The EXA Lexicon

View raw

{
  "title": "The EXA Lexicon",
  "slug": "lexicon",
  "objectNumber": 1,
  "version": "1.0",
  "status": "published",
  "source": "Series One · Series Two",
  "sourceArticleUrl": "https://medium.com/enterprise-experience-architecture/spec-driven-development-certifies-the-code-who-certifies-the-consequence-a3b0503af23a",
  "summary": "Canonical, versioned glossary of every coined term in the Enterprise Experience Architecture framework. Covers all terms from Series One (The 2026 Field Guide) and Series Two (The Specification Economy), both complete and published, plus two role terms — Enterprise Experience Architect and Agentic Experience Architect — already in public use on reidexa.design/exa ahead of their full treatment in the forthcoming Series Three and Series Four.",
  "lastUpdated": "2026-09-10",
  "changelog": [
    {
      "date": "2026-06-30",
      "note": "Founding term set published: 26 terms drawn from Series One and Series Two in full, plus the two forthcoming role terms already live on the site's framework page. Replaces the prior placeholder starter set."
    },
    {
      "date": "2026-07-01",
      "note": "Added a category field to every term (governance/roles/economics/mechanics) to power the findability grouping added this session. Additive only — no definition text changed."
    },
    {
      "date": "2026-07-16",
      "note": "Worked example populated (Build Ledger §9 item 4 — the last founding-release content gap): an object-level diagnosis walking one organization across the Experience Architecture Maturity Model's four stages, threading Dark State, the Agentic Constitution, the Logic-Review Gate, and the Continuous Learning Loop. Reworded from Series Two, Article 4 ('The Only Moat Left Is the One You Specify'). Additive — no term definitions changed, version unchanged."
    },
    {
      "date": "2026-07-25",
      "note": "Five Series Three terms added following Article 1's publication ('Someone Just Asked Who Governs Your AI. Your Org Chart Has No Answer.'): The Governance Vacancy and The Authority Audit are fully defined by Article 1; The Provisional Surface, Specification Authority, and Blast Radius are seeded with Article 1's current wording pending fuller mechanical treatment in Articles 3 and 4. Term count 26 → 31. Resolves the standing exclusion previously noted in Build Ledger §3.2."
    },
    {
      "date": "2026-08-07",
      "note": "Two existing terms extended in place following Article 3's publication ('Your Kill Switch Works. Nobody Has the Standing to Pull It.') — no terms added, renamed, or removed; term count unchanged at 31, both slugs unchanged. The Cultural Veto: the Series One/Two risk framing is retained as the origin, with Article 3's deliberate promotion of that risk to an operational mechanism appended — the standing, executive-backed authority to halt a live deployment running against an uncertified constraint set, and the three conditions it holds on. The Continuous Learning Loop: the original component-consumption stream is retained, with Article 3's second telemetry stream appended — override events resolved to a shared versioned constraint ID and aggregated into one log, the substrate the Amendment Protocol's structural signal is matched across. Each term's source field gained a trailing Series Three clause; firstPublished and sourceUrl are unchanged, since both remain factual records of first appearance."
    },
    {
      "date": "2026-08-16",
      "note": "Enterprise Experience Architect's source field corrected following Article 4's publication ('You Can Describe the Job. You Cannot Name It.'): 'full treatment in Series Three (forthcoming)' replaced with 'full treatment in Series Three, Article 4', since Article 4 is confirmed as the article that delivered the deep-dive treatment, evaluation rubric, and career ladder this term's own definition text already promised. Source-field correction only — definition, seeAlso, firstPublished, and category unchanged; term count and slug unchanged. The Reviewer Interface term's own still-stale source field is a separate, unresolved judgment call (which article fulfills its 'organizational treatment' clause) and was deliberately left untouched pending the author's decision."
    },
    {
      "date": "2026-08-16",
      "note": "Reviewer Interface's source field corrected: 'organizational treatment in Series Three (forthcoming)' replaced with 'organizational treatment in Series Three', dropping the stale forthcoming status now that all four Series Three articles are published. Unlike Enterprise Experience Architect's fix above, this citation stays series-level rather than pinning to a single article — this term's definition makes no single-article promise the way Enterprise Experience Architect's did. The leading 'Series One · Article 2' clause (where the term was originally coined) is unchanged. Source-field correction only — definition, seeAlso, firstPublished, sourceUrl, and category unchanged; term count and slug unchanged. Closes Build Ledger item 36 in full."
    },
    {
      "date": "2026-08-16",
      "note": "workedExample gained one added sentence, cross-linking forward to the new Experience Architecture Maturity Model object (Standard object 11, draft) — the structured mechanism/symptoms/next-move reference table this walkthrough's own narrative diagnosis only implied, built the same session. Additive only: the existing walkthrough paragraph is unchanged, word for word, aside from the one new closing sentence. No terms added, changed, or removed; version unchanged, same precedent as this term's own 2026-07-01 category-field addition and the two source-field fixes above."
    },
    {
      "date": "2026-08-16",
      "note": "The Authority Audit's definition gained one added sentence, cross-linking forward to the new Authority Audit object (Standard object 3, draft) — the structured five-question table with 'what a failing answer sounds like' phrasing per question, content this term's own prose summary never carried. Additive only: the existing definition is unchanged, word for word, aside from the one new closing sentence. No terms added, changed, or removed; version unchanged, same precedent as the Maturity Model's own workedExample cross-link above."
    },
    {
      "date": "2026-08-17",
      "note": "Two new terms added, both ratified by the framework owner — a genuinely different class of entry from every changelog note above, which were all corrections to existing terms. Specification Debt: defined primarily from Series Two, Article 3's own strongest single-sentence statement ('the accumulated weight of unnamed conditions and untranslated business rules that compounds silently until production pays it down at the most expensive possible rate'), cross-checked against Articles 2 and 4 for consistency — published-article language, not the internal Project Brief's working definition, per this repo's standing source-priority rule. Vendor Specification Debt: internal research only (EXA Agentic Workflow, Steps 5-6, ratified 2026-08-11), no public article exists, so no sourceUrl/firstPublished is set — a specific, named species of Specification Debt, not identical to it, and cross-linked to it via seeAlso in both directions. Term count 31 → 33. Both inserted alphabetically (ignoring leading 'The'): Specification Debt between Specification Authority and The Specification Economy; Vendor Specification Debt after Variables Over Pixels, now the final term in the array. No existing term's content touched. Seven further terms from the same ratification packet are explicitly out of scope for this session."
    },
    {
      "date": "2026-08-17",
      "note": "Seven further terms added, the larger, internally-sourced half of the same 2026-08-11 ratification packet this file's own entry above split off from (that entry's two terms were the half with public-article language available; these seven have none and never have — a Series Three Article 4 drafting note confirms they were deliberately kept unnamed pending this ratification). All seven adapted directly from internal, ratified research (EXA Agentic Workflow, Step 6 §6's Lexicon Additions Proposal, cross-checked against each term's fuller mechanical specification in Step 7), disclosed as such in each term's own source field — no sourceUrl/firstPublished set for any of the seven, since no public article exists to cite. Generative Envelope, Generative Grammar Fuzzing, Generative Vendor Envelope: inserted between The Garbage-In Multiplier and The Governance Vacancy. Invention/Assembly Boundary Test: inserted between Infrastructure of Intent and The Legacy Anchor — its definition cross-links in prose to The Amendment Protocol & Veto Charter (Standard object 5), corrected from this session's own brief, which had misattributed the reverse test it's modeled on to the Agentic Constitution Format; the source research's own C8 citation traces to the Veto Charter object, verified directly against that object's live 'escalation path' section before writing the cross-link. Compositional Drift, Constraint Class, Containment State: inserted between Cognitive Endurance and Continuous Learning Loop. Fallback State and Structural Signal, the two terms Containment State and Compositional Drift respectively adapt, were checked and confirmed to have no existing Lexicon entry of their own — described in prose only, not cross-linked, since a seeAlso slug with no matching term fails the build by design. One existing term touched, additively only: Agentic Experience Architect gained one closing sentence recording the ruled Compositional Drift ownership split (Enterprise Experience Architect primary, Agentic Experience Architect named secondary owner on that one assessment, per D-054) and gained 'compositional-drift' as a fourth seeAlso entry — its existing definition text, other seeAlso entries, source, and category are otherwise unchanged. Term count 33 → 40."
    },
    {
      "date": "2026-09-09",
      "note": "Three new terms added, the first in this file sourced from a standalone article rather than a numbered Series entry — 'Your Agents Have Owners. Your Policy Has No Author.' (published 2026-09-09), which also produced Standard Objects 12 (The Specification Authorship Record) and 13 (In-Flight Triage), both live as of this entry. The State Inventory: this concept already existed informally on the site (a bare mention in the Agentic Constitution Format's own worked-example table, and in this file's own Legacy Infrastructure Collision definition) but had no citable definition of its own until this article gave it one — a five-column enumeration (state in operational language, trigger condition, must/must-not, cost if rendered wrong, discovery source) with five named discovery-source targets. Cross-linked to Agentic Constitution (the object whose constraint blocks each row translates into — note this is a correction of the session prompt's own suggested target, 'agentic-constitution-format', which is not an existing Lexicon slug; the Lexicon's term for that object is 'agentic-constitution') and to the new Specification Authorship Record term. The Specification Authorship Record and In-Flight Triage: defined directly from the article and from the built object files, matching each object's own summary. Cross-linked to Authority Audit and to each other; the Specification Authorship Record's third seeAlso target is Cultural Veto rather than a standalone Amendment Protocol term — another correction of the session prompt's own suggestion ('amendment-protocol'), since no Lexicon term exists for the Amendment Protocol on its own, only Cultural Veto, covering the other half of that same object (Standard object 5) — confirmed by checking the full term list directly rather than assuming a slug existed. All three inserted alphabetically (ignoring leading 'The'): In-Flight Triage between Governed Autonomy and Infrastructure of Intent; The Specification Authorship Record between Specification Authority and Specification Debt; The State Inventory between The Specification Scorecard and Variables Over Pixels. This changelog entry is appended at the end of the array rather than inserted at the top — this file's own established convention is chronological, oldest-first (unlike Objects 3/5/12/13's newest-first changelog convention, confirmed by checking this file's actual entry order directly rather than assuming the other objects' convention carried over). No existing term's content touched. Term count 40 → 43."
    },
    {
      "date": "2026-09-10",
      "note": "Build Ledger item 36's full batch scope, sourced directly from the live Article 4 ('You Can Describe the Job. You Cannot Name It.', fetched and read in full, not reconstructed from any prior draft or summary). Two of the task's four parts were found already done by an earlier session and are reported, not repeated: Enterprise Experience Architect's and Reviewer Interface's source fields were both corrected on 2026-08-16 (see those entries above) and their 'source' text was checked directly against this file — both already read correctly, no forthcoming language remains on either. Two genuinely new terms added: The Enterprise Experience Architect Evaluation Rubric (slug eea-rubric) and The Experience Architect Career Ladder (slug career-ladder), both sourced from Article 4 and cross-linked to their live Standard objects (6 and 7 respectively), inserted alphabetically (ignoring leading 'The') — the Rubric between Enterprise Experience Architect and Enterprise Experience Architecture (EXA); the Career Ladder between Enterprise Experience Architecture (EXA) and The Experience Architecture Maturity Model. One existing term substantively rewritten: Enterprise Experience Architect's definition, previously a placeholder stub reading 'the role that ratifies and enforces the Agentic Constitution' plus a forward-looking 'ships with Series Three' promise, is replaced with the article's actual four-responsibility operational definition (ratification authority at the Logic-Review Gate; the Cultural Veto; ownership of the State Inventory and the Amendment Protocol; the standing seat with compliance, operations, and legal — confirmed against the article's exact language, not the ledger's own shorthand verbs alone) plus all four 'not this role' collisions, reworded rather than reproduced verbatim per this repo's copyright discipline: the three already tracked elsewhere on the site (AI Agent Owner, AX Designer, Enterprise Architect) had no existing Lexicon disambiguation entry at all — checked directly, not assumed present — so this is the first time they appear in the Lexicon itself, alongside the fourth, Chief AI Officer, that Article 4 introduced (the CAIO as this function's sponsor, not its rival, per the article's own distinction between owning AI strategy and authoring the constraint set a specific agent executes against). Gained two new seeAlso entries (eea-rubric, career-ladder); existing seeAlso entries, source, category, and slug unchanged. Every new seeAlso target confirmed to resolve by direct check of the full slug list before writing, not assumed. Term count 43 → 45 — two new terms, not the three the originating task brief anticipated, because the brief's own 'three new terms' count included the Enterprise Experience Architect entry as if it did not yet exist; it already did (added at founding-term-set launch, 2026-06-30), so this pass edited it in place rather than adding a fourth term. Reported as a correction to that assumption, not silently reconciled to match it. No other term touched; writing.astro, exa.astro, and the articles collection untouched, per this task's explicit scope."
    }
  ],
  "type": "reference",
  "tagline": "The canonical, versioned glossary of every coined term in the Enterprise Experience Architecture framework — the shared language the Standard is built on.",
  "workedExample": "Read as a diagnosis rather than a summary, the Maturity Model's four stages are the Lexicon locating one organization on the ladder and naming each rung it climbed. Take an energy-trading firm that forked one terminal-status surface four ways across its regional desks — Ungoverned — where a single fork rendered a closed terminal as accepting cargo with a vessel already underway and the demurrage clock running: a Dark State, an operating condition no specification had ever named. Two peers in the same sector, running the same agent fleet, then split: the one that first ran a state inventory shipped clean and crossed into Constrained with an Agentic Constitution to build against, while the one that pointed its agents straight at the design system rendered an embargoed counterparty as clear to load — a sanctions state nobody had enumerated, caught only after the fact, at six figures in remediation, a regulator self-report, and a quarter under board scrutiny. Constrained is not the finish it feels like, since a constitution nobody defends drifts; the move to Instrumented is the Logic-Review Gate standing at the definition of ready, catching a deviation as a constraint check before the build instead of an incident after it. Compounding is the Continuous Learning Loop closing on its own — every regulatory change amended into the constitution before it can become the next Dark State — at which point the terms above have stopped being a glossary and become the instrument the diagnosis was run with. This walkthrough now has a structured companion: the same four stages as a self-diagnostic reference table — mechanism, observable symptoms, and next move per stage — at The Experience Architecture Maturity Model (Standard object 11).",
  "terms": [
    {
      "term": "The Adoption Contract",
      "slug": "adoption-contract",
      "source": "Series One",
      "definition": "The implicit agreement between a central architecture function and the product squads it serves: compliance with the specification follows from the specification's utility, not from a top-down mandate. A squad adopts the constitution because building against it is faster than building around it, not because it was told to.",
      "category": "governance",
      "seeAlso": [
        "design-as-a-service",
        "cultural-veto",
        "enterprise-experience-architecture"
      ],
      "firstPublished": "2026-05-27",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/ai-doesnt-need-a-better-design-system-it-needs-a-constitution-853f89c09038"
    },
    {
      "term": "The Agentic Constitution",
      "slug": "agentic-constitution",
      "source": "Series One · Article 4 — defended in EXA vs. SDD",
      "definition": "The machine-readable document that encodes an enterprise's Infrastructure of Intent for AI agent consumption. It is a constraint file — .md, .cursorrules, or equivalent — that an agent executes against directly, replacing the annotated screen a developer once interpreted by hand. The architect who writes it governs everything an agent subsequently builds. See The Agentic Constitution Format (Standard object 2) for the structured constraint-block syntax.",
      "category": "mechanics",
      "seeAlso": [
        "infrastructure-of-intent",
        "governed-autonomy",
        "semantic-handoff"
      ],
      "firstPublished": "2026-06-05",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/the-new-deliverable-why-your-design-architect-writes-markdown-2601b645ecf7"
    },
    {
      "term": "Agentic Experience Architect",
      "slug": "agentic-experience-architect",
      "source": "Series Four (forthcoming)",
      "definition": "The product-level counterpart to the Enterprise Experience Architect: the function that governs what agents generate in real time, on surfaces that are never assembled the same way twice and therefore have no static specification to review against. Track A of EXA's two practitioner tracks. The full role definition and evaluation rubric ship with Series Four; the term is already in public use. Named secondary owner, consulted rather than co-ratifying, on the Compositional Drift assessment within Generative Envelope certification — the Enterprise Experience Architect remains primary owner of certification throughout.",
      "category": "roles",
      "seeAlso": [
        "enterprise-experience-architect",
        "reviewer-interface",
        "governed-autonomy",
        "compositional-drift"
      ]
    },
    {
      "term": "The Authority Audit",
      "slug": "authority-audit",
      "source": "Series Three · Article 1",
      "definition": "A self-administered, five-question diagnostic that maps where binding AI-behavior decisions actually get made in an organization, and who holds the standing to make them: who ratifies a constraint set before it ships, who can halt a violating deployment, who is in the room when compliance and legal set policy, who owns the constitution when its author leaves, and whose performance review is affected when an ungoverned surface causes an incident. Designed to be run standalone, without reading the source article. Each question now has a structured companion: the same five questions paired with concrete 'what a failing answer sounds like' phrasing, at The Authority Audit (Standard object 3).",
      "category": "governance",
      "seeAlso": [
        "governance-vacancy",
        "enterprise-experience-architect",
        "agentic-constitution"
      ],
      "firstPublished": "2026-07-25",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/someone-just-asked-who-governs-your-ai-your-org-chart-has-no-answer-f332044f64d9"
    },
    {
      "term": "Blast Radius",
      "slug": "blast-radius",
      "source": "Series Three · Article 1 — career-ladder treatment due Article 4",
      "definition": "The breadth of surfaces governed by a practitioner's specification authority — the seniority axis paired with Specification Authority. Also used to describe risk-tiering: ratification requirements scale to blast radius, so low-risk assistive copilots keep moving through standard agile process rather than requiring the same ceremony as high-blast-radius autonomous agents.",
      "category": "mechanics",
      "seeAlso": [
        "specification-authority",
        "governance-vacancy"
      ],
      "firstPublished": "2026-07-25",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/someone-just-asked-who-governs-your-ai-your-org-chart-has-no-answer-f332044f64d9"
    },
    {
      "term": "Cognitive Endurance",
      "slug": "cognitive-endurance",
      "source": "Series One · Article 3",
      "definition": "The design metric that matters for enterprise software: an operator's ability to sustain performance, accuracy, and clarity of judgment across an 8–10 hour shift inside a single application. Cognitive Endurance treats friction as a direct threat to operating margin rather than a polish problem — the relevant failure mode is not 'unintuitive,' it is 'degrades judgment by hour six.'",
      "category": "mechanics",
      "visual": {
        "kind": "stat",
        "value": 6,
        "suffix": "hrs",
        "label": "Until judgment degrades"
      },
      "seeAlso": [
        "experience-utility",
        "margin-multiplier"
      ],
      "firstPublished": "2026-05-21",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/your-design-system-is-not-decaying-it-is-obsolete-54e0dfe52438"
    },
    {
      "term": "Compositional Drift",
      "slug": "compositional-drift",
      "source": "Internal research — EXA Agentic Workflow, Step 7 §5.3 (full specification) and §6.1/D-054 (ownership ruling), cross-referenced in Step 6 §6 (Lexicon Additions Proposal), ratified 2026-08-11.",
      "definition": "The generative adaptation of the Structural Signal, matched on structural clustering rather than exact constraint-ID identity: the same envelope, across three or more independent sessions, producing compositions that cluster around the same runtime failure. Where it fires, an envelope amendment draft is auto-filed for human disposition the same day, mirroring the Structural Signal's own latency target. A mechanism, not a role: the Enterprise Experience Architect owns Generative Envelope certification throughout, with the Agentic Experience Architect named as a consulted secondary owner on this specific assessment.",
      "category": "mechanics",
      "seeAlso": [
        "continuous-learning-loop",
        "agentic-experience-architect",
        "containment-state"
      ]
    },
    {
      "term": "Constraint Class",
      "slug": "constraint-class",
      "source": "Internal research — EXA Agentic Workflow, Step 6 §1.4 (Q-034 ruling) and §6 (Lexicon Additions Proposal), ratified 2026-08-11.",
      "definition": "The authoring-plane type of a governed object — state constraint, generative envelope, vendor envelope, or generative vendor envelope — carried as metadata alongside, and secondary to, its universal enforcement-plane identity. The two planes describe the same object on purpose: a constraint and a generative envelope are authored differently, but at the point telemetry and enforcement actually read them, both are simply a boundary an agent must not cross. Ratified specifically to keep that one identity from splintering into parallel ID namespaces as generative governance was added to the framework.",
      "category": "mechanics",
      "seeAlso": [
        "generative-envelope",
        "generative-vendor-envelope"
      ]
    },
    {
      "term": "Containment State",
      "slug": "containment-state",
      "source": "Internal research — EXA Agentic Workflow, Step 7 §4 (full specification), cross-referenced in Step 6 §6 (Lexicon Additions Proposal), ratified 2026-08-11.",
      "definition": "The generative adaptation of the Fallback State model, for surfaces with no single 'last certified' state to revert to. An assembled surface's fallback can revert to its previous ratified version; a generative surface has none, since no individual render was ever certified — only the envelope was. The set is closed: discard the failed composition and reprompt against the same envelope up to a stated retry ceiling, degrade to a pre-certified static menu, halt and route to a human, or terminate the session — with no auto-resolved default permitted at any severity.",
      "category": "mechanics",
      "seeAlso": [
        "generative-envelope",
        "compositional-drift"
      ]
    },
    {
      "term": "Continuous Learning Loop",
      "slug": "continuous-learning-loop",
      "source": "Series One — extended to override telemetry in Series Three, Article 3",
      "definition": "The mechanism by which real telemetry on component and pattern consumption flows back to the architect, making the specification more precise with every deployment. Engineering delivers consumption signals as a standing handoff; the architect does not need direct dashboard access for the loop to function. Series Three applies the same mechanism to a second telemetry stream: override events, where a squad overrides a governed constraint, resolved to a shared versioned constraint ID and aggregated into a single log. The Amendment Protocol's structural signal — three squads overriding the same constraint independently — is matched across that log rather than across the org chart; without centralized override telemetry it is undetectable by construction.",
      "category": "mechanics",
      "seeAlso": [
        "infrastructure-of-intent",
        "specification-scorecard"
      ],
      "firstPublished": "2026-05-27",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/ai-doesnt-need-a-better-design-system-it-needs-a-constitution-853f89c09038"
    },
    {
      "term": "The Cultural Veto",
      "slug": "cultural-veto",
      "source": "Series One · Series Two — promoted to a mechanism in Series Three, Article 3",
      "definition": "Originally named in Series One and Series Two as an organizational risk — senior practitioners resisting the shift from a production role to a validation role while the metrics that reward them still count screens produced, not specifications governed; a leadership and incentive-design problem, not a training problem. Series Three promotes that risk to a mechanism: the standing, executive-backed authority to halt a live deployment running against an uncertified constraint set. It holds on three conditions — executive air cover above the squad's delivery leadership, so a halt lasts until the constraint set is certified, not until the sponsor is available; cross-silo diplomacy, in which squads author their own constraints against a schema the authority supplies and it certifies, returns, or amends rather than co-authors; and a fallback state ratified at certification time, so a halt degrades rather than detonates. Distinct from the emergency technical stop: this governs what happens after the bleeding stops.",
      "category": "governance",
      "seeAlso": [
        "reviewer-interface",
        "specification-scorecard",
        "enterprise-experience-architect"
      ],
      "firstPublished": "2026-05-27",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/ai-doesnt-need-a-better-design-system-it-needs-a-constitution-853f89c09038"
    },
    {
      "term": "Dark State",
      "slug": "dark-state",
      "source": "Series Two",
      "definition": "An operational condition that exists somewhere in the enterprise system but has never been named, mapped, or encoded in any specification. Dark states are not ambiguous requirements; they are requirements nobody discovered yet — typically surfaced by the people who already handle them informally (support escalations, manual overrides, sales-floor workarounds), at a fraction of the cost of discovering them in production.",
      "category": "mechanics",
      "seeAlso": [
        "legacy-anchor",
        "continuous-learning-loop"
      ],
      "firstPublished": "2026-06-16",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/the-unit-economics-of-intent-0c136df3cd46"
    },
    {
      "term": "Design as a Service (DaaS)",
      "slug": "design-as-a-service",
      "source": "Series One",
      "definition": "The operating model in which a central EXA function acts as a hub of truth that global product teams subscribe to, rather than a gatekeeping committee teams route requests through. Compliance is earned through utility, not imposed through mandate — see The Adoption Contract.",
      "category": "governance",
      "seeAlso": [
        "adoption-contract",
        "enterprise-experience-architecture",
        "90-day-governance-playbook"
      ],
      "firstPublished": "2026-05-21",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/your-design-system-is-not-decaying-it-is-obsolete-54e0dfe52438"
    },
    {
      "term": "Enterprise Experience Architect",
      "slug": "enterprise-experience-architect",
      "source": "Introduced Series One/Two — full treatment in Series Three, Article 4",
      "definition": "The permanent governing function the enterprise org chart is currently missing. Four load-bearing responsibilities: it holds ratification authority at the Logic-Review Gate, certifying which constraint sets may execute; it wields the Cultural Veto, with executive air cover, to halt a live surface running on an uncertified constraint set and specify the fallback that halt degrades into; it owns the State Inventory and runs the Amendment Protocol, applying the signal-to-noise standard to what enters the amendment queue; and it represents the experience-state layer as a standing peer in the room with compliance, operations, and legal — not a guest who presents and leaves. Track B of EXA's two practitioner tracks. Distinct from four adjacent roles: the Chief AI Officer, this function's sponsor rather than its rival, who owns AI strategy and portfolio risk but not the constraint set a specific agent executes against; the AI Agent Owner, accountable to a specification rather than its author; the AX Designer, who designs the interaction surface a person works with, not what the agent is permitted to do at all; and the Enterprise Architect, who maps an experience to a value stream — a diagram, a measurement — rather than ratifying what that experience may do and what it costs when it renders wrong. Full role definition at The Enterprise Experience Architect Evaluation Rubric (Standard object 6) and The Experience Architect Career Ladder (Standard object 7).",
      "category": "roles",
      "seeAlso": [
        "agentic-experience-architect",
        "reviewer-interface",
        "cultural-veto",
        "eea-rubric",
        "career-ladder"
      ]
    },
    {
      "term": "The Enterprise Experience Architect Evaluation Rubric",
      "slug": "eea-rubric",
      "source": "Series Three · Article 4",
      "definition": "A seven-criterion hiring and leveling instrument for the Enterprise Experience Architect function: six scored qualities and abilities — Pragmatic Visionary, Diplomatic Consensus-Builder, Relentless Pattern Matcher, Ecosystem Cartography, Governance & Scale Strategy, Business Translation — plus one gated, unscored baseline, Technical Fluency, that disqualifies a candidate regardless of the other six scores. Each criterion pairs a real evidence question with a named disqualifier, so it evaluates specific answers rather than a checklist anyone can pass. Full instrument at The Enterprise Experience Architect Evaluation Rubric (Standard object 6).",
      "category": "roles",
      "seeAlso": [
        "enterprise-experience-architect",
        "career-ladder",
        "specification-authority"
      ],
      "firstPublished": "2026-08-14",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/you-can-describe-the-job-you-cannot-name-it-74d88bd07aae"
    },
    {
      "term": "Enterprise Experience Architecture (EXA)",
      "slug": "enterprise-experience-architecture",
      "source": "Series One",
      "definition": "The discipline of designing the governed, intelligent infrastructure that defines how digital products are built and experienced at enterprise scale. EXA succeeds the static design system: where a design system specifies what a product looks like, EXA specifies the semantic variables, business rules, interaction logic, and governance protocols that every downstream system — human or AI — builds against. The framework organizes around three pillars: Scale, Intelligence, and Clarity.",
      "category": "governance",
      "seeAlso": [
        "infrastructure-of-intent",
        "experience-utility",
        "semantic-layer"
      ],
      "firstPublished": "2026-05-21",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/your-design-system-is-not-decaying-it-is-obsolete-54e0dfe52438"
    },
    {
      "term": "The Experience Architect Career Ladder",
      "slug": "career-ladder",
      "source": "Series Three · Article 4",
      "definition": "A four-tier ladder for the Enterprise Experience Architect function — Enterprise Experience Designer, Enterprise Experience Architect, Principal Enterprise Experience Architect, Chief Experience Architect — leveled on Specification Authority and Blast Radius rather than tenure or screens shipped. Bands each tier against the engineering ladder, not the design ladder, and holds one structural requirement at every tier: a ratifier cannot report solely into the function whose delivery pressure the ratification exists to check. Full ladder at The Experience Architect Career Ladder (Standard object 7).",
      "category": "roles",
      "seeAlso": [
        "enterprise-experience-architect",
        "eea-rubric",
        "specification-authority",
        "blast-radius"
      ],
      "firstPublished": "2026-08-14",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/you-can-describe-the-job-you-cannot-name-it-74d88bd07aae"
    },
    {
      "term": "The Experience Architecture Maturity Model",
      "slug": "experience-architecture-maturity-model",
      "source": "Series Two · Article 4",
      "definition": "The four-stage self-assessment an organization runs to locate its current state and name its next move: Ungoverned, Constrained, Instrumented, Compounding. Each stage names both the failure mode an organization is exposed to and the specific lever that advances it to the next stage.",
      "category": "governance",
      "visual": {
        "kind": "tiers",
        "stages": [
          {
            "label": "Ungoverned",
            "body": "A static design system only governs states it knew about at launch — everything after that goes unnamed, forked, or becomes a production incident."
          },
          {
            "label": "Constrained",
            "body": "A constitution now exists and rework drops measurably — but with no function defending it, drift starts the day after it's committed."
          },
          {
            "label": "Instrumented",
            "body": "Deviations are caught before merge and governance is finally measurable — real discipline, but still reactive until the loop closes itself."
          },
          {
            "label": "Compounding",
            "body": "The specification improves with every deployment, without anyone hunting for gaps — the architecture compounds correctness on its own."
          }
        ]
      },
      "seeAlso": [
        "90-day-governance-playbook",
        "specification-scorecard"
      ],
      "firstPublished": "2026-06-23",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/the-only-moat-left-is-the-one-you-specify-d55ff210383e"
    },
    {
      "term": "Experience Utility",
      "slug": "experience-utility",
      "source": "Series One",
      "definition": "The reframing of design from an aesthetic choice into high-availability, governed business infrastructure — held to the operational standard of a cloud architecture or a data pipeline, not the standard of a creative deliverable.",
      "category": "economics",
      "seeAlso": [
        "cognitive-endurance",
        "margin-multiplier",
        "enterprise-experience-architecture"
      ],
      "firstPublished": "2026-05-21",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/your-design-system-is-not-decaying-it-is-obsolete-54e0dfe52438"
    },
    {
      "term": "The Garbage-In Multiplier",
      "slug": "garbage-in-multiplier",
      "source": "Series Two · Article 3",
      "definition": "The observation that AI accelerates good architecture and bad architecture without discrimination. An ungoverned specification does not slow agents down — it produces broken systems faster. The operating consequence is a sequencing rule: govern the specification before scaling agents against it, or go slow to go fast.",
      "category": "economics",
      "visual": {
        "kind": "comparison",
        "left": {
          "label": "Governed specification",
          "body": "AI accelerates good architecture. Agents scale safely against a specification that was governed first."
        },
        "right": {
          "label": "Ungoverned specification",
          "body": "AI accelerates bad architecture too. Agents don't slow down — they produce broken systems faster."
        }
      },
      "seeAlso": [
        "retry-tax",
        "governed-autonomy",
        "margin-multiplier"
      ],
      "firstPublished": "2026-06-18",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/hiring-for-the-artifact-instead-of-the-function-is-the-most-expensive-shortcut-you-will-take-f155abe341fe"
    },
    {
      "term": "Generative Envelope",
      "slug": "generative-envelope",
      "source": "Internal research — EXA Agentic Workflow, Step 7 §1 (full specification), cross-referenced in Step 6 §6 (Lexicon Additions Proposal), ratified 2026-08-11.",
      "definition": "The pre-ratified space, certified at design time, inside which real-time generation is permitted. Where a constraint block certifies one named state's required and prohibited behavior, the envelope certifies a bounded grammar — a closed set of semantic primitives, the rules for composing them, and what remains categorically excluded — so every output a generative surface can produce is, by construction, an assembly from that grammar and nothing outside it. It does not certify outputs; it certifies the space outputs are drawn from.",
      "category": "mechanics",
      "seeAlso": [
        "agentic-constitution",
        "invention-assembly-boundary-test",
        "generative-vendor-envelope"
      ]
    },
    {
      "term": "Generative Grammar Fuzzing",
      "slug": "generative-grammar-fuzzing",
      "source": "Internal research — EXA Agentic Workflow, Step 7 §2.5 (full specification), cross-referenced in Step 6 §6 (Lexicon Additions Proposal), ratified 2026-08-11.",
      "definition": "The validation adaptation for a generative surface's bounded-but-unenumerable output space, standing in for standard state-driver validation where no finite state list exists to walk. Rather than exercising every named state, it drives the envelope through adversarial, systematically generated inputs designed to breach its categorical exclusions or complete a workflow without passing through that workflow's own certified constraint set — passing only when every such attempt is caught before render. A Generative Envelope does not reach certification without it, the same way an assembled surface does not reach its own gate without standard state-driver validation.",
      "category": "mechanics",
      "seeAlso": [
        "generative-envelope",
        "invention-assembly-boundary-test"
      ]
    },
    {
      "term": "Generative Vendor Envelope",
      "slug": "generative-vendor-envelope",
      "source": "Internal research — EXA Agentic Workflow, Step 7 §1.4a (full specification), cross-referenced in Step 6 §6 (Lexicon Additions Proposal), ratified 2026-08-11.",
      "definition": "The hybrid artifact governing a third-party platform whose own agentic layer generates UI in real time, where neither a standard Vendor Envelope (assuming static, enumerable routed states) nor a standard Generative Envelope (assuming the enterprise can certify the grammar directly) applies alone. It accepts that the primitive set and grammar belong to the vendor's own opaque architecture and cannot be enterprise-certified — consistent with The Amendment Protocol & Veto Charter's (Standard object 5) existing rule that an enterprise cannot amend a vendor's constitution — and shifts the governance burden onto data reach, authorized actions, states routed, and halted-state fallback instead. A Vendor Envelope Expiration Protocol resolution against this envelope that accepts an interim limitation as permanent, rather than resolving it, is what the Lexicon separately names Vendor Specification Debt.",
      "category": "governance",
      "seeAlso": [
        "vendor-specification-debt",
        "generative-envelope"
      ]
    },
    {
      "term": "The Governance Vacancy",
      "slug": "governance-vacancy",
      "source": "Series Three · Article 1",
      "definition": "The org-design diagnosis at the center of Series Three: the org chart has no slot for the function that owns the Agentic Constitution — no seat with the standing to ratify it, enforce it, or sit alongside compliance, operations, and legal when binding decisions about system behavior are made. Not a design-credibility gap or a talent shortage; a seat that was never built. Interchangeable in running text with 'the org-design vacancy' and 'the empty seat.'",
      "category": "governance",
      "seeAlso": [
        "authority-audit",
        "enterprise-experience-architect",
        "agentic-constitution"
      ],
      "firstPublished": "2026-07-25",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/someone-just-asked-who-governs-your-ai-your-org-chart-has-no-answer-f332044f64d9"
    },
    {
      "term": "Governed Autonomy",
      "slug": "governed-autonomy",
      "source": "Series One · Article 2",
      "definition": "The operating condition in which AI agents act with full autonomy inside strict, mathematically defined architectural boundaries. Agents are not supervised turn-by-turn by a human reviewer; they are constrained by a specification written in advance. The result is execution speed without the chaos that comes from unconstrained agent behavior.",
      "category": "mechanics",
      "seeAlso": [
        "agentic-constitution",
        "reviewer-interface",
        "infrastructure-of-intent"
      ],
      "firstPublished": "2026-05-21",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/your-design-system-is-not-decaying-it-is-obsolete-54e0dfe52438"
    },
    {
      "term": "In-Flight Triage",
      "slug": "in-flight-triage",
      "source": "Standalone — 'Your Agents Have Owners. Your Policy Has No Author.'",
      "definition": "The named procedure for a state the State Inventory missed, discovered only after a manifest is already in production — the normal case, not the exception. Discovery is logged the moment it happens, never silently absorbed into whatever is already shipping. The discovering squad proposes a severity but does not get to confirm its own proposal: confirmation against constraint-library precedent is fast, a lookup rather than a meeting, while a genuinely new shape with no precedent to check against takes the deliberately slower path. Once confirmed, a CRITICAL or HIGH finding holds that surface until a constraint is at least drafted; a MEDIUM finding ships on schedule and backfills the constraint in parallel. Presumes the Specification Authorship Record's four questions already answered for the manifest in question — not usable standalone.",
      "category": "mechanics",
      "seeAlso": [
        "state-inventory",
        "specification-authorship-record"
      ],
      "firstPublished": "2026-09-09",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/your-agents-have-owners-your-policy-has-no-author-8d8f7264d3f4"
    },
    {
      "term": "Infrastructure of Intent",
      "slug": "infrastructure-of-intent",
      "source": "Series One · Article 1",
      "definition": "The complete, machine-readable specification of an enterprise digital platform: semantic variables, design tokens, business rules, interaction logic, and governance protocols, encoded in one place. It is the artifact an architect writes and the artifact every agent, developer, and product team builds against — the layer that makes 'what the system is supposed to do' a queryable fact rather than tribal knowledge.",
      "category": "mechanics",
      "seeAlso": [
        "semantic-layer",
        "agentic-constitution",
        "variables-over-pixels"
      ],
      "firstPublished": "2026-05-21",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/your-design-system-is-not-decaying-it-is-obsolete-54e0dfe52438"
    },
    {
      "term": "Invention/Assembly Boundary Test",
      "slug": "invention-assembly-boundary-test",
      "source": "Internal research — EXA Agentic Workflow, Step 7 §2 (full specification), cross-referenced in Step 6 §6 (Lexicon Additions Proposal), ratified 2026-08-11.",
      "definition": "The checkable, evidence-based test a generative surface must pass before a Generative Envelope may be drafted for it — determining whether the surface's output space is bounded assembly or unbounded invention. A surface that fails is not a candidate for a narrower envelope; it requires redesign to a certified, assembled set, or it is out of scope for agentic execution entirely. Modeled on, but distinct from, the reverse test governing the escalation path in The Amendment Protocol & Veto Charter (Standard object 5): that test checks whether an escalation path is real; this one asks whether a finite, checkable boundary can be stated for the surface at all.",
      "category": "governance",
      "seeAlso": [
        "generative-envelope",
        "generative-grammar-fuzzing"
      ]
    },
    {
      "term": "The Legacy Anchor",
      "slug": "legacy-anchor",
      "source": "Series One · Series Two",
      "definition": "The structural challenge of deploying modern semantic governance against legacy backend infrastructure that was never built to expose the states a constitution needs to govern. Not a reason to delay governance — a constraint the State Inventory process has to plan around.",
      "category": "mechanics",
      "seeAlso": [
        "dark-state",
        "agentic-constitution"
      ]
    },
    {
      "term": "The Logic-Review Gate",
      "slug": "logic-review-gate",
      "source": "Series Two · Article 3",
      "definition": "The governing mechanism that moves a design leader's review function upstream — from approving finished surfaces after the build to approving constraint definitions before it. A leader who governs constraints at the gate governs everything built downstream without reviewing any of it directly. It sits at the 'Definition of Ready' checkpoint inside an existing agile workflow, not as an additional meeting.",
      "category": "governance",
      "seeAlso": [
        "90-day-governance-playbook",
        "reviewer-interface",
        "enterprise-experience-architect"
      ],
      "firstPublished": "2026-06-18",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/hiring-for-the-artifact-instead-of-the-function-is-the-most-expensive-shortcut-you-will-take-f155abe341fe"
    },
    {
      "term": "Margin Multiplier",
      "slug": "margin-multiplier",
      "source": "Series One",
      "definition": "EXA's business-value compression: a single architectural decision that simultaneously increases build velocity, reduces error rates, improves adoption, and shortens time-to-market, rather than trading one against another. The same mechanism runs in reverse under ungoverned conditions — see The Garbage-In Multiplier and The Retry Tax.",
      "category": "economics",
      "seeAlso": [
        "garbage-in-multiplier",
        "retry-tax",
        "experience-utility"
      ],
      "firstPublished": "2026-05-21",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/your-design-system-is-not-decaying-it-is-obsolete-54e0dfe52438"
    },
    {
      "term": "The 90-Day Governance Playbook",
      "slug": "90-day-governance-playbook",
      "source": "Series Two · Article 3",
      "definition": "The operating model that runs specification governance as a standing discipline rather than a one-time hire: a sequenced rollout covering pilot selection, state enumeration, token governance, the Logic-Review Gate, and an ongoing scorecard, executed over 90 days.",
      "category": "governance",
      "visual": {
        "kind": "stat",
        "value": 90,
        "suffix": "days",
        "label": "Sequenced rollout"
      },
      "seeAlso": [
        "logic-review-gate",
        "specification-scorecard",
        "experience-architecture-maturity-model"
      ],
      "firstPublished": "2026-06-18",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/hiring-for-the-artifact-instead-of-the-function-is-the-most-expensive-shortcut-you-will-take-f155abe341fe"
    },
    {
      "term": "The Provisional Surface",
      "slug": "provisional-surface",
      "source": "Series Three · Article 1 — enforcement mechanism delivered in Article 3",
      "definition": "A 'temporary' agent behavior or interface shipped to hit a deadline and never replaced — functionally an unratified amendment to the constitution, deployed straight to production.",
      "category": "governance",
      "seeAlso": [
        "agentic-constitution",
        "governance-vacancy"
      ],
      "firstPublished": "2026-07-25",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/someone-just-asked-who-governs-your-ai-your-org-chart-has-no-answer-f332044f64d9"
    },
    {
      "term": "Reviewer Interface",
      "slug": "reviewer-interface",
      "source": "Series One · Article 2 — organizational treatment in Series Three",
      "definition": "The human role inside a governed agentic system. The practitioner's job shifts from producing screens to validating what an agent produced against what the specification required — governing the output, not operating the surface.",
      "category": "roles",
      "seeAlso": [
        "governed-autonomy",
        "enterprise-experience-architect",
        "cultural-veto"
      ],
      "firstPublished": "2026-05-27",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/ai-doesnt-need-a-better-design-system-it-needs-a-constitution-853f89c09038"
    },
    {
      "term": "The Retry Tax",
      "slug": "retry-tax",
      "source": "Series Two · Article 1",
      "definition": "The metered cost of specification ambiguity. AI inference costs generated when agents loop against an incomplete or ungoverned specification — observed running as high as roughly 50× the token cost of a single clean pass. Every retry cycle is a billable consequence of a definition gap the architecture did not close upstream.",
      "category": "economics",
      "visual": {
        "kind": "stat",
        "value": 50,
        "suffix": "×",
        "label": "Token cost of a clean pass"
      },
      "seeAlso": [
        "garbage-in-multiplier",
        "specification-economy",
        "margin-multiplier"
      ],
      "firstPublished": "2026-06-10",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/good-enough-is-the-most-expensive-thing-you-will-ever-ship-cd076eaf08c2"
    },
    {
      "term": "The Semantic Handoff",
      "slug": "semantic-handoff",
      "source": "Series One · Article 4",
      "definition": "The replacement of visual redlines with machine-readable constraint files. Where a traditional handoff gives a developer an annotated screen to interpret, a semantic handoff gives an AI coding agent a constraint file to execute — collapsing the translation step where intent traditionally got lost.",
      "category": "mechanics",
      "seeAlso": [
        "semantic-layer",
        "agentic-constitution",
        "infrastructure-of-intent"
      ],
      "firstPublished": "2026-06-05",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/the-new-deliverable-why-your-design-architect-writes-markdown-2601b645ecf7"
    },
    {
      "term": "The Semantic Layer",
      "slug": "semantic-layer",
      "source": "Series One · Article 2",
      "definition": "The machine-readable layer that sits between visual design and compiled code. It encodes why an interface behaves the way it does — the structural rules, conditions, and token bindings that make invalid states impossible — not merely what it looks like.",
      "category": "mechanics",
      "seeAlso": [
        "variables-over-pixels",
        "infrastructure-of-intent",
        "semantic-handoff"
      ],
      "firstPublished": "2026-05-21",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/your-design-system-is-not-decaying-it-is-obsolete-54e0dfe52438"
    },
    {
      "term": "Specification Authority",
      "slug": "specification-authority",
      "source": "Series Three · Article 1 — career-ladder treatment due Article 4",
      "definition": "The substantive measure of seniority in this discipline — the scope of the constitution a practitioner is empowered to ratify and enforce. Replaces 'screens shipped' as the career-ladder metric.",
      "category": "roles",
      "seeAlso": [
        "blast-radius",
        "governance-vacancy",
        "enterprise-experience-architect"
      ],
      "firstPublished": "2026-07-25",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/someone-just-asked-who-governs-your-ai-your-org-chart-has-no-answer-f332044f64d9"
    },
    {
      "term": "The Specification Authorship Record",
      "slug": "specification-authorship-record",
      "source": "Standalone — 'Your Agents Have Owners. Your Policy Has No Author.'",
      "definition": "The one-page instrument that answers what an enforcement manifest — a runtime policy manifest, for instance — doesn't: who ratified it and against what authority, which State Inventory rows it governs and which it declined, what severity tier applies and who verified it independent of who proposed it, and what triggers an amendment. Bound to the manifest it governs by a pre-merge check that resolves a specific, versioned Record or fails — an attestation does not satisfy the check, and 'N/A' does not resolve. The gate does not stay locked when production is actively failing: a five-item provisional-resumption checklist, a watched commitment date, and a standing line item on an executive's own report keep that path from becoming a quiet, permanent workaround.",
      "category": "governance",
      "seeAlso": [
        "authority-audit",
        "cultural-veto",
        "state-inventory"
      ],
      "firstPublished": "2026-09-09",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/your-agents-have-owners-your-policy-has-no-author-8d8f7264d3f4"
    },
    {
      "term": "Specification Debt",
      "slug": "specification-debt",
      "source": "Series Two, Articles 2–4",
      "definition": "The accumulated weight of unnamed conditions and untranslated business rules that compounds silently until production pays it down at the most expensive possible rate. Accrues at human speed in a manually built surface — denominated in sprint capacity and remediation cycles — and reprices at inference speed the moment an agent executes against it, without warning or a transition period. Distinct from the Retry Tax: the Retry Tax is the metered cost of ambiguity in the states a specification named; Specification Debt is the cost compounding in the states it never named at all.",
      "category": "economics",
      "seeAlso": [
        "dark-state",
        "specification-economy",
        "retry-tax"
      ],
      "firstPublished": "2026-06-16",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/the-unit-economics-of-intent-0c136df3cd46"
    },
    {
      "term": "The Specification Economy",
      "slug": "specification-economy",
      "source": "Series Two · Article 1",
      "definition": "The economic worldview of the agentic enterprise: once production approaches zero marginal cost, the specification becomes the primary unit of economic value. The only thing left worth paying for precisely is the definition of what agents are allowed to execute against.",
      "category": "economics",
      "seeAlso": [
        "retry-tax",
        "specification-scorecard",
        "infrastructure-of-intent"
      ],
      "firstPublished": "2026-06-10",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/good-enough-is-the-most-expensive-thing-you-will-ever-ship-cd076eaf08c2"
    },
    {
      "term": "The Specification Scorecard",
      "slug": "specification-scorecard",
      "source": "Series Two · Article 2",
      "definition": "The four-metric instrument that makes specification quality legible to finance: Retry Rate, Deviation Caught Before Merge, Escalation Rate, and Onboarding Velocity. Each metric carries a direct dollar translation, which is what makes specification quality defensible in a budget review rather than treated as a design preference.",
      "category": "economics",
      "seeAlso": [
        "retry-tax",
        "specification-economy",
        "logic-review-gate"
      ],
      "firstPublished": "2026-06-16",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/the-unit-economics-of-intent-0c136df3cd46"
    },
    {
      "term": "The State Inventory",
      "slug": "state-inventory",
      "source": "Informal usage on the site prior to this article (Standard Object 2's own worked example; this file's Legacy Infrastructure Collision definition) — formalized with a five-column schema in the standalone article, 'Your Agents Have Owners. Your Policy Has No Author.'",
      "definition": "The five-column enumeration that names every operational state a governed surface can enter and what the agent must do inside it: the state in operational language, the trigger condition that puts the system in it, what the agent must and must not do while it holds, the cost if it renders wrong, and where the state was found. Each row translates one-to-one into a constraint block — the machine-readable unit the Agentic Constitution Format renders. Discovery draws from five specific targets: support-escalation tickets and manual overrides, floor-level workarounds, published regulatory updates, legacy system data models and status enumerations, and production telemetry on retry and consumption anomalies. A State Inventory where every row traces to a stakeholder interview has worked none of those.",
      "category": "mechanics",
      "seeAlso": [
        "agentic-constitution",
        "specification-authorship-record"
      ],
      "firstPublished": "2026-09-09",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/your-agents-have-owners-your-policy-has-no-author-8d8f7264d3f4"
    },
    {
      "term": "Variables Over Pixels",
      "slug": "variables-over-pixels",
      "source": "Series One",
      "definition": "The first structural shift EXA asks of a design practice: at enterprise scale, logical aliases — tokens, variables, conditional bindings — outperform manually managed pixel values. It is the move from visual thinking to conditional thinking, and the precondition for anything in the system being machine-readable at all.",
      "category": "mechanics",
      "seeAlso": [
        "semantic-layer",
        "infrastructure-of-intent"
      ],
      "firstPublished": "2026-05-21",
      "sourceUrl": "https://medium.com/enterprise-experience-architecture/your-design-system-is-not-decaying-it-is-obsolete-54e0dfe52438"
    },
    {
      "term": "Vendor Specification Debt",
      "slug": "vendor-specification-debt",
      "source": "Internal research — EXA Agentic Workflow, Steps 5–6, ratified 2026-08-11",
      "definition": "The named outcome when a Vendor Envelope Expiration Protocol resolution accepts an interim fallback limitation as permanent operational debt — distinct from GM-8, the ratified Specification Scorecard metric that counts occurrences of this outcome. A specific, named species of Specification Debt, not identical to it: the parent term is any unnamed condition or untranslated business rule compounding silently, while Vendor Specification Debt names one particular, governance-tracked instance of it — a vendor-side limitation an enterprise has formally accepted as permanent rather than resolved.",
      "category": "governance",
      "seeAlso": [
        "specification-debt",
        "specification-scorecard"
      ]
    }
  ]
}