Bradley here.

I've got a simple rule for AI content in 2026: if the person who actually does the work never contributed anything, you're probably publishing garbage.

I've watched agencies crank out hundreds of "SEO-optimized" articles that sound perfectly reasonable until you realize nobody at the company has ever said any of it. It's just recycled information wrapped around a keyword.

That's a problem because experience is becoming one of the hardest things to manufacture. AI can write about a roofing problem. It can't tell you what a roofer actually sees when he climbs onto a 20-year-old roof in August and finds three different problems the homeowner didn't know existed.

That's where the opportunity is.

Instead of asking AI, "Write me a 2,000-word article about roof repair," use it as an interviewer.

Have it interrogate your client's estimator, technician, attorney, doctor, salesperson, owner, or whoever actually knows the business.

Then turn those answers into content.

Now you've got something competitors can't simply copy from the same five websites.

IN THIS EDITION

QUICK HITS

THE DEEP DIVE

Here's what most agencies are doing right now.

They get a keyword list.

They give it to ChatGPT or another LLM.

They say something like:

Then they edit the output, add some internal links, throw in a few keywords, publish it, and wonder why it looks exactly like every other plumbing article on the internet.

Because it is.

The model has no firsthand knowledge of that plumbing company. It doesn't know which emergencies they see most often. It doesn't know what homeowners misunderstand. It doesn't know which repairs technicians wish customers would stop attempting themselves.

So it fills the gaps with statistically plausible information.

That's how you end up with 500 articles that are technically correct and completely forgettable.

The old content model treated expertise like something you could fake with enough words.

That doesn't work nearly as well anymore.

I've said this before: AI can fake information all day long, but it cannot create a verifiable history of real-world experience. That's why I care so much about entity relationships and the digital footprint surrounding a business.

Think about a local HVAC company.

The owner has probably answered thousands of questions.

That is an absolute gold mine.

And most agencies completely ignore it.

This is the part I want you to steal.

Don't ask AI to invent expertise.

Ask AI to extract expertise.

Start with one subject-matter expert inside the client's business.

Give the AI some basic context about the person and their role.

Then have it conduct an interview.

For example:

"You are interviewing an experienced roofing estimator. Your job is to uncover firsthand knowledge that can be turned into useful content. Ask one question at a time. Prioritize real projects, unusual problems, customer misconceptions, decision-making criteria, geographic factors, costs, mistakes, and lessons learned. Do not accept generic answers. Ask follow-up questions whenever the answer could be more specific."

Now the conversation gets interesting.

Those answers contain something a generic AI article doesn't have.

Experience.

This is where agencies screw it up.

They interview the expert once, get 30 minutes of answers, then tell AI:

"Turn this into a blog post."

Don't do that.

The interview is your raw material.

Have AI identify:

That gives you a content map based on the business itself.

This fits perfectly with the customer-data approach I've talked about before. Real customer conversations contain the messy language, objections, symptoms, and phrases that traditional keyword research misses.

Now you can use AI for what it's actually damn good at.

But the source material comes from the expert.

That's the distinction.

I'd structure the article around the actual questions the expert answered, with direct answers near the top and deeper explanations underneath.

If the expert tells you about a specific situation they encountered, keep that specificity.

If they explain a mistake customers repeatedly make, use it.

If they mention a particular neighborhood, building type, weather condition, equipment model, or customer concern, don't strip it out just because some generic SEO template didn't include it.

That detail is the value.

This gets even more useful for local agencies.

A local business isn't just associated with a service.

It's associated with what it does, who it serves, and where it operates.

That's the model I've been pushing for a while now: build associations between the brand, its services, and its geography using signals that actually make sense together.

So if the expert talks about working on older homes in a particular part of town, that's useful.

If a doctor explains a common condition they see in their patient population, that's useful.

If a lawyer explains the specific questions clients in their jurisdiction keep asking, that's useful.

The location isn't being stuffed into the article.

It's part of the story.

That's a hell of a lot more useful than replacing "Dallas" with "Plano" fifty times.

Pick your five best clients.

For each one, identify one person who actually knows the business.

Then run a 30-minute AI-assisted interview.

I'd specifically ask about:

Then turn those answers into a content backlog.

You don't need 100 AI articles.

You need a body of content that actually sounds like somebody who has done the work.

Here's where this gets interesting for agency owners.

You're creating an expertise extraction system.

The expert doesn't need to sit down and write a 2,000-word blog post.

They talk.

AI interviews them.

AI organizes the information.

Your content team edits it.

Your SEO team structures it.

Now the agency has a repeatable process for capturing knowledge that previously lived inside one employee's head.

That is a much better use of AI than replacing the expert entirely.

The same principle applies off-site, too. Once the business has a clear body of expert knowledge, you can reinforce those same entities and associations across the wider web instead of publishing disconnected content everywhere.

That's how you start building a real entity.

Not by pretending the business has expertise.

By documenting the expertise it already has.

PRODUCT & SERVICE SPOTLIGHT

This week I'm highlighting Niche Relevant Blog Links.

These are links placed inside content that actually matches the client's vertical, which helps reinforce the topical association around the business.

Let's say you've spent the last three months interviewing a client's experts and building genuinely useful content around their knowledge.

Don't then surround that content with completely unrelated backlinks because some tool told you the domains have a high DR.

That defeats the point.

The source needs to make sense.

If you're building Topical Authority for a medical client, I want the surrounding signals to make sense in a medical context. Same thing for legal, home services, or whatever niche you're working in.

The goal is Defensible Relevance.

I use Niche Relevant Blog Links when a client needs stronger vertical-specific authority signals or when their existing backlink profile contains too many generic placements.

The whole point is to reinforce the entity relationships you're already building on the site.

This is particularly useful for agencies managing clients in competitive niches where generic link acquisition has become a commodity.

If you're doing the expert-led content strategy I outlined above, this is the kind of off-page relevance I'd want supporting it.

AWESOME APPS

Professional headshots shouldn't require scheduling a photographer every time someone joins your team. AmberFace AI generates polished business portraits from a single selfie in just a few minutes.

Here's why it's worth keeping handy:

The Agency Lever: Every agency eventually needs updated team photos. AmberFace lets you refresh your brand without coordinating expensive photo sessions every time someone joins the company.

Check out AmberFace AI here: https://amberfaceai.app/

WINS OF THE WEEK

The biggest win I'm seeing is that agencies are finally getting more comfortable using AI as part of the research and production process without handing the entire job over to the machine.

That's the right direction.

AI is damn good at processing huge amounts of information. The mistake is asking it to manufacture the information in the first place.

We've also seen a clear move toward customer-generated language as a content input. In one example from our own work, a multi-location medical client saw phone calls increase 27% after real patient transcript questions were mapped into dedicated answer blocks on regional pages.

One of our agency partners, David S, summed up his experience with Semantic Links after ordering a Local SEO Audit:

He specifically called out the actionable roadmap covering traditional search, Maps, and AI platforms. That's the kind of feedback I like because it isn't about some vanity metric. It's about getting a clear understanding of what needs to happen next.

THE HIDDEN WIN

Here's the opportunity most agencies are overlooking:

Interview the employees nobody thinks to interview.

Everyone wants the CEO.

Fine. But the receptionist knows what people ask before booking.

Those people are sitting on an enormous pile of first-party content data.

Interview them.

You'll probably uncover better content ideas in 30 minutes than you would in three hours inside an SEO tool.

Good luck copying 15 years of somebody else's real-world experience.

That's the advantage.

Build better links,

Bradley Benner
Founder, Semantic Links