---

title: "Communicating your use of AI to clients"

module: 12

module_title: "Special Situations"

order: 1

access: "paid"

summary: "Decide your AI story before clients ask, and deliver it proactively instead of improvising under pressure. Covers disclosing AI usage at kickoff, keeping accountability for the output, answering the data-safety question with a real policy, handling the 'shouldn't this cost less' pricing question, and working within client-side AI restrictions."

related:

• "proactive-communication"
• "balancing-transparency-with-discretion"
• "having-money-conversations"
• "instill-confidence"

---

Communicating your use of AI to clients

On a project call last year, a client stopped mid-sentence and asked, "Wait, is AI writing this code?" The developer on our side did the worst possible thing: he hesitated, then gave a vague answer about "using various tools to be efficient." You could feel the trust drop in real time. Not because we were using AI, everyone is, but because the hesitation made it look like something we'd been hiding.

Here's what stung: our AI usage was actually a good story. It was making the work faster and the testing more thorough. We just had never decided how to talk about it, so in the moment, it came out sounding like a secret.

Why this conversation is suddenly everywhere

Clients are living in the same world you are. Their CEO is asking why projects still take months. They've read that AI writes code now. Some of them are quietly wondering if they still need you at all, and others are worried you'll feed their confidential roadmap into a chatbot that trains on it. Both anxieties are usually running at the same time, in the same person, which is why client questions about AI come out sideways: "So are you all using AI for this?" is rarely a question about tooling. It's a question about value, or about safety, or about whether they're paying artisan prices for machine output.

If you don't give clients a clear story about how you use AI, they will write their own. And the story they write will either overestimate what you're hiding or underestimate what you're doing. Neither is good. This is a genuinely new communication problem, and almost nobody has figured out their answer yet, which means having a clear one is an actual competitive advantage right now.

How I think about this

Decide your story before you're asked. The core failure in my opening anecdote wasn't AI usage, it was improvising the answer. Your team should be able to answer "do you use AI?" identically, calmly, and in one breath. That takes a decision, made once, in advance: here's what we use it for, here's what we don't, here's how we protect your data, here's who's accountable.

Bring it up before they do. Disclosure that arrives proactively reads as confidence. The same information, extracted by a client's question, reads like a confession. A short paragraph in your proposal or kickoff, saying how you use AI, costs you nothing and takes the entire issue off the table before it becomes charged.

You're accountable for the output, full stop. This is the sentence clients most need to hear, and it's the one that resolves the quality anxiety: every line of code, every design, every document goes out under our name, reviewed by our people, and we stand behind it exactly the same regardless of what tools produced the first draft. AI changes how work gets made. It cannot be allowed to change who's responsible for it.

Answer the data question with specifics. "Your data is safe" is a sentence that means nothing. "We use business-tier tools that don't train on customer data, and your source code and customer records never go into any tool outside that list" is a policy. If you can't state your policy in two sentences, you don't have one yet, and you should fix that before a client asks, because they're going to ask.

Sell the outcome, not the hours saved. The awkward question underneath everything is pricing: "If AI makes you 40% faster, why does this cost the same?" The honest answer is that clients were never really buying hours, they were buying outcomes, judgment, and accountability, and AI raises what a given budget buys rather than discounting the old deliverable. But you have to actually deliver on that framing: faster turnarounds, more included in scope, more thorough testing. If AI makes you faster and clients see zero benefit, that question becomes fair.

Don't oversell it either. There's an equal and opposite failure mode: agencies slapping "AI-powered" on everything and promising magic. Clients have been burned by that already. Understated and specific beats breathless every time.

What this looks like

The proactive disclosure at kickoff

A short section in the kickoff or proposal walkthrough:

"One thing I want to cover directly, because clients rightly ask about it: how we use AI. Our developers use AI tools for drafting code, writing tests, and research. It's a big part of why our estimates are what they are. Two things don't change because of it. First, everything that ships is reviewed by our senior people, and we're accountable for all of it, same as ever. Second, your data: we only use business-tier tools with no-training agreements, and your customer data and proprietary code stay inside that boundary. Happy to go deeper on any of this, and if you have policies on your side about AI usage, tell us now and we'll follow them."

Answering "so is AI just doing this?"

Client asks, with a slight edge, whether AI is doing the work they're paying for:

"Fair question, and I'll give you the straight answer. Yes, AI writes a lot of first-draft code here, the same way it does at most serious shops now. What you're paying us for is everything around that: knowing what to build, catching the ways your inventory system will break it, reviewing everything that ships, and being on the hook when something goes wrong at 5pm on a Friday. Honestly, the tools have made our output better, the test coverage on your project is way beyond what we could've afforded to hand-write. But if any part of this setup doesn't sit right with you, let's talk about it now."

The pricing question

Client says: "If AI makes you this much faster, shouldn't this cost less than last year's project?"

"It's a reasonable question, so let me be honest about how we think about it. Our pricing was never really hours times a rate, it was priced against the outcome, and that hasn't changed. What has changed is what a project like this includes. Two years ago, this budget got you the platform build. Now it gets the build plus full automated test coverage, plus the admin tooling we used to call out of scope, on a timeline about a third shorter. So the answer isn't a discount on the old thing. It's that you're getting meaningfully more thing. If it's useful, I can show you the scope of your 2023 project next to this one."

When the client has their own AI restrictions

Client says their legal team prohibits vendors from using AI on their materials:

"Thanks for flagging that now instead of in month three, seriously. Help me understand the boundary so we scope this correctly. Is the concern about your data going into AI tools, or about AI-generated work product in general? Those are pretty different constraints. If it's data, we can likely satisfy legal with our existing no-training agreements, and I'm happy to get on a call with them. If it's a blanket ban on AI-assisted work, we can do that, but I'd want to re-quote the timeline honestly rather than pretend it changes nothing. Either way, we'll follow your policy. Let's just get it precise."

What goes wrong in AI conversations

The hesitation. The pause before answering is more damaging than almost any answer. It converts a tooling question into a trust question.

Vagueness that reads as evasion. "We use various modern tools" is the kind of sentence people use when hiding something, and clients know it.

Overpromising the magic. "AI lets us do this in half the time at half the cost!" sets expectations that the messy middle of a real project will betray. The disappointment lands on you, not the tools.

Different answers from different team members. If the account lead says one thing and a developer says another on the same call, the inconsistency itself becomes the story. This is what happens when the answer was never decided, only improvised.

Hiding it until discovery. A client who finds an AI artifact in a deliverable, a telltale comment, an odd turn of phrase, after being told or led to assume the work was all-human, doesn't just discount that deliverable. They re-audit everything you've ever sent them.

Getting better at this

Write your AI policy down this week. One page: tools in use, what they're used for, what data can and cannot go into them, who reviews outputs. You can't communicate a policy that doesn't exist, and writing it will surface disagreements on your own team you didn't know you had.

Rehearse the three questions. "Do you use AI?", "Is my data safe?", and "Why does it still cost this much?" Every client-facing person should have fluent, consistent answers. Actually say them out loud in a team meeting. The first attempts will be rougher than you expect.

Put a disclosure paragraph in your proposal template. So it goes out proactively every time, and nobody has to remember or decide in the moment.

Update the story as your practices change. Your AI usage a year from now won't match today's. Revisit the policy quarterly, and mention meaningful changes to ongoing clients before they notice on their own.

How this connects

This is a new topic wearing old clothes. The mechanics are the same ones that run through this whole collection: proactive communication beats reactive, specificity beats vagueness, and balancing transparency with discretion is a judgment call you make out loud. The pricing question is a money conversation. The accountability answer is instilling confidence. What's new is only the subject, and the fact that your competitors haven't figured out their answer yet.

Things to try

• Write your one-page AI policy: tools, uses, data boundaries, review process. If your team can't agree on it, that's the real finding.
• Add a three-sentence AI disclosure to your proposal template this week.
• Role-play the "is AI just doing this?" question in your next team meeting. Have your most junior client-facing person answer it. Coach until it's smooth.
• Ask your current clients whether their companies have AI usage policies for vendors. Better to know now.
• Find one concrete way AI improved a current client's project, better test coverage, faster turnaround, extra scope, and mention it in your next status update, casually and specifically.

The clients aren't going to stop asking. A year from now, "how do you use AI?" will be a standard procurement question, as normal as asking about insurance. The teams that thrive won't be the ones with the fanciest tools. They'll be the ones who can answer the question in one calm breath, because they decided the answer before anyone asked.

Template: Proactive AI Disclosure at Kickoff

Use this when: you're starting an engagement and want to get your AI story out clearly and unprompted, before it becomes a charged question later.

Channel: Email

```template

Subject: How we use AI on your project

Hi [NAME],

One thing I like to cover directly at the start, because clients rightly ask about it: how we use AI.

Our team uses AI tools for [WHAT IT ACTUALLY TOUCHES — e.g. drafting code, writing tests, and research]. It's a real part of why our estimates and timelines are what they are.

Two things don't change because of it:

First, accountability. Everything that ships is reviewed by our senior people, goes out under our name, and we stand behind all of it exactly as we always have. AI changes how the first draft gets made. It doesn't change who's responsible for the result.

Second, your data. We only use business-tier tools with no-training agreements, and your [SENSITIVE DATA — e.g. customer data and proprietary code] stays inside that boundary. I'm happy to share the specific policy in writing.

And if your side has any policies about vendors using AI, tell me now and we'll follow them. Better to get it precise up front than discover it in month three.

Happy to go deeper on any of this whenever you'd like.

[YOUR NAME]

```