First Domino Marketing

Don't know about you, but I use ChatGPT for proofing copy. The problem? It gives a long list of edits, which can be overwhelming. Here's a quick solution to try if you're suffering from advice overload...

1️⃣ Do it in any chat
Paste this line first:
Work step-by-step. Give me one edit or one troubleshooting step at a time. Stop and wait for ‘done’ or ‘next’.
Paste your copy or describe the issue.
Reply “next” after each change.

2️⃣ Make it the default
ChatGPT Settings
Custom Instructions
In “How would you like ChatGPT to respond?” paste:
Work step-by-step. One edit/step at a time. Stop and wait for ‘done’ or ‘next’.
Save

Let me know how you get on...

No secret - I really like ChatGPT and have created projects for almost everything. But...there's a catch. Once you go past 12 projects or so, the system struggles to consistently show them in the menu.

I kept losing work, and as for the search function? It's no Google...

This post shows the simple fix that keeps every project visible and every system findable. All without adding new tools or complexity.

Let me know what you think...and share with your community if you think it would help others.

Confession time. My ChatGPT setup was messy. Too many projects. Files buried in chats, assumptions about what it 'understood/remembered', and having to hunt around to find my latest work.

So, I did some digging. Fixed it with one pattern and one habit.

Swipe through the carousel and steal the setup. Clear names, promote reusable files, fewer projects, cleaner handoffs. Try it on your next upload. If it saves you time, like, comment, and share...

Harvard Business Review’s new “workslop” piece hits home https://lnkd.in/gWBKa_XN. Their researchers found a growing (and familiar) pattern: AI outputs that appear polished but require someone else to fix, clarify, or redo the work.

My take: if you blindly treat your AI model’s output as “done” and pass it on, you’re forwarding algorithmic guesses, not adding value.

In addition to wasting your colleagues' time, I see two major developmental risks to worksloppers...

1. No learning. Copy and paste creates activity without insight. There's no context built. So you're not forming memories.

2. Skill atrophy. Offloading analysis, synthesis, and evaluation erodes the exact muscles you need to develop.

AI should accelerate thinking, not replace it. Avoid hashtag . Build judgment through experience.

P.S. Yes, I get the irony of using an AI image to illustrate this post. But at least it's the fourth generation of the illustration.

Like a lot of people, I didn't really like ChatGPT version 5.0's dynamic model selection. I missed the ability to switch when I wanted to, and welcomed the change back.

But sometimes I let it make the decision to use the 'right' model without switching… which is kinda true, except for agent mode (where ChatGPT orchestrates tools, even opening a browser).

The real choice now is knowing when to use Agent Mode versus Web Search inside ChatGPT. And knowing that Chat can't change to Agent Mode by itself.

👉 Don’t assume ChatGPT will switch for you. Swipe to see when to choose Web Search and when to turn on Agent Mode.

A Semrush study (https://lnkd.in/gfk24UGp) making the rounds says it all: Reddit, Wikipedia, and YouTube dominate as the top sources cited by AI.

But here’s the risk: taking that at face value and assuming “post more on Reddit and you’ll be fine in AI search.”

The reality is more complex. AI visibility isn’t about a single platform. It’s shaped by three layers: training data, citation patterns, and retrieval systems.
That’s why I built this carousel...to show how those layers work together, and what practical steps marketers should take across each.

TL;DR: If you’re still optimizing for Google search alone, you’re already behind.

What happens when you run a car lease negotiation through AI instead of playing the dealer’s game?

In my case: $1,500 saved and free upgrades. By keeping everything in email, refusing to walk into a showroom without a worksheet, and using ChatGPT to structure the numbers and draft replies, I flipped the usual power imbalance.

The end result? the exact car I wanted, savings locked in, and extras thrown in...

Why I’ll Never Negotiate a Car Lease Without ChatGPT Again The reason I hate working with car dealerships is simple: I feel like they’re trying to scam me. The power imbalance makes the process unfair.

I kept running into a ChatGPT memory problem in my work: I’d drop a file into a chat, have a great exchange, and then, the next time, it would forget what we were talking about. So, I dug a little deeper into how “knowledge” actually works inside ChatGPT. Deep breath...

1️⃣ Chats hold context only for that thread.
2️⃣ Project files only persist if you add them at the project level (not inside a chat) and only within that project. THIS IS IMPORTANT!
3️⃣ Custom GPTs carry training across everything, but you must retrain them to reflect updates.

Read on, and let me know how you've been managing knowledge in ChatGPT. And do me a favor (or, favour if you speak the King's), and share this with your connections if you think they'd like it.

I increasingly believe it's a ChatGPT world (but we just live in it). But have one big frustration that I call 'The Goldfish Problem'. ChatGPT has limited memory, and forgets what you tell it.

But there's a fix. It's a balance between 'Knowledge' (the docs you load into a custom GPT), and 'Memory' (what you tell ChatGPT to remember about you).

Take a look at my carousel 👇. And let me know how you're using knowledge and memory in your AI day-to-day.

P.S. No goldfish were harmed in the creation of this post.