Communicators: We're Busy. Do We Have the Bandwidth to Adopt AI?

Happy Wednesday! This week:

Let’s get started!

We're busy. AI promises to make us less busy or, at the very least, more efficient. And whether that proves true, there’s a more immediate issue. Short-term, AI is causing us a problem — we just don't have the time to learn it or, worse, roll it out to our colleagues in ways that it sticks.

And that's an adoption problem if I've ever heard one. I spoke with Ellen Griley about this — she's given it a lot more thought than most everyone else. Her argument is that we're dropping it on people who are already running on empty, and that there's a discipline for exactly that problem. She's the founder of Equilibrious Communications, and she developed her approach, Internal Calms, while getting a graduate degree in trauma-informed leadership.

I hadn't applied that lens to AI adoption before, and it got me thinking.

Let's jump in.

🧠 Trauma-informed practice comes out of healthcare and social work, not communications. What is it, and why does it belong in the way we talk to employees?

💬 Ellen's take: Trauma-informed practice has a long history in healthcare, education, social work, and other disciplines like UX, architecture, and even yoga. It's a commitment to understanding how complex, sometimes invisible experiences shape the way people respond to the world around them.

An estimated 70% of adults have lived through at least one traumatic event. Then you think about things like uncertainty and burnout, or chronic stress — all of these affect how the brain processes information. How we concentrate, focus, or make decisions. Which . . . is what our communications are supposed to help people do, right? Especially employees.

This can't be overlooked, because our playbooks and strategies were built for a world that's business as usual. But as the Institute of Internal Communication declared earlier this year, business as usual is dead.

Trauma-informed communication works with people's nervous systems instead of against them. In practice that's clarity over volume, predictable rhythms people can count on, honest context about what's uncertain and what's within a person's control, real choice in how they receive information, and feedback that actually moves a decision.

Adopting this as a communication strategy doesn't mean that your communications team is now a bunch of therapists — we don't, and shouldn't, discuss any one person's experience of trauma. It also doesn't mean we need to be gentler or soften a message. But you can say, "We understand that there's a lot going on right now. Here's how we're going to help you understand what we're focused on as a company and support you during this time."

That's not soft. That's business continuity.

🔎 My take: This is a pet peeve, and I’m glad Ellen hit upon it: for internal comms particularly, predictable rhythms are often overlooked. Unexpected communications are an interruption, and we should be careful to avoid that.

⚙️ You've built a practice around how people take in information under stress. How does AI actually show up in your own day-to-day work?

💬 Ellen's take: It's definitely evolved. Today, I use AI for lower-lift emails, for light editing, and for fact-checking. Also a lot of automation through Cowork — I worked with someone to get a set of plug-ins built and connected to Notion, Kit, and other tools in my stack.

I don't really use it for writing. The more I use AI, the easier it's become to detect when something sounds like it's been written by AI. I don't like sounding like anybody else, so at this point my content is 100% original, with an occasional lift from Claude for repurposing to social channels.

So, for example, earlier this year I launched a research report, Shifting Ground: Internal Communications in an Age of Polycrisis. I had completed all the primary research myself, through a survey with a few dozen communicators and one-hour follow-up interviews with about a third of those people.

After anonymizing the data, I had Claude help me identify a few major themes from the open text answers. From there, I wanted to cross-check and validate this against secondary research. I pulled about a dozen reports. I used NotebookLM to review those reports, searching specifically for data that spoke to my research — either bolster it or illustrate weaknesses.

From there I worked with a copywriter to shape my findings and the secondary research into a narrative. That was the report. This freed me up to begin working with the consultant I hired to build the workflows. We connected Claude to Kit and created a month-long email campaign to support the report. We gave Claude the report, and it drafted a series of emails that dig deeper into key aspects of the report. Claude then created the individual emails and campaign sequence in Kit, where I reviewed everything and made final edits.

🔎 My take: Longtime readers know I’m a big believer in “AI for operations” over “AI for writing,” too — glad to see another voice in that choir! That said, if you’re using it effectively as a writing tool, keep going.

📣 A lot of communicators are being handed an AI rollout to announce right now. What are we walking into?

💬 Ellen's take: Early in my career, at a newspaper, we were told overnight we were moving to a content management system. Everything digital — no more red pen. I remember the whispered conversations, the confusion moving through the newsroom. This was at the beginning of Web 2.0. And I remember that no one stopped to listen, to answer honestly, or to bring us into shaping any of it. We were just told to adopt.

Now multiply that by the most complex digital transformation of our lifetimes, dropped onto people who never recovered from the pandemic. Nobody knows if the bubble will burst, if AI dulls our thinking, if it's going to replace us.

So employees are trying to survive, and we're forcing new tools on them with no psychological safety, inside trust that's already ruptured, without agency, without two-way voice, without even real peer support for learning the thing.

🔎 My take: Every communicator I know has a version of the newsroom story. The line I'd sit with is "no one stopped to listen." Listen, assess, build, reinforce. That’s a good model for change, but we often skip to step three (and, as an aside, skip step four, too).

🛠 So what does a trauma-informed AI rollout actually look like — the comms plan, not the philosophy?

💬 Ellen's take: A trauma-informed approach helps people move through uncertainty by prioritizing things like safety — psychological safety, like whether experimentation is encouraged, yes, but also recognizing that people arrive in this moment in a variety of states. Excited, nervous, fearful. And that it's all okay. There's no right way.

But if you're trying to move someone through AI adoption, you do need them to participate. So what choice can you give someone? Can you ask them to identify a point where they believe it can help? Can you team them up with a peer to collaborate? Can you have them give feedback on where it's helping, and where it's a pain in the neck? And can you commit to communicating updates about the adoption on a regular cadence, even when there isn't much of an update?

This could look like a weekly update on your intranet with a recurring segment on your monthly all hands. Signpost what was discussed previously — "Last time, we shared three key areas of focus for our AI transformation. Today, we're going to dig in deeper to how we're doing against our goals" — and then tell people what's ahead.

The point is to pick a cadence, stick to it, and remind people of it. All too often we know the cadence, but nobody else does.

🔎 My take: I’d really love to hear from others here — how is your company rolling out AI? As above, I’m a big fan of the regular, weekly cadence. (It’s no accident that The Comms Stack hits your inbox at the same time each week!) But if you’re finding something else that works, let me know?

⚖️ Where can AI genuinely reduce harm in internal comms, and where does it risk eroding trust or retraumatizing people?

💬 Ellen's take: As communicators, we often use personas, which can be helpful for message testing. AI is great for this — and I recommend that no persona is ever assumed to be at 100% capacity, because no human ever is. How does this land for the senior manager with little kids and aging parents, the sandwich generation running on fumes? How does it land for someone six weeks out from a layoff, reading every announcement for the next shoe to drop?

AI can also check our language for plain, human clarity. It can flag ableist phrasing and recommend inclusive language. It can be a tool — so many of our leaders are already cross-checking against it — for introducing empathy before we hit send. That's real harm reduction.

But the danger sits in two places. One is the adoption itself — roll AI out carelessly, on top of ruptured trust, and you can retraumatize the exact people you're trying to reach. We hear so much about how "AI won't replace you — but someone who knows how to use AI will." Well, that's still telling people they're going to be replaced. That there is no choice but to adapt. People who experience the feeling of powerlessness in relation to trauma can become activated in a situation where they feel there is no choice.

The other harm is in the bias that exists in the training for these models — we're seeing that play out in real time in things like recruiting platforms and technology, but it's something we need to be aware of anytime we're using AI.

And there's a line I don't think we should ever cross: in high-stakes moments, and virtually everything is high-stakes these days, you don't outsource the communication itself to AI. In a world this complex and ambiguous, people need to hear from their actual leaders more than ever.

AI isn't living in the world. Humans are. And every time we ignore the humanity of our employees, we risk introducing harm.

🔎 My take: The last couple sentences above are key. For example, I shy away from using the phrase “human in the loop” because we — not the LLMs — are supposed to be the beneficiary and captains of AI powered tools, not just role players.

Thanks to Ellen for the insight. You can find her at Equilibrious Communications or on LinkedIn, and her research report, Shifting Ground, goes deeper on everything above.

5,000+ professionals use Skill Leap to get ahead with AI. Right now, two of their best courses are completely free - Claude 101 and the 14-Day AI Boot Camp.

Claude 101 covers prompting frameworks, Artifacts, file analysis, and real-world workflows in 19 lessons.

The Boot Camp covers ChatGPT, Gemini, Midjourney, and prompt engineering in 16 lessons. Downloadable workbooks. LinkedIn certificate.

Zero cost, no credit card, no catch.

Get 2 Free AI Courses

Axios Communicators reported last week that Reddit averaged 3.83% of ChatGPT Search citations between July 18 and Aug. 7, according to Promptwatch, an AI visibility vendor — and by mid-August it was under 1%, averaging 0.52%. That's an 86% drop in roughly a week. For scale: when the same firm compared social platforms last September, Reddit was the most-cited network in ChatGPT at 3.2%, with LinkedIn second at 0.3%. Reddit wasn't one source among many. It was the source. Promptwatch's co-founder and CTO warns that a data-collection issue can't be ruled out and to treat the size of the drop as provisional, but it's a huge shift and a warning sign. It’s also a reminder: don’t put all your GEO eggs in one basket.

Steve Rubel, EVP for media insights and measurement at Burson, gave Axios the right answer to the whole fantasy of one channel that cracks GEO: "It doesn't exist. It's a long game, and [brands] need to be engaged on many different channels." Every GEO tactic that gets popular enough to reach your plan has already been priced in — you learn a source is winning, you spend a quarter building for it, and the model reweights while you're still in legal review. Chasing the source that's winning this month is often the GEO equivalent of buying high, and the durable version of this work is the boring version: be findable, be accurate, and be where your audience actually is. Or, just do the communications work you’ve already been doing, but do it in ways where the machines can find you and won’t misread you.

Your next step: Re-run the same 10 brand prompts you ran three months ago — and read the sources it cites, not just the answers. See what moved, and then check to see if it mattered. If you had a good, multichannel comms strategy, there may be a shift in sources, but the answers are probably still right. That’s something worth sharing with leadership.

Last week, I shared Gab Ferree's take on Claude's watermark — that the answer isn't panic, it's writing down how you actually use AI. The obvious next question is what goes in that document. She's running a free hour on exactly that.

"Make AI Your Comms Competitive Advantage" is a one-hour interactive session, free, built for comms people from manager to CCO, agency or in-house. You leave having drafted three things: a personal AI operating doc (what you do and don't use AI for, plus the one-line disclosure you can put in your email signature this week), a shared set of rules so your team stops guessing, and the brand-facing version of your company's AI philosophy. There's a segment on what the Claude watermark changes for you — and the part most sessions like this skip, which is the language that gets legal and leadership to say yes instead of "let's table it."

The date on the page right now is Thursday, Sept. 24 at 4:00 PM ET, with other dates available if that one doesn't work.

Full disclosure, since that was rather the point of last week's piece: I'm a paying Off the Record member, and an affiliate on the membership side. On this one I get nothing — Gab offered to set up a commission and I told her not to. I'm passing it along because it's the session I'd have wanted right after writing that story, and it costs nothing to sit in.

Save your seat →

Leadership is asking: where is AI delivering value for us and where is it creating risk? Right now, most teams have no idea.

With Harmonic Security’s Usage Explorer, you get a complete picture of how your organization actually uses AI, automatically categorized into custom use cases with complete tool-level granularity.

See it in action

Looking for a role at the intersection of communications and AI? Here are some opportunities to check out:

Was this forwarded to you by a friend or colleague? Want to get this in your inbox next week? Click here to subscribe!

Quick question: how can I help?

Reply and tell me — I read every response.

Until next Wednesday,

Dan

Advertise | Subscribe