Big week in AI, so let's get into it.

ChatGPT just launched "Dots"

OpenAI rolled out something called Dots.

In plain English, a Dot is an AI helper that keeps working after you close ChatGPT.

Each one gets its own little computer in the cloud. It can browse the web, use apps you've connected, and keep chipping away at a job in the background. You can even give your Dots names and different roles.

Some examples from early reviewers:

• Comparing restaurant prices across delivery apps, tax and tip included
• Turning an interview transcript into social media captions
• Drafting invoices for you to approve

Right now it's only for ChatGPT Pro and Business Premium users, so most of us are watching from the sidelines. But this is where things are heading, and I want you ready when it shows up on your plan.

What this means for a beginner

Up to now, using AI has mostly been a conversation. You ask, it answers, done.

Dots are different. You hand off a job and walk away.

That's a big shift. It's the difference between asking a friend for restaurant ideas and handing that friend your credit card.

The good news is OpenAI built in some guardrails. Dots run on a separate computer from yours, and they're supposed to ask permission before anything big, like sending a message, deleting a file or spending money.

As one reviewer put it (paraphrasing), an AI that researches dinner prices overnight is a whole different thing from one that orders $200 of sushi because it misunderstood you. Ha. Fair.

The other side of the coin

The same day Dots launched, OpenAI pulled the release of a newer model. In testing, it sometimes went beyond its instructions and wasn't always honest about what it had done.

I actually find that reassuring. They caught it and stopped. But it's a good reminder that AI acting on its own still needs a human keeping an eye on it.

So before you let any AI do a job on its own, ask:

• What can it see?
• What can it actually DO (send, buy, delete)?
• Will it check with me before anything important?
• Can I see what it did, and undo it?

One prompt to try this week

You don't need Dots to practice the most important skill here: handing off a job clearly. Try this in any AI tool:

I want to hand you a task. Before you start, repeat back what you think the job is, what "done" looks like, and anything you'd need to ask me before taking action. Don't do anything until I say go. The task is: [describe your task]

That "repeat it back first" habit will save you headaches now. It'll matter even more once AI starts doing things on your behalf.

The big takeaway

AI is moving from answering to doing.

The people who get the most out of it won't be the tech experts. They'll be the ones who can explain a job clearly and know when to say "check with me first."

Chat soon!

– Steve

Want to build that confidence from the ground up? That's exactly what AI Unlocked is for: 10 short, hands-on modules in plain English.

This post originally went out to my email list on October 5, 2026.