I wrote the guide I wish existed when I started integrating AI into real work. No hype. No theory. Just the workflows that actually hold up when the stakes are real.
Start with Lesson 1 →Most people fall into one of two traps with AI: they either trust it too much and ship work they haven't thought through, or they distrust it entirely and miss the leverage it actually provides.
There's a middle path. These 8 workflows come from 30+ years of designing for high-stakes environments — and the last few years of figuring out where AI genuinely helps and where it quietly makes you worse at your job.
Each lesson is a pattern you can apply today. A problem you'll recognize, a workflow you can follow, and the traps to watch for once you start.
Pick one. Apply it. Then come back for the next.
Stop shipping plausible-but-wrong AI output. A 5-step pause-and-audit pattern that catches the 40% of AI work that sounds right but isn't.
Tell AI what NOT to do before asking what to do. Narrow the output space so edits become refinement, not reconstruction.
AI organizes what you've learned — it doesn't learn for you. Maintain epistemic responsibility while accelerating the draft.
Automate the toil. Keep the skills that make you dangerous. The 30% manual rule that protects your future pricing power.
Review is work. If you're spending 30 minutes fixing 10 minutes of AI output, the AI didn't save time — you just commoditized your thinking.
15 variations of mediocre isn't progress. Learn to iterate toward something instead of iterating to avoid deciding.
Know what you know. Mark what you don't. Manage uncertainty honestly instead of pretending AI gave you expertise.
Augmentation degrades gracefully. Automation breaks catastrophically. Build workflows that survive the edge cases.
Are you learning your craft, or optimizing it away? If you haven't manually written a headline in 6 months, you're optimizing away.
Can you ship work without the tool? If no, you're dependent, not augmented.
Are you explaining decisions, or just explaining outputs? If the latter, AI is driving.
Does the client value your judgment, or your output speed? If speed, you're vulnerable to cheaper tools.
Would you stake your reputation on this, AI-assisted or not? If not, you're not done thinking yet.
Start with Lesson 1 and build from there.
Begin The AI Playbook →