AI in marketing has enormous potential to accelerate workflows and content production — but it can also slow things down. In episode 136 of FiredUp!, Morgan McLintic and Chris Ulbrich unpack where AI is actively gumming up agency work, why it’s happening, and the specific places where human judgment and taste have to stay in the driver’s seat. They are, for the record, huge users of generative AI and big believers in it — Firebrand has built agents and AI-enabled apps and is all in on the tooling. The argument is about being judicious about when not to use AI in marketing, not abstaining from it.
Key Takeaways
- AI generated content is creeping into client-agency communications and external communications undetected, and downstream that creates real cost in time, accuracy, and brand voice.
- A short, human-written marketing brief beats a long, AI-generated one. AI fills space with averages and outdated information that someone downstream has to unpick.
- AI generated quotes for the media are a problem. Many publications explicitly prohibit them, and reporters increasingly spot the same AI turns of phrase across submissions.
- When editing material, share the prompt rather than the AI-edited output. The prompt is the real signal of intent; everything else is probabilistic sentence completion.
- AI gives you a commodity opinion. In a sea of brands fighting for attention with the same language, commodity language guarantees you don’t stand out.
- The cultural pressure to use AI is creating a tension where speed is taking a back seat to quality — and the work is moving backward as a result.
Why is AI in marketing creating problems for agency work?
Chris frames two categories where AI is currently doing more harm than good. The first is communications between client and agency teams: when AI writes those communications or the instructions to the agency, it adds words and meaning the client doesn’t necessarily intend or stand behind — but the PR agency has no way to tell the AI’s additions from the client’s actual point. The second is external communications — Q&As, product messaging, ready-to-publish material — where AI inserts strange turns of phrase, false distinctions, and arguments that don’t need to be made.
The cultural piece sitting under both: in tech right now, using AI is treated as virtuous in and of itself. Teams are under pressure to use it, and often the rough first draft AI produces gets passed along as if it were the final cut. The question Chris keeps returning to: are you actually saving time, or pushing uncertainty downstream to someone who has to spend hours undoing it?
Why is a human-written marketing brief better than an AI-generated one?
The traditional definition of “garbage in, garbage out” assumed something was actually going in. AI has introduced a new category: garbage out from the absence of input. Everyone who’s worked with AI intuitively understands that more structure and more guidance produce better output — but the more time you spend assembling that structure, the less time AI is saving you. So there’s a real incentive to skimp on input and let AI fill the gaps.
When that happens, AI does what AI does. It generates messaging based not on anything specific to your company, but on a general average of how a similar company might approach the problem. Worse, it can pull from your own old material — messaging from a press release three product revs ago, discarded years back. An agency that started working with you six months ago has no way of knowing that wording is out of date.
Morgan’s parallel example: RFPs. AI doesn’t generate 10 sensible questions. It generates 25 — bloated with questions about competitors that aren’t competitors and verticals you don’t target. When pushed back on, the response is invariably “oh no, I don’t actually want that information.” The fix is simple: a short human-written brief beats a long AI one. Put the time in at the front, save the time at the back.
Should you use AI to write quotes for the media?
In most cases, no. Many publications explicitly prohibit AI generated quotes, and reporters are actively screening them out. Feeding AI quotes into media outlets damages the company’s reputation, the agency’s standing, and the well for every other client.
The nuance Chris adds: AI generated quotes do find their way into the media, and so do entirely AI-generated articles. The safer working rule is to assume you need to write something original unless you know otherwise. If you do use AI to start, flag it — “Claude generated this, here’s the prompt I gave it” — and then rewrite it in a human voice so the agency can check it’s on message and sounds like a person.
There’s also a practical problem beyond the ethical one. The characteristic AI turns of phrase — quietly, shifts, the benefit compounds — are increasingly recognizable. Reporters spot them. Readers spot them and treat the content as AI slop. The whole point of giving a reporter a point of view is to be distinctive. AI, as is well-documented, gives you the lowest common competent response. It can sound polished. It will not sound different from the spokesperson at your competitor, who is prompting the same model on the same opportunity.
In rapid response situations where up to 60 sources are submitting quotes to a single call, the reporter receives 50 similar-sounding submissions. Speed without distinctiveness wastes the opportunity. The better discipline: pre-approved, human-developed messaging on the topics your PR agency tracks, so when an opportunity lands, the starting point is original.
Download the Multiplier Marketing Megapack today. Exclusive offer for all our listeners — get all our Startup Guides in one go! Over 50 pages of advanced tips and advice that dive deep into content marketing, search advertising, and marketing attribution.
Should you use AI to edit marketing copy?
This is the editing trap. A piece of material has gone through approval, comes back from the client edited top to bottom by AI — not because anything was deeply wrong, but because a simple prompt like “make this tighter” or “add more about X” caused the model to restructure the whole thing. The agency receives a press release riddled with AI-inserted errors and extraneous arguments and no way to tell which changes the client cares about and which are AI filler.
The bigger principle Chris lands on: in a lot of cases, the prompt the client gave AI is the most important communication they could have given the agency. If you handed the agency the same words you handed Claude, the agency can write the press release — slower than Claude, granted, but able to explain every word choice and stand behind it. Anything AI generates beyond restructuring information that already exists is superfluous. The prompt is the signal; everything else is probabilistic sentence completion the agency has to filter out, on the client’s budget.
Morgan adds the byline-specific failure mode: a handwritten contributed article gets a quick AI smoothing pass before being sent back, which trips AI detection screens at publications. What looked like a ten-minute time-saver makes the article unusable, and the whole thing has to be rewritten.
Why does AI produce commodity language?
AI doesn’t generate distinctive language. It generates the statistical middle. The structure of an AI sentence is often so punchy and smooth that the eye glides past on first read — and then you actually read it and the word choice is three degrees off what any native speaker would have picked. Chris calls it the uncanny valley of prose: polished enough to seem competent, not quite human enough to land.
Chris also points to a familiar AI tic: pushing for definitive this-not-that arguments where the truth is more nuanced. When you challenge a claim it has made, the model often shrugs and says something like “I just thought this is the kind of thing you would say.” The positions AI inserts have to be checked carefully, because large language models will frequently hallucinate.
How is daily AI use changing how we read and write?
Morgan raises the question. Handwritten LinkedIn articles and blog posts now read as lumpy and less polished than the AI smoothness we’re acclimating to. Are we starting to expect — and accept — AI-shaped language as the default?
Chris’s read: there’s a counter-movement. People are intentionally roughing up their writing, leaving in typos, using double dashes instead of em dashes to signal “a human wrote this.” But LinkedIn is, in his words, an absolute fire hose of AI slop — and some of the most cynical examples are PR people posting about the need for a human voice in posts that were themselves clearly AI generated content. The mass phenomenon is the move toward using AI to communicate. Resistance to it may stay niche, like spinning vinyl.
Morgan’s concern lands the broader argument: when you stop writing, you stop forming the thoughts that writing forces you to form. The thinking gets worse alongside the writing. That’s the doom spiral risk for thought leadership specifically, and for marketing teams more broadly. Putting in the time to write sharpens the point of view; outsourcing it dulls the brand.
Why does this matter for brand awareness?
This connects directly to Firebrand’s broader argument about brand awareness in a saturated market. External communications around your product is exactly where you want to be most distinctive. AI will make you most average. In a sea of brands fighting for attention with more or less the same language, you cannot stand out using commodity language — and commodity language is what AI gives you.
The principle that ties the whole conversation together: AI is a tool. Marketers are paid for taste, judgment, creativity, and experience. The places where AI helps are the places that don’t require those things — structuring information, accelerating well-defined tasks, generating variations. The places where it hurts are the places where those things are the entire point.
FAQ: AI in Marketing
When should marketers use AI?
For internal-facing tasks: brainstorming, outlining, structuring arguments, summarizing research, and accelerating repetitive work. It can also be useful for first-pass drafts a human will substantially rewrite. The common thread is that a human reviews and shapes the output before anything goes external.
When should marketers avoid using AI in marketing?
For external communications where voice, accuracy, and point of view matter — quotes for the media, thought leadership, core messaging, contributed articles, and marketing briefs. Many publications also explicitly prohibit AI generated quotes, so submitting them risks the placement and the relationship with the reporter.
Why is a human-written brief better than an AI-generated one?
AI briefs tend to be long because the model fills in gaps with averages — generic framing, outdated messaging it pulled from your old press releases, and information that isn’t actually relevant. A short, human-written marketing brief gives the agency exactly what’s intended and saves time on the back end. Put the time in at the front.
What's the problem with AI-edited material?
A simple prompt like “make this tighter” can cause AI to restructure the whole document, introducing errors, extraneous arguments, and AI-characteristic turns of phrase that trip AI detection screens at publications. The agency has no way to tell which changes the client cares about and which are AI filler — so the edit kicks an hour or two of work back over, on the client’s budget.
What's the best practice if you do use AI to edit?
Share the prompt, not just the output. The prompt tells the agency what the client actually wanted changed. Without it, the agency is guessing what is intention and what is probabilistic sentence completion — which uses a lot more cycles than the edit appeared to save.
Why does AI produce commodity language?
AI generated content reflects the statistical middle of how things are typically written because large language models average patterns across enormous datasets. That’s the opposite of a distinctive brand voice. In rapid response situations where dozens of sources are submitting quotes on the same topic, AI ensures everyone sounds the same — which means no one stands out.
Does using AI hurt thought leadership?
It can, when it replaces the writing process entirely. Writing and editing are where marketers sharpen a point of view and surface original insights. Outsourcing that work removes the friction where strategic thinking happens, leaving polished copy with no perspective behind it.
Find Us
Connect with Firebrand
Firebrand is a brand awareness agency for B2B AI companies. We help AI startups become visible, discoverable, and memorable to the buyers, investors, talent, and partners that matter.
Related reading:
- Awareness Wins: Why Brand Is the #1 Growth Lever for AI Companies
- The Great GEO Debate: Is this just SEO or is it More?
- Firebrand Blog
Find us on YouTube, LinkedIn, and X. Work with Firebrand or apply to be a podcast guest.