How the Top 1% of Marketers Are Using AI Right Now

A while back, the executive in charge of HubSpot's entire AI product line said something that shook me.

He said: a year ago, when he first took over this division, he went around telling everyone that AI was "cool, but not essential." Now? Essential.

Not "might become essential someday." Essential already.

I asked — that fast?

He said, fast. Way faster than you think. Because AI's capabilities have already outpaced the level at which most people actually use it.

After that conversation, I went back and thought about it for a long time. Then I decided to take the truly valuable insights from that conversation and break them down for you, one by one.

The First Word Is "Context"

What does context mean?

You join a company, and they spend 6 to 12 weeks training you. They teach you how the company does things, how to sell the product, how to communicate with customers, what the brand voice sounds like.

That body of knowledge — that's context.

Now here's the problem: you hire an AI assistant, and it knows nothing. You ask it something, and it gives you an answer that's "correct but useless." You're disappointed and think AI is nothing special.

But have you ever wondered why it responds that way?

Because it has no context.

You spend 12 weeks training a new hire, yet you expect an AI with zero background to deliver professional-level work right out of the gate. Is that reasonable?

That's the realization HubSpot had when building their Breeze assistant: for an AI assistant to be truly useful, it first has to "know" you. It needs to know your company, your data, your brand style guide, your marketing playbook. Not by making you re-explain everything in every conversation — but by digging through your CRM, your documents, your PDFs on its own.

Let me tell you — this isn't some distant future.

This is a piece of the puzzle that every serious AI marketing tool will fill in over the next year.

Image and Video Models Suddenly "Actually Work"

Have you ever tried handing AI a photo and asking it to make changes?

Here's a fun story. Someone posted online that they had a photo with an old movie billboard in it — could the text be changed to a wedding date? A bunch of people started grinding away in Photoshop.

Then someone dropped the image into Google's Nano Banana. Thirteen seconds. Done.

Thirteen seconds.

They then tossed the image into Google Veo, Google's video model, and typed just two words: "surprise me." And the person in the video actually moved, smiled, stood up.

Where's the real impact in that?

It's not that "AI can edit images." It's that you suddenly realize everything that used to require hiring a photographer, a designer, an editor — you can now do all of it yourself, in a single afternoon.

If you're in marketing, what does this mean for you?

Think about it. Super Bowl ads — they could be made by AI in the future. Your company's product videos could be customized into 100 different versions for 100 target clients, each one deeply researched and tailored so precisely it feels bespoke.

Could you do that before? No. Can you do it now? Yes.

Because production costs just collapsed.

The "Customer Service" on Your Website Is Becoming a Marketing Weapon

Here's something counterintuitive.

HubSpot first built their website chat feature as part of their sales product. The logic was simple: a customer shows up, wants to chat, that means they want to buy.

The result? Dead wrong.

Nobody places an order directly in a chat window. Later they pivoted, repositioning chat as customer service — answering post-sale questions, handling complaints.

But guess what? What users kept asking about was never "my thing is broken." It was "how do I use this," "what's the difference between this and that," "why is this so expensive?"

These are pre-sale questions. These are marketing questions.

Even more interesting — one customer told HubSpot something that really hit home for me:

I genuinely believe that if my website can't answer these questions, customers will go to the competition to find the answers.

Let that sink in.

Traffic is getting harder and harder to come by. Every visitor is hard-won. They want to learn a little more, but your website has no answers, and they're afraid of getting a sales pitch if they ask customer service. So do they leave?

Gone.

So now, a group of smart marketers has started doing something: transforming the AI chat on their websites from "processing support tickets" into "answering pre-sale questions, identifying potential customers, and even automatically scheduling appointments."

And here's the key:

There's no such thing as a stupid question when you're asking AI.

You wouldn't dare ask a sales rep "what does this acronym mean" — afraid of looking unprofessional. You wouldn't dare ask "why is this so expensive" — afraid of getting pitched. But you'll ask AI. You can talk to it for eight minutes, ten minutes — nobody rushes you, nobody pressures you.

HubSpot ran their own experiment: the average conversation with a human sales rep lasted less than four minutes. With AI, it was eight minutes.

Why?

Because human sales reps are chasing quotas. "That's a great question! Shall we set up a meeting?" That's pressure.

AI has no quota. It just quietly stays with you and chats.

Agents Handing Off to Agents

This next one — I think it's the trend most worth watching over the coming year.

A customer has been chatting with AI on a website for a long time, asking a bunch of questions. Finally, the AI says: "Would you like to schedule a time to talk with one of our experts?"

The customer says: No thanks.

The traditional approach ends right there. The lead is lost.

But there's a new playbook: after the customer says "no thanks," the AI doesn't just close the conversation and move on. It packages up the entire conversation and hands it off to another AI agent.

The second agent takes the conversation, digs deep into researching the company on its own. Then a couple of days later, it sends an email: "I noticed you were exploring this area a while back — we just built something related. Would you like to take a look?"

Agents handing off to agents.

Let me tell you what this means.

It means marketing is eating into sales' territory. All the work sales reps hate doing — scheduling meetings, qualifying leads, following up — AI handles all of it. And sales? They're thrilled to just take the high-quality meetings AI feeds them.

Go tell a sales team today, "I'm going to fill your calendar with qualified meetings" — see if they're not delighted.

The Rules Have Changed. You Make Your Own.

As we were wrapping up, that HubSpot executive said something that I think was the most valuable line of the entire conversation.

He said that for the past fifteen years, there's been a massive "information gap" in the market: not enough information flowing between businesses and customers. So you had to hustle — writing articles, creating content, desperately filling that gap.

Now, large language models have closed that gap once and for all.

But video hasn't.

How many topics are out there where people want to watch a video but can't find a decent one? Too many.

And AI is driving the cost of making videos down at a speed you can't even imagine.

So what happens next? Video will come pouring in like a tide. B2B companies will stop playing it straight with stiff tutorials — they'll start making dramas, thrillers, content you can't look away from. Because entertainment is becoming a new customer acquisition channel.

The Old Rules	The New Rules
Not enough information — write more articles	Enough information — make more videos
B2B does information, B2C does entertainment	Everyone can do entertainment
Marketing and sales are sharply divided	Marketing is eating into sales' territory
High production costs, scarce content	Production costs collapsed, creativity is king

Do you see it now?

Technology has caught up with the hype. Those "six-fingered monsters," those obviously AI-generated crude images — they're in the past. The image models today are better than the photographer you're paying for.

What's left is up to you.

Your creativity, your ambition, whether you're willing to pick up these tools and practice like a beginner all over again.

He mentioned his own kid — a teenager who wanted to start a soccer training camp. The kid used free AI models to make materials, and the results were a mess — nearly had a breakdown. He switched them to the paid version, and within a couple of tries they were producing beautiful work, posted it online, and the camp was up and running.

Where's the gap?

It's not in the AI. It's in whether someone guides you to build those muscles.

These days, what limits you isn't the tools. It's you.

So stop waiting. Take the best AI tools available today, and practice. Experiment. Take every "that's impossible" thought in your head and overturn them, one by one.

That teenager's training camp is already up and running. Among the kids enrolled, there might be the child of the client you most want to land.

You never know. But the prerequisite is — you have to start practicing first.