The Sunday Work Reset One AI decision. One useful resource. One clearer workweek.

The free lesson I teased last week has a date. Design ChatGPT 5.6 Skills, Not Prompts, live on Zoom, Wednesday July 29 at noon Eastern. Details at the bottom. This issue is the argument behind it.

WHAT CHANGED?

There's a word going around for the flood of mediocre AI output filling up inboxes, feeds, and shared drives. AI slop. Generic emails. Reports that say nothing in more words. Proposals that read like every other proposal. Most people blame the model. The model is not the problem.

Slop is what you get when you hand AI a prompt instead of a skill.

A prompt is a single move. You type it, you get one result, and the quality depends entirely on how well you framed that one moment. A skill is different. A skill knows the sources to pull from, the steps to follow, the boundaries not to cross, and what a good result actually looks like. Run a prompt and you get an answer. Run a skill and you get your answer, the way you would have done it, every time.

Here is where most people are stuck. They have a folder of prompts. Each one worked once, so it got saved. And every week they open the folder, paste one in, and it does not do what it did the first time. So they go find a better prompt. The folder gets longer. The work does not get easier.

That folder feels like progress because it keeps filling up. It is not progress. It is hoarding. You are collecting moves without learning the game.

Think about the last follow-up email you asked AI to write. You had a prompt that worked for one client on one deal. This week the client is different, the deal is further along, the tone needs to shift. The old prompt returns something generic, so you rewrite it twice and lose twenty minutes. You did not need a better prompt. You needed the thing underneath it: where the deal stands, what the last conversation was, what you want to happen next. That is the skill. With it, you can write the prompt fresh in ten seconds for anyone. Without it, you are back in the folder next week.

This is the line between people who save real time with AI and people who just stay busy with it. Prompts are disposable. Skills compound. And slop is simply what a prompt produces when there is no skill behind it.

Stop feeding the folder. Build the skill it was standing in for.

What this changes:

Stop judging AI by the prompt you typed. Start judging it by the skill you gave it. Slop is a skill problem, not a model problem.

Who should pay attention:

Anyone sitting on a prompt folder they keep adding to and rarely reuse. Any leader whose team shares prompts but still redoes the work by hand.

What to do this week:

Pick one prompt you saved and actually reused. Write down why it worked, the judgment behind it, not the wording. That judgment is the skill. The words were one version of it.

The Prompt-to-Skill Test

Three questions. Run any saved prompt through it. It tells you whether you built a skill or got lucky once. Five minutes.

Start with a prompt you actually reused, not one you saved and forgot. The reused one is the honest test because you already trusted it with real work. If it holds up here, you have something worth keeping. If it does not, you just found the shortcut that has been costing you time without you noticing.

1. Can I rewrite this from scratch and get the same quality?

Close the folder. Open a blank window. Write the prompt again without looking.

If the new version gets you the same result, you own the skill and the saved prompt is just one expression of it. If the new version comes back weaker, you memorized words you do not understand, and they will break the moment the client, the task, or the model changes.

2. Can I name why it worked, not just that it worked?

Say the reason out loud in one sentence. Not "it gave me a good email." The reason. What did the prompt tell the AI that made the output land?

A prompt that works for a reason you can state is a skill you can repeat. A prompt that works for reasons you cannot explain is a coincidence you are hoping happens again. Coincidences do not scale.

3. Could I teach this to my team without handing them the prompt?

Picture the conversation. You are walking a team member through this task. You cannot copy and paste the prompt into the chat. You have to explain what you do and why.

If all you can pass along is the words, you have a prompt. If you can pass along the judgment, you have a skill. Skills survive when the tool updates and the old prompt stops working. Prompts do not.

How to score it:

Three yeses means you have a real skill under the prompt. Write the skill down. The judgment is the asset, not the text.

One or two noes means you have a prompt you got lucky with. That is not a reason to throw it out. It is a reason to slow down, figure out what actually made it work, and turn that into something you can rely on.

Three noes means the prompt worked once by accident and you have been paying that accident forward every time you paste it in.

How to run it:

Take the prompt you picked. Open a notebook or a doc. Write the three questions across the top. Answer each one in one or two sentences, honestly. If you cannot answer a question, that is the answer. Do not skip the ones that feel obvious. The obvious ones are usually where the judgment is hiding.

The live lesson on July 29 picks up where this test leaves off. We take one of your real tasks and build the full Skill around it, sources, steps, boundaries, and review, using a blueprint you can reuse for every task after.

Last week I said I was hosting a free live ChatGPT lesson. Here it is.

Design ChatGPT 5.6 Skills, Not Prompts 

Wednesday, July 29, 12:00 PM EDT One hour, live on Zoom. Free.

In an hour you will learn to find your way around the new ChatGPT 5.6 setup and know when to use Chat, Projects, Work, Skills, Sites, or Codex, design one reusable Skill with trusted sources, repeatable steps, clear boundaries, and a review step, and map one of your own tasks into a Skill you can run again.

Sign up here.

How I help:

AI Advisory. Executive alignment, AI roadmaps, and decision frameworks for leaders navigating AI.

AI Governance. Safe-use policies, risk review, and governance playbooks built for organizations where the cost of getting it wrong is real.

AI Training & Adoption. Role-based training and playbooks that hold after the rollout ends.

AI Speaking. Signature keynotes (The Operator's Edge, The Hidden Hour), executive briefings, and workshops.

Reach me: keenallianceconsulting.com set a time or reply to this email.

See you next week!

Andrew Keener 

Founder, Keen Alliance Consulting 

Follow Andrew Keener for where AI belongs, and where it doesn't.