Issue #9 | September 3, 2026

A couple of weeks ago, I discovered an AI tool that I think may prove to be a breakthrough equal to—or possibly more important than—the smartphone.

That's a pretty big statement coming from me.

Thinking back, I remember a time when I didn't know how to attach a PDF file to an email and grumbled about having to learn how to do it. Fast forward twenty-plus years, and today we're witnessing the beginning of another technological revolution.

This discovery actually started with an ordinary business problem.

I was chatting with Claude, trying to figure out how to eliminate the need for me to manually pull data from one application and connect it to our database without adding another piece of software like Zapier.

I asked a few questions and suddenly Claude responded with something along the lines of: Claude for Chrome is currently in beta testing and might be able to solve your problem.

Wait a minute.

Claude could actually operate Chrome?

Not tell me where to click. Not give me a twelve-step set of instructions.

Actually operate it.

I Decided to Test It

I was a little concerned about using Claude for Chrome because of the warnings. In essence, you're turning AI loose to manipulate your computer on your behalf. I've been using it for about two weeks now, and no disasters yet…

What could possibly go wrong?

The first real test was Meta Ads. Anyone who has spent much time inside Meta's advertising platform knows how incredibly complicated it can be. It's complicated enough that there are courses and certifications devoted to learning how to navigate it.

Normally, figuring out a new system inside Meta means research, trial and error, several wrong turns and probably dragging another member of my team into the problem.

This time I simply opened a tab logged into Meta Ads, opened Claude for Chrome and explained what I wanted to accomplish.

Then I watched it take over.

Claude began navigating through Meta, opening menus, finding settings and moving through screens while I watched.

What could easily have turned into a two-day project for my team was set up, running and tested in about two hours.

That got my attention.

But one successful experiment doesn't prove much. So I gave it another problem.

Then I Tried Google

Next, we needed to work on our Google Business Profile.

Claude navigated the account, found old photos and posts that needed to be removed and helped deal with violations associated with the profile. It even wrote the email and sent it to Google's support team.

If Meta is the most difficult platform on the internet to navigate, Google might be a close second. And once again, Claude handled much of the navigation for me.

That's when this stopped feeling like another interesting AI tool.

I'm not exaggerating when I say that using Claude for Chrome has at least doubled our team's productivity in certain areas. Tasks that used to take two or three people multiple days, we're now finishing in hours.

And that's when I realized I may have been thinking about AI backwards.

The Human Bottleneck

For the past few years, we've been asking how intelligent AI will become. How much better will it get? What jobs will it replace? What will ChatGPT, Claude, Gemini and whatever comes next be capable of doing?

Those are important questions. But I'm beginning to think there's another question we should be asking:

What happens when we're the slow part?

Think about how we've used computers for the past several decades.

Human → Software → Result

The computer might have been incredibly powerful, but the human still had to operate it.

We had to learn where the buttons were.

We had to learn the menus.

We had to understand spreadsheets, databases, CRMs, websites, advertising platforms and dozens of other pieces of software. The machine could only move as quickly as we could operate it.

AI is beginning to change that relationship.

Increasingly, we're moving toward:

Human → AI → Software → Result

Instead of learning every button inside a complicated piece of software, we can tell AI what we're trying to accomplish and allow it to operate the software.

And that creates an interesting possibility.

The limiting factor may no longer be how quickly the computer can do the work. It may become how quickly the human can decide what needs to be done.

That's The Human Bottleneck.

The Takeaway

How this all ends, I don't know.

What does business look like in two years? Education in five? Our personal lives in ten?

I don't pretend to know.

I've used ChatGPT, Gemini and Claude, and each has different strengths and weaknesses. Sometimes I'll brainstorm with one, cross-check the answer with another and use a third to attack the same problem from a different direction.

But as a reluctant early adopter of technology, the most important lesson I've learned isn't which AI is currently the best.

It's that we have to adapt.

Humans didn't get this far because we were the fastest or strongest species. We adapted to the environment around us. We learned to use tools. We cooperated. And when our environment changed, eventually we changed with it.

I suspect we're about to be tested on that ability again. Only this time, the environment may be changing faster than we are. You don't have to predict exactly what AI will look like ten years from now to begin adapting.

Here's a simple experiment.

Think about one task you repeatedly do that frustrates you, takes too long or requires you to navigate software you hate.

Then instead of asking AI how to do it, ask a different question:

Can AI actually do it with me—or for me?

You may be surprised by the answer.

Intelligence Brief

Why Rates Are Climbing Again

Mortgage rates have had a rough week. The 30-year fixed has pushed up to around 6.74%–6.82% nationally as of this morning—up roughly a full point from the 5.75% many buyers were seeing back in March.

Several forces are contributing to the pressure, including geopolitical uncertainty, energy prices, persistent inflation concerns and expectations surrounding future Federal Reserve policy.

Houston's housing market, meanwhile, continues to give buyers something the national headlines sometimes overlook: inventory and negotiating opportunities.

Insider takeaway: In a softer market, don't look only at the asking price or wait endlessly for mortgage rates to fall. Ask your agent and lender to explore seller concessions and rate-buydown strategies. Depending on the transaction, improving the financing can sometimes be more valuable than negotiating another small reduction in price.