← Field Guide / Insights

Field Guide · Reference Index

Enterprise taxonomies.

Ten lookup tables, designed to be quotable as shorthand — “that’s a V3 task at A2 autonomy.” They compose: an action’s risk class plus its verification class determines its allowable autonomy tier and required oversight mode.

Rios Applied AI · Taxonomy set v1 · August 2026

T1 — AI OPERATING MATURITY (M0–M4)

LevelNameDescriptionTelltale
M0ExploratoryIndividuals use AI tools ad hoc; no institutional footprintShadow ChatGPT usage
M1AssistedSanctioned copilots; humans do the work, AI accelerates itLicenses purchased, workflows unchanged
M2OrchestratedDeterministic workflows with embedded model stepsFirst real engineering; evals appear
M3DelegatedAgents own bounded tasks within tiered autonomyAction inventories, provenance, oversight design exist
M4EngineeredAI portfolio managed on unit economics, eval coverage, and absorption rate as standing functionsAI indistinguishable from "how operations works"

The load-bearing distinction is M1→M2 (tool adoption becomes engineering) and M2→M3 (engineering acquires governance). Most enterprises claiming M3 are at M1 with demos.

T2 — AGENT AUTONOMY TIERS (A0–A5) — assigned per action class, never per system

TierNameBehavior
A0AdvisoryProduces drafts and recommendations only; humans execute
A1GatedExecutes, but every action requires pre-approval
A2Exception-gatedExecutes freely; policy triggers route defined cases to humans
A3SampledExecutes; humans review a statistical sample post-hoc
A4MonitoredExecutes; automated monitoring with human escalation on anomaly
A5DelegatedOutcome-level review only

A single agent legitimately operates at different tiers simultaneously — A4 for reading and drafting, A1 for payments. A single system-wide tier does not show whether an action inventory was performed — ask for it.

T3 — HUMAN OVERSIGHT MODES

ModeHuman positionAppropriate when
In-loopApproves each action before executionRisk class IV; early trust gradient
On-loopWatches live, can interveneTransitions; high-tempo workflows
SampledReviews statistical sample after executionRisk class II with evidence base
Exception-onlySees only policy-triggered casesMature A2/A4 operation
Outcome-onlyReviews results, not actionsRisk class I; verification is mechanical

The design failure to name in workshops: nominal in-loop — a human formally in the loop who, once approval volume outruns the attention available, is functionally outcome-only while carrying in-loop accountability.

T4 — ENTERPRISE KNOWLEDGE CLASSES

ClassDescriptionAgent-readiness
K1 StructuredSystems of record; transactional truthHigh — the trust anchor; everything derives from here
K2 RecordedDocuments, wikis, policiesMedium — abundant, stale, contradictory; needs source-of-truth tiers
K3 ProceduralHow work is actually doneLow until made legible; the target of legibility programs
K4 TribalUndocumented know-how in tenured headsZero until extracted; the largest hidden dependency
K5 JudgmentTacit criteria for good decisionsPartially extractable as resolved ambiguities and graded examples; the rest is what humans are for

The strategic insight: RAG programs concentrate on K2, the least differentiating class. Competitive advantage lives in converting K3/K4 to legible form and encoding K5 into evaluation criteria.

T5 — AI GOVERNANCE CONTROL TYPES

TypeNatureExample
StructuralWhat the system can doScoped credentials, egress controls, tool allowlists
Runtime-preventiveDeterministic gates on actionsThreshold checks, policy-as-code, approval routing
DetectiveDiscovers deviationEvals, canaries, distribution monitoring, reconciliation
CorrectiveUndoes or contains damageRollback, compensating transactions, kill switches
ProceduralHuman processReview boards, attestations, training
DocumentaryEnables reconstructionProvenance, traces, decision records

Governance maturity is the migration of weight leftward: immature programs are procedural and documentary; mature ones are structural and runtime, with procedure reserved for judgment.

T6 — EVALUATION MATURITY (E0–E4)

LevelNameDescription
E0AnecdotalDemos and spot checks; confidence by vibes
E1StaticFixed test set run at release time
E2RepresentativeProduction-sampled inputs, graded scoring, periodic refresh
E3ContinuousRuns on every change — including vendor model updates; drift alarms; correlated-grader controls
E4InstitutionalEval assets versioned and owned; grader–human agreement tracked; production disagreements feed the suite; eval results gate change management

E3 is the minimum for anything in risk classes III–IV. The E2→E3 jump is organizational, not technical: it requires someone to own evaluation as a standing function.

T7 — WORKFLOW COMPLEXITY CLASSES (W1–W5)

ClassShapeAutomation posture
W1Linear, deterministicClassic automation; models add nothing structural
W2Branching, rules-expressibleOrchestration with model steps for perception/extraction
W3Exception-heavyValue concentrates in exception mass; census before building
W4Judgment-denseAgent prepares, human decides; encode judgment as eval criteria over time
W5Open-endedGenuine agentic planning; smallest category, highest risk, most hype

The most common scoping error is treating W3 as W2 (the exceptions surface in production) and W4 as W5 (deploying open-ended agents where judgment placement was the actual problem).

T8 — VERIFICATION COST CLASSES (V0–V4)

ClassVerificationExamplesImplication
V0Mechanical, freeCompiles, schema-validates, reconcilesLong chains safe; automate aggressively
V1Rule-checkablePolicy lint, balance checks, format contractsCheap gates; near-V0 economics
V2Spot-checkableStatistical sampling sufficesA3 autonomy viable; verification budget models the sample
V3Full expert reviewLegal positions, strategic analysisVerification ≈ generation cost; agent drafts, human owns
V4Outcome-delayedAdvice quality, credit decisionsCorrectness observable only later; requires counterfactual baselines and cohort monitoring

This is the taxonomy that does the most portfolio work: V-class × risk class predicts deployability better than any capability benchmark.

T9 — ACTION RISK CLASSES (I–IV)

ClassReversibilityBlast radiusMinimum posture
IReversibleContainedA3–A5; outcome/sampled oversight
IIReversibleBroadA2–A3; pre-execution checks, monitored rollout
IIIIrreversibleContainedA1–A2; runtime gates, full provenance
IVIrreversibleBroadA0–A1; human approval always, dual control, documentary completeness

The bands above are risk-only: they assume verification is not the binding constraint. Where it is, the composed rule governs — the Workflow Diagnostic assigns autonomy on risk class and verification class together, and tightens below these bands as verification gets more expensive. A Class I action that requires full expert review to check (V3) does not get A3–A5; it gets A2. The published matrix is on How this tool decided. Where this table and that matrix disagree, the matrix is what actually runs.

T10 — OPERATIONAL INTELLIGENCE LEVELS (OI1–OI5)

LevelCapabilityQuestion answered
OI1DescriptiveWhat happened?
OI2DiagnosticWhy did it happen?
OI3PredictiveWhat will happen?
OI4PrescriptiveWhat should we do?
OI5Closed-loopThe system acts — governed by autonomy tiers and risk classes

OI5 is where operational intelligence merges with agent governance: the moment analytics acquires hands, every taxonomy above applies to it.