I've been paying pretty close attention to what's happening with AI lately, and one shift is getting hard to ignore:

AI is starting to do more than just answer questions.

It's beginning to work across websites and apps, knock out parts of everyday tasks, and help people take on things they may not have the experience (or the time) to handle on their own.

Here are a few updates that caught my attention, and why I think they actually matter.

Gemini can now handle certain web errands

Google added a feature called "auto browse" to Gemini Spark.

With your permission, it can work through websites to research options and start tasks like comparing flights or setting up apartment viewings. When it hits something more sensitive, especially a purchase or payment, it's supposed to hand control back to you.

This is a pretty big change.

Up until now, most of us have used AI by asking a question and getting an answer. Now we're moving toward AI where we give it a goal, set some boundaries, and let it handle parts of the process.

That doesn't mean turning everything over and hoping for the best. Quite the opposite. A sensible approach:

1. Clearly explain what you want.
2. Tell the AI what it can and can't do.
3. Let it handle the low-risk, reversible steps.
4. Review everything before approving anything important.

The tech is getting more capable, but your judgment still matters, especially when passwords, money or personal info are involved.

Talking to AI is getting more useful

Google's latest Gemini update also lets Mac users talk to AI while working in other apps. You can talk through an idea, dictate text, rewrite something you've highlighted, summarize info or ask for an image without bouncing between windows.

This is great news for anyone who feels like they need the "perfect prompt" before using AI.

You really don't.

Sometimes the easiest way in is to just talk through what you're thinking and let AI help you make sense of it. A simple workflow:

Speak your rough idea → ask AI to organize it → check the facts → rewrite it in your own voice.

That works for emails, meeting notes, travel plans, social posts, or pretty much anything you're trying to get out of your head and into usable shape.

AI education isn't just for tech people anymore

Colleges are starting to build AI courses and requirements for students outside computer science. Purdue introduced an AI graduation requirement. Ohio State now has an AI-fluency requirement. Community colleges are building shorter programs around how AI gets used in specific careers.

I find this super encouraging, because I don't think most people need to become AI experts (or learn to code) to benefit from it.

But we're getting to the point where a basic understanding of AI will be like knowing how to use email, search the internet or work a spreadsheet.

The important skills are pretty straightforward:

• Knowing what AI is good at
• Giving it enough context to be useful
• Recognizing where it can go wrong
• Checking important info
• Staying responsible for the final decision

That's really what AI literacy means to me.

AI can help us "borrow" skills we don't have

This was probably the most interesting stat I came across.

OpenAI analyzed more than 800,000 work-related ChatGPT messages. Among messages tied to a particular job, 43.5% involved tasks that would normally belong to someone in a different field.

In plain English: people are using AI to venture outside their usual expertise.

A small-business owner thinking through a marketing plan. A salesperson analyzing data. Someone prepping for a big meeting on a topic they barely know.

This doesn't suddenly make any of us a lawyer, accountant, marketer or software developer. And it definitely doesn't replace a qualified pro when you need one.

But it can help you ask better questions, understand your options, and walk into a conversation way more prepared. That feels like one of the most practical uses of AI out there.

A lot of the rest is still noise

New models, new benchmark scores, new "super apps" that haven't actually launched yet... the announcements never stop.

Some of it will matter eventually. Some of it won't.

For most of us, the better question is:

Does this help me do something useful more easily, reliably or safely?

If not, you probably don't need to worry about it yet.

One prompt to try this week

Think of something you'd like to accomplish but don't have much experience with. Then adapt this:

I'm trying to [describe what you want to accomplish], but I don't have much experience with it. Before giving me a recommendation, ask me five questions about my situation, goals and limitations. Then give me a simple plan, explain any unfamiliar terms in plain English and identify anything that should be reviewed by a qualified professional.

The big takeaway

Being good at AI isn't about memorizing every feature or keeping up with every new model.

It's about knowing how to bring AI into a task, give it the right context, question what it gives you, and use your own judgment at the end.

That's way more useful, and honestly, way less intimidating.

– Steve

Want to get comfortable using AI every day, minus 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 August 3, 2026.