THE AI EDGE · WEEKLY EDITION
The AI Edge, the weekly edition from Hub365 AI. Friday, August 28, 2026 · 5 min read · by Naty Ross.
This isn't legal advice. It's a heads-up that these rules exist; the link to the official text of each one is inside this week's blogs, so you can review it with your attorney or your licensing board. Where we name a rule, it's Florida's.
The read of the week
Five pieces and one argument underneath all of them: what your website is allowed to say about you. Here's what isn't in any of the five blogs: those five pieces are three layers plus two warnings, and they go in ORDER.
- Delete. First out is whatever your rule doesn't let you say. It isn't the fun layer and it's the only one that can cost you a letter from your board. The superlative about yourself, the bare price with none of the variables that move it, and if you're a Florida attorney, the testimonial you drafted yourself. It comes out before you write anything.
- Write what's verifiable. Only then. An AI engine can't repeat an adjective about yourself, because it isn't a data point: there's nothing to repeat. It repeats facts. Year, service by its actual name, city, languages, what happens on a first visit. What you deleted leaves the exact hole all of that goes into.
- Ask who wrote it. And what they have access to. Because somebody wrote the two layers above, and that somebody almost always holds the keys to your Google profile, sometimes your email, and increasingly some AI tool connected to one of those.
Most people start at 2 because 2 feels productive. And they end up publishing, beautifully written, a sentence that shouldn't have been there.
And two things that aren't layers, they're warnings. The first is measurement. One search on ChatGPT tells you nothing: repeating the same question in Gemini returned the same sources only around 40% of the time, and the same business in first place around 7%. If you're going to decide anything based on what an engine answers, make it three questions, on three different days, across two engines, signed out.
The second is reading properly what other people write about you: when writing a review is free, several of them saying the same thing no longer proves anything.
Start where you're weakest, not where you're comfortable.
The play of the week (2 steps, 5 minutes)
Not a diagnosis and out. Here's the fix. Do it in English and in Spanish.
You come out with two things: what has to go today, and the verifiable sentence that takes its place. That's exactly what an AI engine reads when somebody asks about you.
This week on the blog
All the detail, in one place:
-
Monday: One search tells you nothing. That's how almost everyone is checking. - An analysis of 14,472 AI citations found that repeating the same question in Gemini returns the same sources only 40% of the time. Inside, the three-question protocol across two engines, and the two-minute rule for the day an AI asks to see a window of your screen.
-
Tuesday: Two reviews backing each other up can be the same person. Or nobody. - Consensus stopped being proof: when writing a review is free, several of them saying the same thing no longer means it's true. Inside, the coordinates filter for reading what people say about you, and the two lines you don't cross when you ask for reviews in healthcare or legal.
-
Wednesday: What your website can't say in Florida - Five regulated professions and the map of which rule reaches each one, with sentences almost everybody publishes without knowing their board restricts them. Inside, the rule for each one with a link to the official text, and the typical sentences that touch it.
-
Thursday: How an accounting firm in Fort Lauderdale gets new clients - The three answers that get an AI to recommend you, filled in for an accounting firm. With the rule almost no CPA knows exists, and that changes how you can write your fees and the word "specialist."
-
Friday: The five questions your marketing agency should be able to answer in 60 seconds - The standard marketing playbook is written for businesses with no licensing board above them. Five questions to find out whether whoever runs your marketing knows which rules you're under. And all five answered by us, in public.
The tool of the week
Meta AI for Mac now lets you share a window during a session via screen capture, so the AI can see what you're working on and answer with that context. It's genuinely useful. And there's one line worth reading twice: Meta uses interactions with its AI features to train its models. If you work with other people's information, apply the clean-window rule from Monday's blog before you share anything. It takes two minutes and saves you the headache.
Thank you for reading. The owners who go far are not the ones who know it all, they are the ones who keep learning when everyone else stops. That is why we do this, and what we do for our clients we treat as if it were our own business. We do not create just to create. We create to make an impact.
