The AI Governance Canon — Reading Path

This is not a blog archive. It is a curriculum.

Eleven entries, ordered by the story they tell together. Each builds on the ones before it. You can start anywhere, but the signal compounds when read in sequence.

If you only read one path, read the one for your role. If you read two, read your role path and then the full arc.


The Full Arc (Chronological)

The canon is organized in four movements: Foundation → Diagnosis → Framework → Economics. This is the order in which the problems were identified and the solutions were built.

Part 1: The Foundation

Entry #1 — The AI Gateway: Governance as Enterprise Control Plane March 24, 2026

The thesis that started everything. Every enterprise AI interaction must route through a governed API proxy layer — not direct user-to-model access. Governance is the architecture, not a compliance document. Published before most enterprises had deployed their first AI agent.

Entry #2 — RAG Data Integrity Model April 22, 2026

Hallucinations are data architecture failures, not model failures. When conflicting source records enter a vector index without upstream reconciliation, the RAG pipeline embeds all versions and outputs garbage. Core principle: vector indexes are derived views, not sources of truth.

Entry #3 — AI Hype Cycle May 29, 2026

A timestamped note of caution. Three first-principles tests — commodity moat, last-mile accuracy deficit, physical infrastructure constraints — applied to generative AI specifically. Historical comparisons to dot-com, blockchain, and metaverse cycles. The patterns are the same. The speed is faster.

Part 2: The Diagnosis

Entry #4 — Enterprise AI Failure Pattern July 10, 2026

Why 95% of enterprise AI deployments will collapse. Not because AI is immature — because the enterprise is. Most companies bolt AI onto fragmented systems, building adaptors and stitching layers that become permanent. The debt is compounding and irreversible.

Entry #5 — Post-Peak AI Economics August 1, 2026

The money story begins. Front-loaded gains in Years 1–3, sharp ROI flattening in Years 4–5. Inference, retraining, and monitoring costs do not decline proportionally. Most AI business cases assume the early curve continues. That assumption will be revised downward.

Entry #6 — The Data Layer Is Where AI Goes to Die September 8, 2026

AI does not clean, unify, or reconcile data — it amplifies whatever mess already exists. Garbage semantics in, garbage semantics out. The fix is upstream data contracts, not better prompts. Directly feeds into the RAG Data Integrity Model (Entry #2).

Part 3: The Framework

Entry #7 — GATES AI Governance Framework October 26, 2026

The prescription. Five runtime gates — Secure, Govern, Authenticate, Trace, Enforce — each scored 0–5 for maturity. Built from 12+ years of enterprise API gateway operations, not AI lab research. Designed to complement NIST AI RMF and ISO/IEC 42001, not replace them.

Entry #8 — Who Governs the Governor? October 26, 2026

The threat that proves why the framework exists. Stored prompt injection through the API gateway — adversarial text enters a database through a legitimate write, then an AI agent reads it as instruction. Every API call was authorized. The exploit lives in the gap between two correct decisions.

Part 4: The Economics

Entry #9 — The Collapse Curve™ November 3, 2026

The tool. A four-step DIY framework to calculate when your AI program crosses into negative contribution. Six inputs, four steps, one answer. Most programs hit the wall in Year 4–5. Run your own numbers.

Entry #10 — The Knowledge Balance Sheet™ November 7, 2026

The hidden cost. A diagnostic lens for two parallel risks: AI economics that decay on one timeline, and institutional memory that depletes on another. The Triple Liquidity Collision™ (~2028) is where they converge — and no one is left who can recover.

Entry #11 — Costs You Can’t Attribute, You Can’t Govern November 14, 2026

The mechanism. The party creating the AI cost is often not the party paying, so nobody has a reason to ask for a limit. Four questions to fix it. Attribute first, cap second. Rough beats nothing.


Role-Based Paths

Not everyone needs the full arc. Here is where each role should start and what to read next.

For CFOs and Finance Leaders

You need to understand the cost structure before it surprises you.

  1. Entry #5 — Post-Peak AI Economics (the curve exists)
  2. Entry #9 — The Collapse Curve (calculate your own timeline)
  3. Entry #11 — Costs You Can’t Attribute (who owns the bill?)
  4. Entry #10 — Knowledge Balance Sheet (the cost that doesn’t show up)

Total reading time: 25 minutes. What you will know: whether your AI program is on track to create or destroy value.

For CTOs and CAIOs

You need to build the governance before the regulators arrive.

  1. Entry #1 — AI Gateway (governance is architecture)
  2. Entry #7 — GATES Framework (the 5-gate solution)
  3. Entry #8 — Who Governs the Governor? (the threat you must prevent)
  4. Entry #2 — RAG Data Integrity Model (hallucinations are data problems)

Total reading time: 30 minutes. What you will know: what to build, in what order, and why.

For Board Members and Audit Committees

You need to know what questions to ask and what liabilities exist.

  1. Entry #4 — Enterprise AI Failure Pattern (why most fail)
  2. Entry #10 — Knowledge Balance Sheet (the hidden liability)
  3. Entry #11 — Costs You Can’t Attribute (the ungoverned spend)
  4. Entry #7 — GATES Framework (the maturity scorecard)

Total reading time: 20 minutes. What you will know: whether your organization is accumulating AI debt or building AI capability.

For PE Partners and Due Diligence Teams

You need to assess AI risk in a target before you sign.

  1. Entry #9 — The Collapse Curve (stress-test their ROI claims)
  2. Entry #11 — Costs You Can’t Attribute (find the margin leakage)
  3. Entry #4 — Enterprise AI Failure Pattern (is their foundation sound?)
  4. Entry #8 — Who Governs the Governor? (do they have runtime controls?)

Total reading time: 25 minutes. What you will know: whether the target’s AI program is an asset or a liability.

For New Readers (Start Here)

You have no context. This path builds it from zero.

  1. Entry #1 — AI Gateway (the thesis)
  2. Entry #4 — Enterprise AI Failure Pattern (the diagnosis)
  3. Entry #7 — GATES Framework (the solution)
  4. Entry #9 — The Collapse Curve (the economics)
  5. Entry #10 — Knowledge Balance Sheet (the hidden cost)

Total reading time: 35 minutes. What you will know: the full landscape of enterprise AI governance.


Dependencies

Some entries assume you have read others. Here is the dependency graph:

Entry #1 (AI Gateway) → Entry #7 (GATES)
  GATES extends the gateway thesis into a working framework.

Entry #2 (RAG Data Integrity) → Entry #6 (Data Layer)
  The Data Layer article provides the root cause; RAG Data Integrity provides the model.

Entry #4 (Failure Pattern) → Entry #9 (Collapse Curve)
  The Failure Pattern diagnoses why; the Collapse Curve gives you the tool to measure it.

Entry #5 (Post-Peak Economics) → Entry #9 (Collapse Curve)
  Part 1 sets the stage. Part 2 gives you the calculator.

Entry #7 (GATES) → Entry #8 (Who Governs the Governor?)
  GATES provides the framework. Who Governs the Governor? stress-tests it against a real attack.

Entry #9 (Collapse Curve) → Entry #10 (Knowledge Balance Sheet)
  The Collapse Curve measures the economic timeline. The Knowledge Balance Sheet measures what else you are losing.

Entry #10 (Knowledge Balance Sheet) → Entry #11 (Cost Attribution)
  The Knowledge Balance Sheet identifies the hidden liabilities. Cost Attribution gives you the mechanism to fix one of them.

Supporting Material

These are not part of the numbered canon but provide context, depth, or alternative angles.


How This Page Is Maintained

New entries are added to the canon in chronological order as they are published. The reading paths are updated when new material changes the recommended sequence. The dependency graph is reviewed with each new entry.

Last updated: November 18, 2026. Entry #11 of 11.

Back to AI Governance · Glossary · About