Notes from working AI-natively

I'm Max. I run my week with Claude Code.
This is what I've built and what it's taught me.

A library of skills and prompts I use every day. Writing on what works and what doesn't. If something here is useful, take it and run.

See the library Read the writing →

How I work

Claude Code at the centre of the week.

  1. 1

    Wired into the work

    MCP servers for calendar, docs, and email. Claude Code reads what I need before I ask, drafts what I'd otherwise write from scratch, and keeps the context across sessions.

  2. 2

    Skills, not scripts

    The recurring jobs that fill the week (board prep, strategy memos, hiring scorecards, investor updates) run through skills I've built and refined. Most of them live in the library, MIT-licensed and free to fork.

  3. 3

    A cadence for noticing

    Friday reviews, prompts I ask myself on a schedule, agents that watch for things that would otherwise slip. The system is for noticing, not for output.

Library

Take what's useful.

Skills, schedules, and CLAUDE.md fragments from my own setup. Every one with a working example and a template you can adapt. MIT-licensed.

Browse the library →

Writing

From the road.

Essays on AI, leadership, and the gap between knowing something is possible and actually doing it. A new one every couple of weeks.

Read the writing →   Get new essays by email →

Things people ask

Same questions, every time.

Do you need to be technical to work like this?

Less than you'd expect. The terminal looks intimidating for about a week, then becomes invisible. Most of the work is writing clearly about what you want, and senior leaders already do that for a living.

What does Claude Code cost?

It's part of an Anthropic subscription. The Max plan at around $100–$200/month covers most personal use. Heavy use can go higher. Billed by Anthropic, nothing to do with me.

Is the data safe?

Claude Code runs on your machine. The MCP servers I use only have the access I grant them, and that access is per-tool, not per-account. Nothing leaves the environment that I haven't asked it to send.

How long did this take to set up?

The first useful version took an afternoon. The version I run now is the result of about six months of small adjustments. Most of what I learned in those six months is somewhere on this site.

Follow the work.

A new essay every couple of weeks, on AI-native leadership and the system behind it. No spam, no pitch, unsubscribe anytime.

Get new essays by email

Or, if something here lands or you've built something similar, message me on LinkedIn. I read everything.