Artifacts.
Diagrams, tables, and frames from the analytical work. Each artifact stands on its own and travels between decks, one-pagers, and conversations. Most also appear inside one or more library essays; those essay pages link back here.
25 artifacts, each on its own page with a downloadable SVG and cross-links back to the essays that cite it.
The decision chain
The chain every diagnostic piece in the library traces, and the links a piece can name as the place it stops working: construct, representation, inference, decision, action, consequence, and the return leg that carries what the outcome showed back into the next measurement.
Decision-System Architecture — the four disciplines
The umbrella frame of the Analytic Bytes Library at a glance.
The Contract at the Seam
Integration moves the data.
The Decision System — reference architecture
A tool-agnostic reference architecture: sources through integration, warehouse, and the semantic-layer keystone to AI and the reporting surfaces, with a governance rail across every layer and a learning loop that closes the system.
One Architecture, Three Stacks
The same six-layer architecture instantiated three ways — Microsoft/Fabric, the modern data stack, and lean/open — showing the tools swap while the architecture holds.
The Agent System
An agentic-AI architecture: five named agents — Data, Analysis, Insight, Execution, Monitoring — operating the Signal–Decision–Action loop, with monitoring closing the loop and a human-in-the-loop rail across every agent.
Decision Load vs Decision Capacity
AI raises both an organization's decision load and its decision capacity.
The Data Role Landscape
Data-leadership roles distributed across the five stages of the decision arc — build the system, govern the system, interpret the signal, support the decision, own the decision.
Reliability vs Validity
The four-target view of the AI scoring trap: a model can agree with human raters at a high rate (reliable) and still measure the wrong thing (invalid) — a tight cluster, off the bullseye.
The Validity Ladder
Five rungs of evidence for an AI system.
Fair for Whom?
Fairness reframed as validity asked one subgroup at a time.
The Evidence Spine
The measurement-and-evaluation architecture that turns monitoring into learning: a living theory of change as keystone, harmonized assessments, and one semantic layer so every audience sees numbers that agree.
Measurement = Diagnostics
A sixteen-row translation table from educational measurement vocabulary to medical diagnostics — across foundations (validity, reliability), models (IRT and ROC, standard setting and thresholds, equating and calibration), bias and equity, stakes and decisions, standards and integrity, and the inferential closer: validity argument and differential diagnosis.
Higher Ed = Healthcare
An eighteen-row translation table mapping higher-education data and analytics vocabulary onto healthcare equivalents — across outcomes, throughput, advising and care navigation, support programs, infrastructure (SIS/EHR, NSC/HIE, 1EdTech/FHIR), regulation, accountability, equity, and integrative philosophy.
K-12 = Healthcare
An eighteen-row translation table mapping K-12 data and analytics vocabulary onto healthcare equivalents — across outcomes, intervention workflow, infrastructure, regulation, accountability, and integrative philosophy.
Commercial = Mission-Driven
A fourteen-term translation table from commercial vocabulary — GTM, audience, segmentation, funnel, conversion, KPIs, OKRs, ROI, LTV, runway, churn, A/B testing, MVP, CI/CD — to its mission-driven equivalents.
The essential minimum — mission-driven AI evaluation
Five components of one discipline: task decomposition, ground-truth benchmarking with constrained data, deployment-context evaluation, downstream impact evaluation, and escalation with human-in-the-loop discipline.
Three lenses on the same practice
Six task categories from AB's 90-day audit, read through three measurement lenses: raw session count (throughput), weighted-proxy volume (attention estimate), and real token share (cost measurement).
Multi-tool selection map
Where the work lives now — a snapshot of current-state multi-tool selection across Claude, GPT, and Gemini.
Governance Craft Card
One-page synthesis of AB's 90-day audit and its methodology anchors — the Executive Card.
Governance Craft Log
The 23 discipline moves catalogued during the audit's own execution, phase by phase, plus 7 emergent methodology principles the close surfaced.
Dialogue Maturity Curve
116 threads scored on a six-metric rubric across the twelve-month arc (88 GPT + 28 Claude).
AB Governance Maturity Scorecard
Seven governance dimensions scored against current baseline and next-90 targets.
The Validity Quadrant
A validity quadrant for any cost-per-outcome claim.
Decision Rights at the Door
Three panels of the same sixty-six job postings, each answering a different question about decision rights: what decisions the hire may make unaided, what an AI system may decide, and whether the hire may halt or gate a model.