Every agency has a position on AI right now. Some are using it for everything. Some are avoiding it entirely. Most are somewhere in the middle, figuring it out as they go without a clear principle behind the decisions.
Here’s the principle we use at MORE: AI can replicate. It can’t ideate.
That one distinction tells you almost everything you need to know about when AI belongs in your workflow and when it doesn’t. If a task involves finding, organizing, or reproducing patterns that already exist somewhere, AI can help. If a task requires original thinking (like judgment, empathy, strategy, or creative instinct), AI is the wrong tool. Not because it’s bad software, but because that’s not what it does. AI predicts the most statistically likely output based on patterns it has seen before. That’s fundamentally different from human reasoning, and no amount of prompting changes it.
“My favorite thing to use AI for in our business is hands down recording my meetings and making my tasks from them. Fun fact: if you use Google Workspace, it’s built in and free!” -Cole
Where AI earns its place
AI genuinely helps a marketing agency on tasks where the value is speed and accuracy, not originality.
Summarizing meetings and long documents is a good example. If you have 90 minutes of call notes that need to be distilled into three action items, AI can do that faster than any person. What to do with those action items is still your call.
Proofreading and grammar checks fall into the same category. Finding inconsistencies in a 40-page document, catching a client’s brand name spelled two different ways, and flagging passive voice are pattern-matching tasks that AI handles well.
Debugging code and writing small scripts belong here too. If you are building a tracking setup, automating a report, or trying to figure out why a form is not firing correctly, AI can walk you through the fix or write the code. It’s very good at this because code has right and wrong answers.
Analyzing data for patterns is another strong use case. Pull six months of Google Search Console queries into a document and ask AI what themes are emerging. It will find them faster than you can. The strategy you build on those themes is still your job.
Other places where AI earns its keep: building templates and SOPs for repetitive internal processes, looking up how to do something specific in a tool you are learning, quickly getting up to speed on an industry before a strategy call, and generating multiple variations of something low-stakes like email subject lines so you have options to test.
All of these tasks share something. The parameters are already defined. You know what good looks like. You’re asking AI to do the mechanical work of getting there faster.
Where AI has no business being
AI breaks down on tasks that require knowing things it simply cannot know.
Published copy is the clearest example. AI doesn’t know your client’s voice. It doesn’t know the inside joke their audience would recognize, the phrase their founder uses in every interview, or what their customers actually complain about on review sites. It produces text that sounds like marketing copy because it has read a lot of marketing copy. That’s not the same as knowing what this brand should say.
Campaign ideation and strategy are off limits for the same reason. A great campaign idea comes from understanding a specific audience, a specific market moment, and a specific brand’s position. A great campaign idea requires judgment AI hasn’t earned. AI can generate ideas, but they will match the patterns of every other campaign it has seen. Standing out requires thinking that breaks from the pattern, not one that follows it.
Tone of voice and brand messaging belong in the same column. They’re essential decisions about what a company believes, who it talks to, and how it wants people to feel. That requires empathy, earned through real conversations with real clients.
Discovery conversations can’t be replaced or skipped. When a client tells you about their business, the useful information is often not in what they say directly. It’s in what they emphasize, what they come back to, what they hesitate on. A human picks that up. AI can’t be in the room.
Don’t use AI for anything involving sensitive client information, especially anything touching HIPAA-regulated data. Most AI tools are not built for that kind of data handling and privacy, and the risk isn’t manageable.
Generating imagery or visuals for a client’s brand falls under the same rule. The same logic that keeps AI out of logo design keeps it out of any creative output. You can’t prompt your way to a visual that is authentically right for a specific brand.
The question to ask
Before you use AI for any task, ask yourself: does doing this well just mean finding an existing pattern and following it accurately?
If yes, AI can help.
If doing it well means understanding something specific about this client, this audience, this moment, or this brand in a way that requires judgment and empathy — that task stays with your team.
AI is a powerful execution tool. It’s not a substitute for the thinking that decides what to execute. An agency that understands that will use AI to free up its team for the work that actually requires them. An agency that doesn’t will find out what happens when replication replaces ideation. Their clients will notice first.
If you want to work with an agency that uses AI where it helps and keeps humans where it matters, reach out here. We offer digital marketing strategy, branding, social media management, and more, all built on work your team actually did.