Skip to content

PROCESS · AI

How I work with AI

A look at the tools I use daily, the principles I follow, and the lines I won't cross.

The tools I use

Four tools that earn their weight in my daily workflow.

Claude

Long-form thinking

Where I do my heaviest drafts — articles, briefs, and strategy docs. The place where AI thinking actually changes the output.

ChatGPT

Quick analysis

Fast Q&A, summaries, and ad-hoc exploration. My pocket calculator for things I don't need to think hard about.

Gemini

Translations

First-pass Spanish translations from the English originals. Every line gets a human pass before it ships.

Cursor

Code

For this site and small side tools. Speeds up the boilerplate, doesn't replace the architecture decisions.

My principles

Three rules I keep when AI is in the loop.

01

Human in the loop

AI drafts, I edit, I ship. Nothing goes out that I haven't read and re-shaped in my own words.

02

Disclose when it matters

Anything client-facing gets a note. Anything that shapes an opinion, a strategy, or a recommendation — disclosed.

03

Don't trust, verify

Every output gets checked before it lands. Sources, numbers, and quotes get verified by hand, not by the model that produced them.

What I won't do

Lines I keep. Non-negotiable.

  • No client work goes through AI without my hand on the output.
  • I won't pretend AI wrote something I wrote, or vice versa.
  • I don't share client data, contracts, or unpublished strategy with model providers.

Want help thinking through AI for your team?

A 45-minute call is usually enough to find the leverage points. No pitch — just a working session.

Book a call
Copied