Using AI in internal communications: 5 tips from our AI Masterclass 

88% of organizations are now using AI in some capacity, and internal communicators are among those who can benefit most from embracing it. But, as we’re all learning, using AI and using it well are very different things.  

Interact’s Director of Engagement, Greg Stortz, recently hosted an AI Masterclass for comms professionals, which received rave reviews from attendees and spread essential AI knowledge to communicators around the world. We’ve put together five practical tips that stood out, drawn straight from the session. 

What does AI adoption look like right now? 

Most organizations are using AI, but very few are seeing meaningful results from it. According to McKinsey’s November State of AI report, 88% of businesses are running AI in some capacity, but only one in three has started to scale it, and just 39% report seeing any real business impact. As Greg put it, most teams “are running on activities and not real outcomes.” 

Greg also touched on how the rise of AI is impacting the people comms teams support. Research from AMBS shows that 55% of senior leaders are worried about staying relevant as technology evolves, and 73% say their role has become more complex since 2020, with AI cited as the number one reason. Meanwhile, according to SHRM’s State of AI in HR 2026, 40% of senior decision-makers say learning to use AI effectively is their top priority right now. 

This data makes it clear that senior stakeholders are stretched, uncertain, and hungry for support. The comms professionals who can bridge that gap are becoming genuinely indispensable – and that’s on top of reaping the benefits of AI in their own day-to-day work. So what’s getting in the way? 

What are internal comms professionals actually experiencing with AI? 

We polled over 200 attendees during the AI Masterclass so that we could get a sense of how they’re using AI in their roles. First, we asked how they would describe their current use of AI at work. More than half the room said they were already using AI regularly, putting this audience noticeably ahead of broader benchmarks. It was clear that attendees are eager to get the most out of AI, even if some feel behind or unsure. 

How would you describe your current AI use at work?	Percentage of respondents
Using it regularly	56.0%
Just starting	28.7%
Feeling like I’m falling behind	11.6%
Not sure where to start	3.7%

Another question we asked dove a bit deeper. We probed into the audience’s biggest frustrations around AI:

When you use AI for work, what’s your biggest frustration?	Percentage of respondents
Results feel generic	39.0%
Outputs need too much editing	29.8%
I don’t trust the outputs	18.9%
I don’t know what to ask	12.3%

If you’ve spent any time using AI for internal comms, these frustrations are likely quite relatable. Nearly 70% of respondents reported wrestling with output quality: results that feel generic or need heavy editing before they’re usable. A further 18.9% don’t trust what comes back at all, which points to a deeper confidence gap around AI that goes well beyond any single technique.  

Put it all together and you’ve got a room of people, most of whom are actively using AI, want it to work better for them, and are running into very similar obstacles. That’s the gap Greg set out to bridge with this AI Masterclass. Read on for five great takeaways from the event, from tips for better prompts to clear use cases showing what AI can do for comms teams. For the full experience check out the free on-demand recording here. 

What are the best tips on using AI for internal communications? 

Here are five specific applications Greg walked through in the Masterclass, each with a sample prompt you can adapt straight away: 

Tip 1: The prompt is almost always the problem, not the tool 

It doesn’t really matter which AI tool you’re using. The output quality usually comes back to the prompt. Generic instructions produce generic outputs, and no amount of editing will save a draft that started from the wrong brief. 

The fix is context. The more relevant information you give the tool up front (think role, audience, tone, format, real examples to learn from) the less time you spend cleaning up what comes out. Greg demonstrated this live using two versions of the same request: 

	Vague prompt	Context-rich prompt
The prompt	“Write me an email to employees about our new intranet”	“You are an internal comms expert writing to skeptical middle managers at a manufacturing company — write a 150-word email announcing a new intranet, in a warm but direct tone, like this example: [example]”
What you get	Generic, corporate-sounding output. No reflection of your org, your voice, or your audience.	A first draft that’s ~80% there and needs one round of review, not five.
Time spent editing	Heavy: Most of the draft needs reworking	Light: The process is about refining rather than rewriting

The richer prompt specifies the role AI should take, the audience, the organizational context, the tone, the length, and a real example to work from. Add those elements and you’ll feel the difference immediately, with less editing required to get an output that sounds like your organization and speaks to your audience. 

Once you have a prompt that works, save it as a template. A library of great templates can give a small comms team (and let’s face it, that’s most comms teams) more power, and a large one more consistency. 

Another great mantra from the Masterclass: if you can describe it, you can build it. This applies to anything, whether it’s a campaign plans, newsletter template, or even code. Greg demonstrated this live, generating code for a working intranet widget in under five minutes from a single prompt, no developer required.  

Tip 2: Let AI pressure-test your comms before your employees do 

Every draft has blind spots, but getting a genuine second opinion before a sensitive message goes out isn’t always an option. AI as your “first reader” solves that, and it’s one of the most immediately useful applications Greg covered. 

As always with AI, setting the stage matters. Before you ask it to read anything, teach it about your organization. Feed it employee pulse survey results, engagement data, a summary of recent sentiment, and whatever else you have that will convey a sense of who your organization is. Ask it to confirm what it’s learned before moving on. Now it knows your workforce well enough to give you something more useful than a generic reaction. 

From there you can ask for three different reads of the same draft. An initial, general read, can answer basic questions: How will this land? What questions does it raise? What might you have missed? What’s the likely emotional reaction? After that, try segmented reads. Run it through the lens of your most engaged team, and then your most skeptical. Your detractors are going to read a reorg announcement in a very different way than your promoters will, and AI can point out how. You can also ask for a role-based read: How does this land for a frontline employee versus someone desk-based? Does it actually work for both? 

This approach is valuable, especially for small, rushed teams and diverse workforces. When timing is tight, it can be hard to put yourself in the shoes of every one of your colleagues, and with AI, you no longer have to. 

Tip 3: Build a leader voice profile and let AI ghostwrite the rest 

With the right inputs, AI can write in nearly anyone’s voice, and it’s better at it than most people expect. The mistake is trying to describe the voice in the abstract (“warm but authoritative, conversational but credible…”), which yields generic results. What does work is feeding it real examples and letting it figure the voice out for itself. 

Greg’s process is to find four to six examples of the leader’s past communications, such as old emails, blog posts, meeting transcripts, or even a voice note. Share them with AI and ask it to describe back to you how that person communicates: the rhythm of their sentences, the vocabulary they reach for, how they open and close, their relationship with the reader, and what they would never say. Ask it to confirm the voice profile before writing a single word. If anything seems off, correct it. 

Then give it the leader’s rough notes or bullet points (not a polished draft) and ask it to produce a first communication in that voice for a specific audience, landing a specific emotional tone. If all goes according to plan, you’ll end up with a first draft that sounds like the leader and needs very little refining. The best part about this approach is that the leader’s involvement is genuinely minimal, with a few bullet points and a final read-through, rather than the usual back-and-forth. 

Tip 4: Your survey results have a story, and AI can find it 

You run a survey. Results come back. Leadership wants a readout by end of day. And you’re staring at a spreadsheet of open-text responses wondering how to turn this into something coherent before the all-hands. Sound familiar? 

AI can help. Feed it your results, share your organization’s strategic priorities for the year, and ask it to find the three biggest themes, flag the biggest gaps between departments, and connect each finding to a business goal. This gives you a story arc that tells leadership what the data means, rather than just reporting back what it says. 

Greg was direct about why this matters beyond just efficiency. Surveys are a crucial employee listening touchpoint that let employees know that their opinions matter. When employees hear nothing back after a survey, they feel less heard, and they’re far less likely to participate in the next one. AI-assisted analysis means you can share results with leaders and employees faster and start taking action based on them, with more precision than manual review alone allows. 

Tip 5: Build your crisis comms plan before the crisis hits 

Most comms teams know they should have a crisis comms plan ready to go, but very few actually do in practice. That’s because almost nobody has time to build one until something has already gone wrong. AI can make preparation significantly more accessible. 

In the AI Masterclass, Greg shared a two-step approach for using AI to build a crisis comms plan:  

1. The brainstorm. Ask AI to identify the ten most likely crises for your organization, categorized by type (people and culture, operations, technology, reputation, external, etc.) with a likelihood rating for each. The more context you give it about your industry, workforce size, and known vulnerabilities, the more relevant your output will be. This step often surfaces scenarios you haven’t considered or thought through yet. 

2. The response plan. For your highest-priority scenarios, ask AI to generate a full communications strategy: immediate actions for the first two hours, key audiences in priority order, a core holding message, a draft all-staff from the CEO, three things not to say, and a 48-hour checklist of what good looks like. 

The output won’t be perfect first time, and that’s fine. After a first review, take it to your leadership team, get their input, and use AI to refine it. You’ll end up with an organization-specific playbook rather than a generic template. 

AI won’t do the strategic thinking, the relationship-building, or the judgment calls that are essential to great internal communications work. But for so many elements behind the scenes, like drafting, analysis, and planning, it’s already changing the function in meaningful ways. The five applications above are practical, accessible, and applicable right away. Try them out for yourself. 

Want to see AI in internal communications in action? 

While these five tips cover the highlights, the full masterclass goes a lot further. From building a live intranet widget from scratch to running a newsletter engine that anyone on the team can operate, Greg packed the session with practical demonstrations and a complete prompt library you can take straight into your next working week. 

Watch the full AI Masterclass on demand and walk away with everything you need to start using AI in internal communications more confidently, more creatively, and more strategically.