This week is about one line being drawn: whose side is your AI on?

The Super Bowl made the AI trust war mainstream. Anthropic spent millions on four spots mocking the idea of ads inside AI assistants. Hours later, OpenAI officially launched ads in ChatGPT for free and Go-tier users. Sam Altman called the Anthropic ads "clearly dishonest." Scott Galloway called them "a seminal moment." The real story is what Galloway said next: the dominant use case for AI is therapy, not productivity. People reveal their deepest concerns to these tools. Introducing monetisation into that relationship is not a CPM decision, it's a trust architecture decision. Marketers need to watch this closely, because the question of whether your AI works for you or for an advertiser is no longer theoretical. It launched on Monday.

91% of marketers use AI. Only 41% can prove ROI. Jasper's 2026 State of AI in Marketing report surveyed 1,400 marketers and found that ROI confidence actually dropped from 49% last year, not because AI is underperforming, but because expectations jumped. The teams that can prove returns report 2x or greater. The gap: governance, legal review, and brand standards are now the primary blockers to scale, not adoption. The CMO-IC divide is widening too. 61% of CMOs say they can prove ROI; only 12% of individual contributors agree. If your team is using AI without measurement infrastructure, you're generating activity, not evidence.

Oracle launched 16 AI agents for marketing, sales, and service. Built into Oracle Fusion Cloud Applications at no additional cost, these role-based agents cover campaign planning, buying group targeting, audience analysis, deal advisory, and renewals. The important move: Oracle's agents pull data from across the entire enterprise, not just CRM. Finance data, contract data, inventory, supply chain. As Oracle's Katrina Gosek put it, AI hits a ceiling if it's CRM-first only. This is the same pattern as HubSpot's Breeze push, but aimed at global enterprise. The race is to own the context layer, not the model layer.

The IAB says US ad spend will grow 9.5% in 2026, and five of the top six buyer priorities are AI. Two-thirds of buyers are now focused on agentic AI for ad buying and campaign execution. 73% are prioritising content optimised for AI-generated answers. Creator partnerships surged to 57% focus, up from 48% last year, with brands using human storytelling to rise above AI-generated content noise. Meanwhile, customer acquisition dropped 10 points as a priority while driving repeat purchases doubled. The implication: the industry is shifting from reach to retention, and agentic systems are enabling it.

Cowork plugins turned Anthropic from model provider to workflow competitor. Anthropic open-sourced 11 specialised plugins for its Cowork agentic tool, spanning sales, marketing, finance, legal, and more. The legal plugin alone triggered a market panic, with Thomson Reuters stock dropping 16% and Wolters Kluwer falling 10%. The signal for marketers: when a foundation model company can ship a marketing plugin that connects to your CRM, follows your brand playbook, and executes campaign workflows, the competitive threat is no longer another SaaS vendor. It's the model layer eating the application layer.

The pattern: Trust is becoming the differentiator, not capability. The winners will pair agentic execution with clear governance, prove ROI beyond time savings, and build context moats that AI systems can actually use. Everyone has agents now. The question is whether yours work for you, and whether you can prove it.

If AI tools are as powerful as everyone keeps telling us, why aren’t more marketers actually building things with them?

Not prompts.
Not posts.
Not “10 ideas for LinkedIn”.

I mean real things. Functional tools. Live URLs. Something you can send to a client, a team, or your email list.

So I decided to test it properly.

I built a full web application from scratch. It’s called the AI Marketing Roadmap. It’s live. It works. And it was built using Claude Code.

No development team.
No agency.
Just me, Claude, n8n, and a few focused evenings.

This is what I learnt and why it matters.

I spend a lot of time speaking to marketing teams about AI.

The pattern is always the same.

They know AI matters.
They know they should be doing more.
They’ve experimented with a few tools.

But when I ask three simple questions, things get vague:

Silence. Or a shrug.

The data reflects it.

According to the 2025 State of Marketing AI Report, 75% of teams still don’t have a formal AI roadmap for the next one to two years. Salesforce found that 43% of marketers admit they don’t know how to get the most value from generative AI. Marketing Week reports that 80% of CMOs are concerned about the AI skills gap.

Awareness is high.
Execution is low.

That gap is the opportunity.

I didn’t want to write another article telling people AI is important. We’ve got enough of those. I wanted to build something that helps a marketer understand where they are and what to do next.

Yes, commercially, it works as a lead magnet.

But it had to be genuinely useful. Something you would actually forward to your team. Not another PDF that dies quietly in a downloads folder.

Here’s the bigger belief behind this project.

AI tools like Claude Code are not just productivity enhancers. They change who gets to build.

I’m not a developer. No computer science degree. A year ago, the idea of shipping a web app would have felt unrealistic.

Now it feels normal.

Claude Code lets you think in outcomes, not syntax. You describe what you want in plain English. It translates that into structure, logic, and working code.

There’s a lot of noise about these tools at the moment. Plenty of theory. Plenty of “look what AI could do”.

Far fewer examples of marketers shipping real products.

That’s the shift.

The marketer who can prototype tools, build micro-products, test ideas, and deploy working assets without waiting for engineering is operating on a completely different level.

This is not about saving time.

It’s about leverage.

The idea was simple.

Build an interactive assessment tool that:

Not a chatbot.

Not a fluffy quiz.

A proper diagnostic tool grounded in real marketing methodology.

The AI Marketing Roadmap asks eight targeted questions covering:

It then scores responses across four dimensions:

Users receive:

Results are delivered instantly on screen, plus a formatted PDF for internal sharing.

Three minutes in.
A strategic action plan out.

That was the goal.

I used Claude Code Opus 4.6 for the entire development process.

Here’s what that actually looked like.

First, I described the product in marketing terms. User flow. Scoring logic. Experience goals.

The first build session created:

Then I iterated.

Each session focused on a specific component:

Claude handled:

n8n runs the email workflow.
The app is deployed on Netlify.

What surprised me most was how collaborative it felt.

It wasn’t like briefing a developer and waiting a week. It felt more like pair programming with someone patient and extremely fast.

I could say:

“I want the results page to feel premium, not like a generic scorecard.”

And it would translate that into layout changes, spacing adjustments, hierarchy shifts, and design refinements.

That speed of iteration changes how you think.

You try.
You test.
You adjust.
Within minutes.

Yes, things broke. Logic needed reworking. Outputs needed tightening.

But friction was low. And that makes building addictive.

This is the part that matters most.

The AI Marketing Roadmap is not just a prompt wrapped in a form.

The framework behind it is mine.

The scoring matrix.
The maturity levels.
The way answers map to recommendations.
The phased structure.

That all comes from years of hands-on strategy work.

AI powers the analysis layer. It takes structured inputs and generates nuanced, contextualised output quickly.

But the thinking is human.

There’s a huge difference between:

“Ask an LLM for a marketing plan.”

And:

“Use an LLM inside a structured methodology.”

The value is not in the text generation.

It’s in the framework shaping the text.

Anyone can generate words.

Very few people build systems.

If you’re a marketer, here’s the uncomfortable truth.

The teams who learn to build with AI will outpace the teams who only consume AI outputs.

They will:

They won’t ask, “Can we build this?”

They’ll ask, “Should we?”

That’s a very different posture.

You don’t need to become an engineer.

But you do need to become comfortable building.

The AI Marketing Roadmap is currently in Beta.

It’s free.
No credit card.
No bait and switch.

If you’re a marketing leader, founder, or operator working through the AI question, try it.

Three minutes.
You’ll get a personalised action plan.

More importantly, I’d value your feedback.

What feels useful?
What feels generic?
What’s missing?

That feedback will shape the next iteration.

👉 roadmap.goodvibemarketer.com

The real experiment here is bigger than one tool.

It’s this:

What happens when marketers stop just talking about AI and start building with it?

That’s the shift I’m interested in.

Ready to transform artificial intelligence from a buzzword into your personal revenue generator

HubSpot’s groundbreaking guide "200+ AI-Powered Income Ideas" is your gateway to financial innovation in the digital age.

Inside you'll discover:

Download your guide today and unlock a future where artificial intelligence powers your success. Your next income stream is waiting.

Get Your Guide