Notes from the operating layer

AI execution under real constraints

The writing draws on a decade of operating work, now centered on AI execution: the leadership, infrastructure, reliability, cost, and governance systems that determine whether AI becomes durable business capability or organizational theater.

Read the operating thesis Operating context →
Canon10
Topics49
Covers2016–26
Latest2026.09.25

A decade of practice

These notes span ten years of operating work, from container reliability and security incident response to the AI operating model. The throughline is the same discipline applied to a moving target.

2016–2018 Infrastructure discipline Containers, databases, reliability: the production fundamentals.
2019–2021 Platform and scale Internal platforms, observability, multi-region, FinOps.
2022–2026 The AI operating layer When models met production reality, and discipline became the differentiator.

The recurring question

// The recurring question

What has to be true for this system, team, or strategy to keep working when the model, the vendor, the cost curve, or the organization changes?

The answer is rarely a better model. It is usually a clearer operating model. AI does not remove the need for operating discipline; it raises the cost of operating without it.

// Canonical reading

  1. No. 01 Unsupervised AI-NATIVE OPERATING MODEL · AI-OPERATING-MODEL Producing work is close to free; knowing whether it is any good, undoing it, and answering for it are not, and that is where the weight of a company now sits.
  2. No. 02 Build the System the Model Cannot Break AI-NATIVE OPERATING MODEL · AI-OPERATING-MODEL An AI-native company is not the one that adopts the model fastest; it is the one whose operating model the model cannot break.
  3. No. 03 The Throughput Engineer: Why Headcount Is a Lagging Metric THROUGHPUT CULTURE · AI-OPERATING-MODEL Headcount is a lagging metric; the real throughput ceiling is how fast an organization can decide.
  4. No. 04 The CTO Communication Protocol for AI Programs CTO COMMUNICATION PROTOCOL · AI-OPERATING-MODEL AI programs fail when leadership communication stays ad hoc instead of becoming an operating protocol.
  5. No. 05 Why Most AI Platform Teams Become the New Bottleneck PLATFORM BOTTLENECKS · AI-OPERATING-MODEL A central AI platform team becomes a liability when every workflow improvement has to wait in its queue.
  6. No. 06 AI Roadmaps That Survive Contact With Reality REALITY-TESTED ROADMAPS · AI-OPERATING-MODEL An AI roadmap is only real if it can survive latency, ownership, and workflow constraints in production.
  7. No. 07 Decision Latency as a P&L Variable: The Leadership Metric Nobody Owns DECISION LATENCY · AI-OPERATING-MODEL Decision latency is a P&L variable because slow organizational decisions destroy AI leverage before the model does.
  8. No. 08 Designing the AI Leadership Bench: Roles, Interfaces, and Failure Boundaries LEADERSHIP BENCH · AI-OPERATING-MODEL Serious AI execution needs a leadership bench with explicit role interfaces, not a heroic single-threaded leader.
  9. No. 09 The Operating Cadence: Turning AI Leadership Interfaces Into Predictable Output OPERATING CADENCE · AI-OPERATING-MODEL Leadership interfaces only compound when the organization runs them on a predictable cadence.
  10. No. 10 The Post-Prototype AI Org: Operating Models That Survive Year Two POST-PROTOTYPE AI ORG · AI-OPERATING-MODEL The hard part of AI starts after the prototype, when the company has to become an organization that can actually run it.

Latest writing

/blog →
A Backup Model You Never Exercise Is Not a Backup · 4 min Vendor count is inventory; measured failover time is resilience. Shadow replay, a bounded canary, and a recovery clock with defined start and stop. ai reliability strategy The Handoff Tax · 4 min 80% report personal AI gains while EBIT impact stays flat. The handoff tax: task time down, throughput flat, queues rising at one nameable transition. leadership operations ai The Statistic Nobody Can Reconstruct · 4 min I tried to reconstruct '95% of AI pilots fail.' It isn't in the report it cites. Card the numbers that steer your decisions: licensed, expired, or retired. metrics ai executive Don't Book the Cut Before the Work Disappears · 4 min Expected AI headcount cuts outran reported ones last year. Let forecasts drive scenarios; book a saving only once the work has observably gone. leadership teams ai Autonomy Without a Demotion Path Is Permission Creep · 4 min Teams can describe how agents earn autonomy. Few can take it back. Write the transition contract: trigger, authority, mechanics, re-entry. ai governance reliability The Software You Didn't Buy Still Needs an Owner · 4 min Teams skip software purchases and build with agents. Each build is rational; the fleet is the risk. Agents compress upkeep but can't own it. strategy ai cost MCP Grew Up. Your Integration Debt Has a Clock. · 4 min The July MCP revision started a 12-month deprecation clock. Use it to find every agent-to-tool boundary, give each an owner, and test it. ai architecture platform-engineering The Rung Contract · 4 min Competency matrices describe skills. Apprenticeship needs a contract per rung: permission, evidence, reviewer, a promotion rule two managers apply alike. leadership teams ai
See all 350 entries →

Coverage

Where the writing concentrates. Every topic is grounded in production work, not commentary.

AI Operating Systems 151 AI and Systems Architecture 95 Engineering 73 Technical Leadership 57 Go 53 Infrastructure 53 DevOps 51 LLMs 47 Security 45 Teams 43