 are illustrative fiction written in November 2026.

Landscape model. The machine-readable model in Landscape of one organisation's AI uses on a date; Nordvik Region's has thirteen entries, two of them planned on 2027-01-01, so that the picture of that day has eleven marks.

Ledger. The machine-readable list in data/ledger.jsonl of the identifiers, events and figures of the dossier, each with the volume and chapter that state it.

Mandate manifest (MND). A record of what an agent or assistant may do, with its permissions and review date, for example MND-NR-01 for the security architect's tools.

Mode (usage mode). The series' scale of how AI is used, from Mode 0 Ambient AI through Mode 1, 2, 3 and 4 to Mode 5 Autonomous production agents. The Region permits no Mode 5.

MS-1, MS-2. The annexes to the managed security provider agreement NR-2025-044; MS-2 is to be signed by 2028-03-31 at the latest and its status is open.

Net benefit principle. The principle of AI Environment that a use is justified only when its benefit per unit of environmental cost is shown with evidence, by the tests Count, Compare and Cap.

NIS 2. Directive (EU) 2022/2555 on measures for cybersecurity; its position at the Region is an open thread (OT-NR-21).

Nordvik Region. The fictional regional public authority of this dossier: about 6,000 employees, about 450,000 residents, in a Nordic member state.

Observable, expected, speculation. The three labels of AI Governance, Chapter 22, that every statement about the future carries.

Observer principle. The principle of AI Operations that what verifies an action must be independent of what performed it.

Open thread. A point on which the volumes disagree or that a volume leaves without outcome; collected in Chapter 27 with a recommended resolution.

Placement policy (PLC). A record in AI Environment of where and when work may run to reduce its footprint, with thresholds (10 per cent and 1 kg) and the classes it covers (PLC-NR-0001).

Playbook manifest (PBK). A record of an executable response playbook, for example PBK-NR-ISOLATE-01.

Post-incident review (PIR). A record of a review after an incident, for example PIR-NR-2030-0006.

Provider. The AI Act's term for an organisation that develops an AI system or has it developed and places it on the market or puts it into service under its own name; the Region is provider of its explanation service.

Purple team. The joint loop between red and blue, not a third team; the subject of AI Purple.

Quota. In AI Environment, the monthly volume at which an environmental budget is reached, recalculated when a re-measurement drifts by more than 0.05 (EBG-NR-0001, 160,650 resolved enquiries, then about 176,600).

Real counterpart. A real organisation of the same kind, named only in Part V as a comparator.

Red team. The people who emulate the adversary; at the Region this is bought, not internal.

Restore test (RST). A record of a test that a service can be restored, with time and result, for example RST-NR-2027-002.

RoE (rules of engagement). The signed scope and limits of an emulation or test, for example RoE-NR-2027-001.

Role charter (ROLE). A record that fixes a role's duties and limits, for example ROLE-NR-SA-01 for the security architect.

Series order. The rule that each volume refers only to earlier volumes; this dossier refers to Books 1 to 14.

Stewardship principle. The principle of AI Developer that AI may draft, generate, refactor and, within a mandate, execute, but may not own; the tests are Name, Understand and Revert.

Stop path. The means of halting a system from outside it, with an owner, a time and a test; recorded for each class of system in the volumes.

Stop right. The right of a named role to stop a system without asking, for example the Head of Benefits Unit for the ranking of claims.

Supplier disclosure record (SDR). A record in AI Environment of what a supplier discloses about energy, grid, embodied carbon and water, with what is not given, valid for 12 months (SDR-NR-0001 to -0003).

Takeover test. In Vibe to Dev, the test that a named colleague can in a working day run, change and ship a build and switch it off and restore it.

Tier (A to D). In AI Developer and Vibe to Dev, the four classes of build: personal, team, business and project; Tier D is not a citizen build.

Tools. The companion that catalogues every tool of Books 1 to 13; cited here as source 16.

Unit. In Landscape, one AI use, drawn as one mark.

Working value. The value this dossier uses while an open thread is unresolved, usually the later volume's ledger or the profile.