Writing on AI, leadership, and the work itself.
Field notes from rebuilding a business on AI, and from helping founders build an AI team inside their own companies.
-
Prompt Engineering vs Context Engineering: What a Business Owner Needs to Know
Prompt engineering is how you word the ask. Context engineering is what the AI knows when you ask. What to check when an AI answer comes back weak, and a two-question check to run before you type.
-
Automate or Augment? A Job-by-Job Test for Small Business Owners
Should you automate a job or augment the person doing it? Run four yes/no questions on one job at a time. Some jobs should be automated. The rest get AI in a supporting role, like a reviewer.
-
Why Do AI Pilots Stall After the Demo?
Most AI pilots in a 5 to 50 person firm stall for reasons that have little to do with what the tool can do. Nobody owns the first job, and nobody said what 'used' means. Here is a smaller start that holds: one job, one owner, one week-two test.
-
Your Website Is a Business Card
A referred client trusted me before our first call. Where website credibility fits, and how I now write regularly from calls and notes I already had.
-
Start With Meeting Prep, and the Follow-Up You Keep Dropping
The first AI job a founder should hand over is meeting prep and the follow-up loop after it. Not because meetings are special, but because that prep is work you already do badly under pressure and the follow-up is a loop you already drop.
-
The Vacation You Never Actually Take
Most founders already know the fix for bad handoffs. Here's why knowing it has never been enough.
-
You Sent the AI Memo. Your Org Chart Didn't Move.
Every founder who rolled out ChatGPT seats can name the tool. Almost none can name the seat it replaced. Here's the test that catches the gap before your next job req does.
-
AI Is a Tool, Not a Genie
The machine did exactly what I asked and the result was unusable. That's not a bug in the tool. It's the oldest story we have about getting what you asked for, and it's the reason your people can't leave the room.
-
AI Didn't Get Too Expensive. We Deployed It Like Idiots.
The AI bill came due at Meta and Microsoft. The tech didn't fail, the rollout did. Four mistakes that made AI cost more than the people it replaced, and the fix.
-
I Had a Frontier Model for a Week. I Used It to Audit My Own Systems.
I had Anthropic's Fable 5 for a short window. Instead of building more, I pointed it at my own systems. The model is 10 to 20 percent of the work. The system it plugs into is what survives the swap.
-
The Second Desk: The AI Team I Built Because My Ideas Kept Dying in the Handoff
Working Genius named the exact gap I'd had for years. No genius in Tenacity or Galvanizing. So I built an AI team to close it. This is what that looks like.