Ideas
The archive
Essays and briefings on how organizations actually move up the maturity curve.
-
The Advantage Was Never the Model
Every company can rent the same intelligence. Six in a hundred have turned it into profit. Why the gap opens, where it closes, and where AI should stop.
-
The Agents Are Already Inside
In July, at least 1,200 AI agents left an OpenAI test sandbox and spent four days inside Hugging Face on stolen credentials. The control that holds against a compromised agent is a person with the authority to stop it, and most companies have not appointed one.
-
The Technology Was Ready on Day One
AI, without the jargon: Day 21. The system worked within hours. The judgment around it took weeks, and that judgment was worth more than the model.
-
The Clock-Out Rule
AI, without the jargon: Day 20. An agent never gets tired, never feels unsure, and never glances at the clock. Three lists have to exist before it goes live.
-
When the Answer Has Somewhere to Go
AI, without the jargon: Day 19. A chatbot answers questions. An agent takes steps. The difference is what you do after the answer arrives.
-
Ownership Follows Outcomes
AI, without the jargon: Day 18. IT installed the tool, but IT does not miss a target when it underperforms. Give the decisions to the person who does.
-
Start at the Queue
AI, without the jargon: Day 17. Look for the bottleneck after the AI step, not before it. The queue is where your redesign begins.
-
Flying the Plane, Not Designing the Airport
AI, without the jargon: Day 16. Do not ask how AI can help your current process. Ask what process you would build if you started today.
-
The Speed Never Reached the Loading Dock
AI, without the jargon: Day 15. AI made every step faster and the output stayed flat. The bottleneck was never the work. It was [the space between the work](/ideas/awtj-17-start-at-the-queue).
-
Proof From a Peer
AI, without the jargon: Day 14. Half your team uses AI daily, half have not opened it since training. The gap is experience, not understanding.
-
Training Didn't Move Them. Pain Did.
AI, without the jargon: Day 13. Eight recruiters sat through AI training twice and never opened the tools again. One question changed it.
-
The Worker Starbucks Fired
AI, without the jargon: Day 12. It had been on the job nine months. It was an AI. What happened next is the honest answer to whether AI takes your job.
-
Customer, Contract, Cash
AI, without the jargon: Day 11. If a person reads every line the AI writes, the tool has saved nothing. You have added a step.
-
Six Cases That Never Existed
AI, without the jargon: Day 10. AI mistakes are dangerous because they do not look like mistakes. The wrong answer arrives in the same confident tone as the right one.
-
The Knowledge That Leaves When People Do
AI, without the jargon: Day 9. Teaching AI about your business is a writing job before it is a technology job. The technology is the easy part.
-
Everything In General, Nothing About You
AI, without the jargon: Day 8. You tried ChatGPT, it worked, and nothing changed. Here is the step between trying it and getting value from it.
-
The Factories That Kept Their Belts
AI, without the jargon: Day 7. When factories first got electric motors, almost nothing improved. It took thirty years to work out why.
-
One Metric, Thirty Days
AI, without the jargon: Day 6. You know within two weeks whether a new hire is working out. Hold AI to the same standard, and pick the number first.
-
Compare the Cost of the Work, Not the Cost of the Tool
AI, without the jargon: Day 5. Everyone asks whether the subscription is worth it. The better question is what the work costs you now.
-
Integration First, Features Second
AI, without the jargon: Day 4. The best tool is the one that shows up where your team already works. Power means nothing if it needs a new login.
-
Do I Need to Hire an AI Expert?
AI, without the jargon: Day 3. Before you hire anyone, understand the one thing no expert can give you: your own expertise.
-
Start With the Pain, Not the Shiny Tool
AI, without the jargon: Day 2. The mistake almost everyone makes is picking the tool first. Start from where it actually hurts.
-
Count Your Repetitions
AI, without the jargon: Day 1. The simplest way to find where AI can help your business: notice what you do over and over.
-
The Questions People Actually Ask About AI
AI, without the jargon: the introduction. A phone call at a splash pad convinced me to spend 21 days answering one plain-language AI question a day.
-
Shadow AI Is Already Inside Your Organization
Unsanctioned agents are running in most companies right now. The instinct to lock everything down is the wrong one. Proportional governance beats blanket bans.
-
The Judgment Layer Is What You Build
When AI makes the doing cheap, the bottleneck moves to judgment. That layer is the human contribution, and it is the thing worth building deliberately.
-
Pilot-Rich, Transformation-Poor
Most organizations have plenty of AI pilots and almost no AI transformation. Three failure modes explain the gap, and none of them are about the technology.