Mental Models & Perspectives

Thinking

How I think about problems, make decisions, and lead teams. Some ideas here are frameworks I've refined over years, they're stable and I apply them daily. Others are perspectives, my current thinking on where things are heading, which may evolve as I learn more.

Frameworks

Stable mental models I apply daily. Refined over time, not tied to a moment.

Mental Models

Thinking Tools for Decisions

A working collection of mental models I reach for when making decisions, diagnosing problems, or evaluating opportunities.

Goal Setting

S.M.A.R.T. + Stretch + Pyramid

Three rules for goals that compound: S.M.A.R.T. + Stretch (both required), three-tier difficulty (~70% target), and pyramid alignment from individual to mission.

Management at Scale

Leading Cross-Functional Teams

Principles for scaling operations across countries, cultures, and functions without losing speed or quality.

Coming soon

Lifelong Learning

The Feynman Technique in Practice

Why I build things to understand them. Learning by doing, teaching to solidify, and staying hands-on as a leader.

Coming soon

Perspectives

Point-in-time views that may evolve. Dated to show when I thought this.

Your AI Is Confidently Quoting Stale Docs. The Fix Is 20 Years Old.

Jun 2026

Single-sourcing (write once, publish everywhere) is a decades-old discipline. AI didn't make it obsolete, it raised the cost of ignoring it, because a machine can't tell it's reading the stale copy and repeats it confidently, at scale, to everyone who asks.

AI Has Been Reading the Web for Free. That's Starting to Change.

Jun 2026

AWS just shipped the infrastructure to charge AI per request (HTTP 402, paid in stablecoin at the edge). The catch: the crawlers can't pay yet, and for most sites you wouldn't want them to. The web is splitting into two economies.

I Gave My Website an MCP Server. No AI Can Find It Yet.

Jun 2026

I exposed this site as an MCP server, tools an agent can operate, not just text to scrape, in an afternoon. As of mid-2026 no LLM auto-discovers it, and part of that gap is deliberate. The endpoint is given, not discovered.

Vibe Coding: What 6 Months Taught Me

Jun 2026

A product leader, not an engineer, on building real software with AI. The capability jump that 10x'd the work, the pixel-art game that walled me out, and what to tell a Chief AI Officer.

AI-DLC

May 2026

The methodology is real. The tools are usable. The bottleneck isn't tooling, it's measurement. Humans systematically misjudge AI output quality.

Where AI is Heading

May 2026

Capability is on a clear curve. Enterprise absorption isn't. The companies that win the next five years will solve absorption, not chase capability.

Connect your AI assistant

Pull these articles & playbooks straight into your tools via MCP

This site runs a Model Context Protocol server. Add it to Claude, Cursor, or any MCP-capable client and your assistant can read and search my full thinking, perspectives, frameworks, the product-management playbook, and my professional profile, as first-class, citable sources. Public, read-only.

Claude Code

claude mcp add --transport http arthur-brejon https://arthurbrejon.com/api/mcp

Claude Desktop / Cursor (config JSON)

{
  "mcpServers": {
    "arthur-brejon": {
      "url": "https://arthurbrejon.com/api/mcp"
    }
  }
}

Endpoint: https://arthurbrejon.com/api/mcpTools: list_content, get_content, search_content