I run Margle Media, a marketing agency working with brands like Petco, Post, Cousins, Qdoba, Johnsonville, and others. I also run GAS Studio, where AI is embedded in everything we build. Between the two, I have a practical — not theoretical — perspective on AI in marketing agencies.
The conversation around AI marketing tools usually falls into two camps: breathless enthusiasm ("AI will replace all marketing jobs!") or dismissive skepticism ("AI can't do real creative work"). Both are wrong. The truth is more useful and more nuanced. Here's where AI actually handles the work at Margle, where it doesn't, and how we decide which is which.
Where AI Handles Real Work
These are the areas where AI has become a genuine part of our agency AI workflow — not experimental, not optional, but essential to how we operate daily.
Research and briefing. Before any campaign, strategy session, or client presentation, someone needs to assemble context — competitor analysis, industry trends, audience insights, historical performance data. This used to take a junior team member half a day. Now AI does the first pass in minutes.
We feed AI the client brief, relevant data sources, and a set of questions. It returns a structured briefing document that's 80% complete. A human reviews, adds judgment and nuance, and we have a brief that's better than the old version in a fraction of the time. This alone has transformed our agency AI workflow.
First-draft content. Blog posts, social media copy, email sequences, ad copy variations. AI generates the starting point — sometimes a rough outline, sometimes a nearly complete draft. The key word is "starting point." Every piece gets human editing for brand voice, strategic alignment, and the kind of specificity that comes from actually knowing the client's business.
At GAS Studio, this is how the Journal operates too. AI helps produce content at scale, but the perspective, the experience, and the editorial decisions are human. That's the same model we use at Margle.
Reporting automation. Weekly and monthly performance reports used to consume significant team hours. Now AI pulls data from ad platforms, analytics tools, and CRM systems, structures it into report templates, and highlights the metrics that matter. The human layer adds interpretation: "Here's what these numbers mean for your business and what we recommend doing next." Marketing automation AI at its most practical.
Meeting summaries and follow-ups. Every client call generates action items, decisions, and follow-up tasks. AI transcribes, summarizes, and creates structured action lists. This sounds small. It's not — it eliminates the gap between "we discussed it" and "we documented it," which is where agency balls get dropped most often.
Competitive monitoring. AI continuously scans competitor activity — social posts, ad creative, website changes, content publishing — and surfaces relevant changes. Instead of manually checking competitor feeds, we get an automated intelligence briefing. This feeds into the strategic frameworks we use for paid advertising and content planning.
Where AI Stays Out
These are the areas where we've tried AI, evaluated the output, and deliberately decided to keep humans in control. Not because AI can't do something here, but because what it produces isn't good enough for clients paying premium rates.
Brand strategy. Defining a brand's positioning, voice, values, and strategic direction requires the kind of holistic judgment that AI doesn't have. AI can analyze market data. It can't sit in a room with a founder and feel the energy of what they're trying to build. It can't detect the gap between what a brand says and what it actually is. Building a memorable brand requires human insight that goes beyond data synthesis.
Creative direction. The strategic decision about what to create — not the production of it, but the why and what — remains fundamentally human. Which story do we tell? Which angle resonates with this specific audience at this specific moment? Which creative risk is worth taking? AI can generate options. Choosing between them requires taste, experience, and cultural awareness that AI mimics but doesn't possess.
Client relationships. This might sound obvious, but it's worth stating: the relationship between agency and client is human. The trust, the difficult conversations, the reading between the lines when a client says "it's fine" but means "I'm worried" — these are social and emotional skills that define great agency work. No AI marketing tools replace the phone call where you sense something's off and address it before it becomes a problem.
Final-mile editing. AI gets content 80% of the way there. The last 20% — the sentence that needs to land differently, the word choice that carries the wrong connotation for this specific brand, the paragraph that should be cut because it dilutes the piece — that's editing, and editing requires a human who understands the strategic context, the audience, and the brand's specific voice.
Ethical judgment. Should we run this campaign? Is this messaging potentially harmful? Does this creative direction cross a line? These questions don't have algorithmic answers. They require values, empathy, and a willingness to say "we can do this, but we shouldn't." AI for advertising that lacks ethical guardrails is a liability, not an asset.
The Framework: AI as a Team Member
The mental model that works best for us — both at Margle and at GAS Studio — is treating AI as a team member with specific strengths and specific limitations.
You wouldn't ask your best data analyst to develop brand strategy. You wouldn't ask your creative director to build a reporting dashboard. AI is the same: powerful in its lane, counterproductive outside it.
The strengths: speed, consistency, pattern recognition, synthesis of large information sets, production of variations at scale, and elimination of repetitive tasks. In our agency AI workflow, these strengths are deployed daily across every client engagement.
The limitations: originality, judgment, emotional intelligence, cultural nuance, strategic creativity, and the ability to know when something technically correct is strategically wrong. These limitations aren't temporary gaps that will be solved with the next model update. They're fundamental to the current paradigm.
Understanding these boundaries clearly — and communicating them honestly to clients — is what separates agencies using AI responsibly from agencies using AI as a cost-cutting shortcut.
Knowing When to Stop
The most important skill in deploying AI in marketing agencies isn't knowing when to use it. It's knowing when to stop.
There's a seductive pull toward "let AI do more." Every time it produces something adequate, you think: maybe it can handle the next step too. And sometimes it can. But the quality gradient between "adequate" and "excellent" is where client value lives, and that gradient is where human expertise earns its premium.
At Margle, we draw the line at the point where AI's output requires more editing to meet our standards than starting from scratch would. If an AI-generated strategy deck needs 60% revision, we're not saving time — we're creating a worse version of what a human would have produced directly.
The honest calculation: AI saves us roughly 30% of production time across the agency. That time gets reinvested into strategic thinking, client relationships, and creative development — the human-only activities where our value is highest. Marketing automation AI doesn't replace the team. It amplifies what the team does best.
What This Means for Agencies (And Everyone Else)
If you're running a marketing agency, the question isn't whether to use AI. You're already behind if you aren't. The question is how — with what boundaries, what quality standards, and what transparency to clients.
If you're a business evaluating agencies, ask how they use AI. Not whether they use it — they do, or they're lying — but where they draw the line. An agency that uses AI marketing tools for production but keeps humans in control of strategy, creativity, and relationships is using it right. An agency that uses AI to cut corners on the work you're paying premium rates for is not.
And if you're a builder like me, running ventures where every efficiency matters, the principle is the same: automate the production, protect the perspective. AI handles the predictable. Humans handle the meaningful. That division of labor is the foundation of everything we build at GAS Studio.
Follow the Journal for more honest takes on AI, marketing, and building ventures. Or get in touch to talk about how AI fits into your workflow.
This entry is part of our Systems & Scale series. Subscribe to the GAS Studio Journal RSS feed to stay in the loop.
