How I Use AI to Run Marketing for a 70-Outlet Chain

AI in marketing is the use of artificial intelligence tools to execute creative, analytical, and operational tasks that previously required either a large team or painfully slow workflows. For me — running brand and growth at FiLLi Cafe, an 80+ outlet tea chain across the UAE — AI didn’t replace my team. It replaced the gap between having an idea at 9am and seeing it live by noon.

That distinction matters. Most AI marketing content is written by people who demo tools. I use them to run a real operation with real revenue, real customers, and real deadlines. This is what that actually looks like.

How Does AI Actually Help Run Marketing at Scale?

The honest answer: AI helps with the parts nobody wants to do.

Not the strategy. Not the brand taste. Not the instinct for what will resonate with your customer. Those are still mine. AI handles the execution layer — the part where good ideas used to go to die because the friction of producing them was too high.

Before AI, here’s what launching a seasonal campaign looked like at FiLLi: brief the designer (2 days), wait for creative options (3-5 days), review and revise (2 more days), brief the media buyer, set up the campaign in Meta, wait for results, manually pull reports. Two weeks from idea to live ad, minimum.

Now, I can concept a campaign, generate ad creative variations, set up targeting, and launch — often in the same day. Not because the quality bar dropped, but because the bottlenecks disappeared.

The work that used to require five calendar days of back-and-forth now happens in a focused two-hour block. That’s the real unlock. Not replacing humans. Compressing time.

What Tools Actually Work and Which Are Just Noise?

I’ll be specific because vague AI advice is useless.

Claude Code is the tool that changed my workflow the most. I used it to build ashmo.io — this entire website — from architecture to deployment. Not as a developer. As a brand strategist who described what I wanted and iterated in real time. The site runs on Astro, deploys through Netlify, and the entire build happened through conversation. I described the structure. Claude wrote the code. I reviewed, adjusted, pushed live. No agency. No three-week sprint cycle. No “we’ll get back to you with a quote.”

That experience rewired how I think about digital projects. The cost of building something dropped so dramatically that I now prototype ideas I would have previously dismissed as “not worth the development time.”

For Meta ad creative, I use AI to generate copy variations at volume. When you’re running campaigns for a chain with 80+ locations, you need localised messaging — what works for a Mall of the Emirates audience doesn’t always land the same way in Deira. AI lets me produce 15 copy variations in the time it used to take to write three. I still pick the winners. The taste is still mine. But the raw material comes faster.

For content repurposing, a single long-form piece — like this article — becomes social posts, newsletter sections, ad copy hooks, and internal talking points. AI doesn’t create the original thinking. But it’s excellent at reformatting that thinking for different channels and audiences.

For campaign analysis, I pull performance data from Meta and ask AI to surface patterns I might miss when I’m deep in the numbers. Not replacing analysis — augmenting it. A second pair of eyes that doesn’t get tired at 11pm.

What Can AI Replace and What Can’t It?

This is where most AI content gets dishonest. They either oversell it (“AI will replace your entire marketing department!”) or undersell it (“AI is just a toy, real marketers don’t need it”). Both are wrong.

Here’s my honest breakdown after 18 months of daily use:

What AI handles well	What still needs a human
First drafts of ad copy and social posts	Brand voice and final editorial judgement
Code generation for websites and tools	Architecture decisions and UX instinct
Data formatting, reports, and summaries	Interpreting what the data means for your specific business
Creative variations at volume	Knowing which variation actually fits the brand
Repurposing content across formats	Creating the original insight worth repurposing
Routine campaign setup and structure	Budget allocation and strategic trade-offs
Research synthesis and competitive scans	Relationship building with partners and customers

The pattern is clear: AI is exceptional at production and poor at judgement. It can build, but it can’t taste. It can write, but it can’t decide what’s worth saying.

“AI doesn’t give you better ideas. It removes the excuses for not executing the ones you already have.”

That’s the sentence I keep coming back to. The bottleneck in most marketing operations isn’t creativity — it’s the friction between the idea and the execution. AI collapses that friction. What you do with the cleared space is still entirely on you.

How Did Building a Website with AI Change My Perspective?

Building ashmo.io with Claude Code was a turning point. Not because the website itself was revolutionary — it’s a personal platform, a place for my thinking and work. The turning point was the process.

I’m not a developer. I’ve managed developers for years, briefed agencies, reviewed technical proposals. But I’d never built a production website myself. With Claude Code, I didn’t need to learn a programming language. I needed to know what I wanted and describe it clearly.

That’s a different skill. It’s closer to creative direction than coding. You’re steering, not typing. You’re making decisions about structure, flow, and user experience — and the AI handles the implementation.

The entire site — content architecture, page templates, responsive design, deployment pipeline — was built through iterative conversation. When something didn’t look right, I described the problem. When I wanted a new section, I described what it should do. The feedback loop was minutes, not days.

This matters beyond websites. It means that every digital project I evaluate now gets filtered through a different lens: “Could I prototype this in a day instead of scoping it for a quarter?” Often, the answer is yes.

How Does AI Fit Into Running Meta Ad Campaigns?

Meta ad campaigns are where AI has the most measurable impact for me.

Running ads for FiLLi across 80+ outlets means constant creative production. We’re not running one campaign — we’re running dozens, with different locations, audiences, offers, and seasonal moments. The volume alone used to be the constraint.

Here’s what a typical workflow looks like now: I identify a campaign objective, describe the target audience and the offer, and use AI to generate a batch of headline and body copy options. I review them against our brand voice, pick the strongest three to five, and set them up as variations. Within 48 hours, performance data tells me which ones to scale.

The speed advantage compounds. Because I can test more creative faster, I learn what resonates quicker. That learning feeds the next campaign. Over months, the pattern recognition — mine, not the AI’s — gets sharper.

But here’s what AI can’t do in this process: it can’t tell me whether to run the campaign at all. It can’t sense that a particular promotion will cheapen the brand. It can’t feel that a certain tone is wrong for this moment. Those are brand positioning decisions, and they require years of context that no model has.

What Do Most People Get Wrong About AI in Marketing?

Three things, consistently.

First, they use AI to replace thinking instead of to accelerate it. If you ask AI to “write a marketing strategy,” you’ll get something that sounds professional and means nothing. Strategy requires knowing your specific business, your specific customers, and your specific constraints. AI doesn’t know any of that unless you bring it.

Second, they treat AI output as final. First drafts from AI are starting points, not deliverables. Every piece of copy, every campaign structure, every analysis needs a human pass. The value isn’t in the first output — it’s in the speed of getting to that first output so you have time to actually refine it.

Third, they focus on the wrong tools. I’ve seen marketers spend weeks evaluating 15 different AI tools instead of getting deep with one. Depth beats breadth. I use Claude for almost everything — writing, coding, analysis, brainstorming. Knowing one tool deeply is more valuable than knowing ten tools superficially.

Is This the Future of Small Marketing Teams?

I think so, but not in the way most people mean.

The future isn’t “AI replaces marketers.” The future is that a small team with AI tools can compete with a team five times its size — if the small team has taste, judgement, and strategic clarity.

That last part is the filter. AI amplifies whatever you bring to it. If you bring clear brand positioning, AI helps you execute it faster. If you bring confused positioning, AI helps you produce confused content at scale. The tool doesn’t fix the thinking. It just removes the excuse of “we don’t have the bandwidth.”

I run brand and growth for an 80+ outlet chain. I build digital platforms. I write, produce, and ship marketing daily. A few years ago, that workload would have required a team of six or seven people. Today, it requires clear thinking, strong tools, and the discipline to know what still needs a human touch.

AI didn’t make me a better marketer. It made me a faster one. The difference between good marketing and great marketing is still the same as it always was — taste, timing, and the willingness to say no to most things so you can say yes to the right ones.

The tools changed. The work didn’t.