"Do You Use AI to Write?" — How Agencies Should Answer the Question Every Client Is Now Asking
It usually arrives mid-meeting, dropped in almost casually.
"Quick question — do you guys use AI to write our stuff?"
And in the two seconds before you answer, an entire internal debate plays out. Say no, and you're lying — and probably provably lying, because they've seen the word "delve" in a draft. Say yes with no framing, and you can hear the follow-up already: "Then what exactly am I paying you for?"
Most agencies fumble this moment. They hedge, deflect, mumble something about "leveraging tools," and leave the client more suspicious than before the question was asked. Which is a shame, because handled well, this question isn't a threat. It's one of the best trust-building openings a client will ever hand you.
Here's how to handle it — what the question actually means, what the data says about disclosure and trust, the answers that backfire, the answer that works (with word-for-word scripts), and how to get ahead of the question so it never ambushes you again.
First, Understand What the Client Is Actually Asking
Almost no client who asks this question is conducting a philosophical inquiry into machine authorship. The question is a proxy for three underlying fears:
"Am I paying human prices for machine output?" This is the big one. The client has seen ChatGPT. They know a draft can be generated in forty seconds. If your deliverable is indistinguishable from that, your invoice suddenly looks like arbitrage — and clients hate feeling arbitraged far more than they hate AI.
"Is this going to hurt me?" They've read headlines about AI content penalties, hallucinated facts, and plagiarism suits. They want to know someone competent is standing between the model and their brand.
"Will you tell me the truth?" Sometimes the client already knows the answer. They've spotted the tells — we've catalogued the exact phrases that give AI writing away — and the question is a test. Not of your workflow. Of your honesty.
Read the question this way and the right response becomes obvious: you're not defending a tool choice. You're addressing value, risk, and candor. Any answer that handles all three wins. Any answer that dodges even one loses.
The Data: Why "Just Deny It" Is a Terrible Strategy
Let's be blunt about the landscape, because it explains why honesty isn't just ethical — it's the only strategically sound option.
AI use in marketing is effectively universal. Per Salesforce's State of Marketing research, 87% of marketers now use generative AI in at least one workflow, up from 51% in 2024 — non-adoption is now the exception. McKinsey's global survey puts organizational AI use at 88%, with marketing among the top deployment areas. One German study of 600 marketing professionals found AI adoption at literally 100%. Your client knows this. If you claim to be the one agency on Earth running a fully artisanal, hand-typed operation, they will not believe you — and if they later find evidence otherwise, the relationship is done.
But trust in AI content is falling, fast. In Fractl's Q2 2026 survey of 1,008 U.S. consumers, 39% said heavy AI use in a brand's marketing would make them trust that brand less — roughly double the 20% who said so a year earlier, with the penalty strongest among Gen Z and women. A Klaviyo survey of 8,000 consumers found much the same: visible AI in content is a trust risk, not a trust builder. Your client's nervousness is not irrational. It's directionally correct.
And the disclosure gap is enormous. The same Fractl research found that 84% of consumers want AI-written content labeled — yet only 20% of organizations always disclose AI use, and 33% never do. That gap is a reputational liability waiting for a catalyst, and clients increasingly know it. When they ask you the question, they're partly asking whether you're the liability.
The nuance that saves you: the academic research on AI disclosure is more forgiving than the headlines. A 2026 systematic review of 35 studies found that while AI disclosure can erode trust, the effect is "neither universal nor uniform" — perceived authenticity is what actually mediates the outcome. And experimental work on AI labels shows that framing AI as a support tool for human work lands very differently from framing it as a replacement. In other words: "AI-generated" scares people. "Human-led, AI-assisted, and here's exactly how" generally doesn't.
That's your entire playbook in one sentence. Now let's build the answer.
The Four Answers That Backfire
Before the script, the anti-patterns. Agencies reach for these under pressure, and every one of them makes things worse.
1. The flat denial
"No, everything is 100% human-written." If it's true, fine — but for the overwhelming majority of agencies in 2026, it isn't, and clients can increasingly tell. One AI-flavored draft later, you haven't just been caught using a tool. You've been caught lying about your process, which contaminates every other claim you've ever made — your reporting, your timelines, your results.
2. The panicked over-apology
"We do sometimes, I know, I'm sorry, we can stop—" This answer accepts a premise the client never stated: that AI use is inherently shameful. It signals you don't understand your own workflow well enough to defend it, and it invites the client to start dictating your production process — a road that ends with you doing worse work, slower, at the same price.
3. The vague deflection
"We leverage a variety of cutting-edge tools in our tech stack." Ironically, this is the most AI-sounding answer available. Clients hear vagueness as concealment. If your process is defensible, describe it. If you can't describe it, that's the actual problem.
4. The tech lecture
Fifteen minutes on transformer architectures and prompt engineering. The client asked a trust question and received a TED talk. They leave more confused and no more reassured.
Notice what all four have in common: they answer the surface question ("is AI involved?") instead of the real ones (value, risk, honesty).
The Answer That Works
The winning structure has four beats: honest yes → where AI sits in the process → where humans are irreplaceable → what the client is actually paying for. Delivered confidently, it takes about sixty seconds.
Here's the word-for-word version. Adapt the details to your actual workflow — and only say it if it's true, because the entire strategy rests on it being true:
"Yes — and I'd be worried about any agency in 2026 that told you otherwise. Nearly ninety percent of marketing teams use AI somewhere in their workflow now, so the honest question isn't whether an agency uses it, it's how, and whether they'll tell you straight.
Here's exactly how we use it. AI helps us with the parts of content work that are research and assembly: synthesizing source material, generating outline options, producing rough first-pass drafts, and pressure-testing structure. That's maybe the first 30% of the work, and it's the 30% that used to eat hours without making the content any better.
Everything that determines whether the content actually performs stays human: the strategy behind what we write and why, the original data and examples, every claim getting verified against a real source, the editing pass that gives it your brand's voice instead of a chatbot's, and the judgment about what your buyers actually need to hear. No draft ships to you without a human who understands your business rewriting and standing behind it.
What you're paying us for was never typing. It's knowing what to say, proving it's true, and making it perform — in rankings, in AI citations, and in pipeline. AI made the typing cheaper. It made the judgment more valuable, not less."
That's the whole move. You've told the truth, reframed the question from tool to process, drawn a bright line around human value, and reconnected the conversation to outcomes.
Handling the Follow-Ups
The first answer usually triggers one of four follow-ups. Have these ready.
"So why am I paying agency rates?"
"Because the deliverable was never a document — it's the outcome the document produces. Anyone can generate words now, which is exactly why generic words have stopped working. What moves rankings and pipeline is everything AI can't supply about your business: positioning, proof, original data, and an editor who knows the difference between content that fills a page and content that converts. If anything, the flood of cheap AI content has made that layer more valuable — it's the only thing left that differentiates."
This is also a moment to show, not tell. We wrote a full breakdown of what actually converts: AI copywriting vs. human copywriting — send your equivalent, or send data.
"Won't Google penalize AI content?"
"Google's stated position is that it rewards helpful content regardless of how it's produced and penalizes content made primarily to manipulate rankings — regardless of how it's produced. What actually gets punished is unedited, generic, zero-expertise content, which describes most AI output published raw, and none of ours. The bigger opportunity is on the other side: AI search engines cite sources with original data and real authority. That's a bar raw AI output can't clear by definition, and it's the bar we build to."
If the client wants depth, point them to a real resource — ours is the complete guide to getting found in AI search.
"Could I just do this myself with ChatGPT?"
Don't get defensive — agree, precisely:
"You could absolutely generate drafts yourself, and for some internal content you probably should. What's hard to replicate is everything wrapped around the draft: knowing which topics can actually rank and get cited, the keyword and entity strategy, the editing that removes the AI fingerprints, the fact-checking, the internal linking architecture, the authority building, and the measurement loop that tells you what's working. That system is the product. The draft is one input to it."
"Will you disclose this in the content itself? Do we have to?"
"That's your call to make with full information, and here's the information: 84% of consumers say they want AI-written content labeled, and disclosure norms — including regulation like the EU AI Act's transparency rules — are tightening. Our view: because every piece is human-verified and human-finished, 'AI-assisted, human-edited' is both accurate and, per the research, the framing audiences accept. We'll follow whatever policy you set, and we'll put our process in writing so you always know exactly what you're publishing."
Get Ahead of It: Stop Waiting for the Question
The best version of this conversation is the one the client never has to initiate. Three moves take the question off the table permanently:
1. Write an AI use policy — one page, plain English. What AI is used for (research, outlines, first drafts), what it is never used for (final copy, factual claims, strategy, anything client-confidential without approved tooling), and the human review every deliverable passes through. Attach it to proposals. The agencies losing this conversation are the ones improvising it live; per Fractl, only about 20% of organizations have consistent disclosure practices — meaning a one-page policy alone puts you ahead of roughly 80% of the market.
2. Put it in the contract. A short clause covering AI use, confidentiality (what client data can and cannot be entered into which tools), IP ownership of outputs, and accuracy accountability. This converts "do you use AI?" from an awkward gotcha into a documented, already-answered term of business.
3. Raise it proactively in the sales process. One slide, thirty seconds: "Here's how we use AI, here's what stays human, here's why that combination outperforms both extremes." Prospects are comparison-shopping agencies who dodge this topic. Being the one who volunteers it is memorable — and it pre-frames every competitor's hedging as evasion.
Why We'd Tell You to Answer This Way (and Why We Do)
Full disclosure, since this whole article is about disclosure: this is how we operate at Ritner Digital, and it's not incidental to our positioning — it is our positioning.
We use AI across our content systems for the acceleration layer, and humans for everything that earns rankings, citations, and trust: strategy, original research, verification, voice, and the copywriting judgment that separates answer-ready content from filler. Then we do the thing almost no agency does — we publish our own Search Console data every month, uncomfortable parts included, from our first 90-day report card onward.
We do that because the deeper lesson of the "do you use AI?" question applies to everything an agency says: claims are cheap now. Anyone can generate confident-sounding promises at scale — which means proof is the only currency left. The agencies that thrive over the next five years won't be the ones that hide their AI use or the ones that automate everything. They'll be the ones whose answer to every uncomfortable question is "here's exactly what we do, and here's the data."
Transparency stopped being a virtue. It became a moat.
Frequently Asked Questions
Should an agency admit to using AI for client content?
Yes. With roughly 87–88% of marketers and organizations using generative AI, denial is neither credible nor sustainable — and getting caught in a denial damages trust far more than the AI use itself. The winning answer is an honest yes, immediately followed by a specific explanation of where AI sits in the workflow and where human strategy, verification, and editing take over.
Do agencies have to disclose AI use to clients?
Contractually, that depends on your agreement — but practically, proactive disclosure is the safer and stronger position. 84% of consumers want AI content labeled while only 20% of organizations consistently disclose, regulation like the EU AI Act is formalizing transparency requirements, and clients discovering undisclosed AI use tend to treat it as deception rather than a process detail. A one-page AI policy attached to your proposal solves this before it becomes a problem.
Will clients pay less if they know AI is involved?
Not if the value is framed correctly. Clients pay less when they believe the deliverable is raw machine output; they don't when they understand the deliverable is an outcome — rankings, AI citations, pipeline — produced by a system in which drafting is one small, accelerated step. Research shows audiences respond far better when AI is positioned as a support tool for human work rather than a replacement, and the same framing holds in client conversations.
Does Google penalize AI-generated content?
Google penalizes unhelpful content, not AI involvement per se — its guidance targets content produced primarily to manipulate rankings, whatever the production method. In practice, the content that gets hit is generic, unedited, expertise-free output published at scale. Human-verified, experience-rich, well-edited content performs regardless of whether AI assisted the draft — and it's the only kind that earns citations in AI search engines.
What should be in an agency AI use policy?
Four things, in plain English: the tasks AI is used for (research synthesis, outlines, first drafts); the tasks reserved for humans (strategy, factual verification, final copy, brand voice); data handling rules (what client information may be entered into which tools, under what confidentiality terms); and the review standard every deliverable passes before the client sees it. One page is enough. The point is that it exists in writing before anyone asks.
How should an agency respond if a client wants zero AI involvement?
Take it seriously, price it honestly, and clarify scope. Some clients — regulated industries, sensitive legal contexts — have legitimate reasons. Explain what "zero AI" actually changes: timelines lengthen, costs rise, and some competitive tooling (research, optimization) may be off the table. Many clients discover their real requirement isn't "no AI anywhere," it's "no unreviewed AI output and no confidential data in public tools" — both of which a good policy already guarantees.
The Question Is Coming. Decide What Your Answer Says About You.
Every agency will be asked this question dozens of times over the next year. Most will fumble it, and their clients will quietly start comparison shopping. The agencies that answer with specifics, confidence, and proof will convert the most awkward question in the industry into their sharpest differentiator.
If you'd rather be on the second list — or you're a business tired of agencies that can't give you a straight answer about their own process — that's exactly the conversation we like having. We'll walk you through how we work, show you our published data, and give you a real read on where your search and AI visibility stand.
Book a free 30-minute strategy call → No pitch, no obligation — and yes, a human will be on the call.
Sources
Fractl — AI Search Consumer Trust Study: Brand Visibility Strategies for 2026 (survey of 1,008 U.S. consumers and 150 marketers)
ContentGrip — AI Trust Drops as Usage Rises: Fractl's 2026 Survey
Digital Applied — AI Marketing Statistics 2026 (compiling Salesforce State of Marketing 2026, HubSpot, Gartner, McKinsey)
Whitehat — AI in Marketing 2026 Research Report (McKinsey Global AI Survey, Duke/Deloitte CMO Survey)
Nuremberg Institute for Market Decisions — Consumer Attitudes Toward AI-Generated Marketing Content
American Impact Review — Consumer Trust in AI-Generated Marketing Content: A Systematic Literature Review(35 studies, 2020–2026)
MDPI — AI Labels, Perceived Authenticity, and Consumer Trust in User-Generated Reviews
eMarketer — Shoppers Aren't Impressed by AI-Generated Marketing (Klaviyo survey of 8,000 consumers)
YouGov & Meltwater — Trust in the Age of Generative AI (survey of ~10,000 consumers)