Big shift this week... AI is getting a lot less "chat window" and a lot more "here, let me build you something."

Let me break it down in plain English.

ChatGPT can now build a tool around your question

On October 7, OpenAI launched GPT-6 with something they're calling "Intelligent UI." Depending on what you ask, ChatGPT can answer with a diagram, a calculator, a chart, a form, or a comparison tool instead of a big block of text. It's rolling out on free and paid plans (the model behind it varies by plan). (Source: OpenAI Help Center)

What this means for you: you don't have to hunt for the perfect app anymore. Just say "build me a calculator that shows how much I need to save each month" and... it does. Pretty cool. (I've been building dashboards in Claude for the past week and it's been so fun.)

Honestly? Ignore the benchmark scores everyone's yelling about. This is the part that matters for us regular folks.

Your voice memos just got a job

As of October 6, paid ChatGPT users can upload audio files and get a transcript, summary, meeting recap, or follow-up email. You can even ask questions about what was said. (Source: OpenAI Help Center)

What this means for you: Record → upload → organize → review.

That rambling voice memo from your walk? Now it's an outline. That meeting? Now it's a checklist.

One heads-up: it can still get names, dates and promises wrong. Always double-check the important stuff against the actual recording.

Free Gemini got a little less free

Starting October 9, Google limited free Gemini users to its Flash Lite model. The beefier models now need a paid plan. (Source: The Verge)

What this means for you: the free stuff is going to keep changing. That's why I don't teach "memorize this one app." I teach a workflow that works anywhere:

• Explain your goal clearly
• Give it some context
• Ask questions
• Review what you get
• Try another tool if the first one can't handle it

And no, "free AI is over" isn't true. Plenty of good free options are still out there.

AI agents might have to tell on themselves

On October 6, OpenAI and Anthropic told Australian lawmakers they'd support mandatory reporting when AI agents cause data breaches. Right now that's mostly voluntary. (Source: Reuters)

What this means for you: as AI starts clicking around websites and your accounts for you, you deserve to know what it's up to. Before you connect any AI to your stuff, ask:

• What can it see?
• What can it do?
• Will it ask me first?
• Can I disconnect it?

More automation should mean more visibility, not less.

One prompt to try this week

Pick something that's been confusing you and paste this in:

Help me understand [topic], but don't give me a long block of text. Create a simple, interactive explanation with one visual example, a short step-by-step walkthrough and three questions to check my understanding. Use plain English and let me ask questions as we go.

Holy moly, it's a totally different experience than reading a wall of text. Try it on something you've been putting off learning.

The big takeaway

Don't just ask AI for an answer.

Ask it to build the tool you need.

Chat soon!

– Steve

Want to get comfortable doing this stuff every day, without the overwhelm? 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 12, 2026.