AI Agents for Internal Comms Teams: 75% Are Still Stuck in Ad Hoc AI

Internal communications teams were some of the first to pick up AI. Drafting an all-hands recap or tightening a CEO email is an obvious fit. But a year or two in, most of those teams are still using AI the same way they started.

That's the picture in Gallagher's 2026 Employee Communications Report, the latest edition of its long-running State of the Sector study. It draws on more than 1,300 communications and HR professionals in 40 countries, surveyed in November 2025, plus roundtables and focus groups. On AI, the summary is blunt: three-quarters of functions remain stalled in early-stage discussions or ad hoc AI experimentation.

Drafting is solved. The rest of the job isn't.

Gallagher found that nearly everyone is using AI for drafting and summarizing. Those tasks are "relatively simple, safe, and immediately helpful," the report says, "but while they can save time, they don't impact how the function operates."

The gap shows up in confidence. Only 36% of respondents say they have the skills, literacy, and critical thinking to use AI effectively. That jumps to 61% for teams Gallagher classes as AI-ready, meaning routine users with structure, support, and governance. In Gallagher's March release, 63% of HR and comms professionals said they were still experimenting, and just 5% reported optimized integration.

The report's summary line is worth repeating to any comms leader: AI "is not a shortcut to maturity; it simply amplifies whatever maturity already exists."

The problems AI hasn't touched yet

Look at what comms teams say is actually hard, and it isn't writing:

• Managers are the missing channel. 87% say manager capability is a risk to their goals, but only 21% give managers toolkits to cascade messages.
• Change keeps coming. Change is the most wanted skill, yet 61% have no formal change communication approach.
• One size fits nobody. Fewer than 1 in 5 (18%) are satisfied with their ability to personalize content, and 77% don't regularly use audience profiling like personas.
• Volume backfires. High-volume environments see a 16% spike in perceived burnout risk, and organizations that send a lot of messages see a 30% jump in rating leadership trust as a significant risk.

None of these are drafting problems. They're coordination, targeting, and follow-through problems. That's where an AI agent, one that can work across your tools rather than just answer a prompt, earns its place.

Governance is the unlock, not a brake

The strongest finding in the AI section is about structure. Teams with high governance, clear policies, dedicated training, and active champions are 10x more likely to reach Gallagher's "Enabled" maturity stage. Those in the most mature group were over 20x as likely to agree their leaders had a clear AI strategy.

Small teams are most exposed. 50% of organizations with fewer than 500 employees said they were experimenting with no formal oversight or governance. There's a people side too. Senior leaders tend to see AI as an opportunity, while support and administrative staff are twice as likely to see it as a threat. That matters for a function whose job is to explain change to everyone else.

Sign up for our newsletter

Get the latest updates, product launches, and helpful tips delivered straight to your inbox.

Email addressSubscribe

Where AI agents help internal comms first

Start where the report says the pain is, and keep a communicator approving every message that goes out:

• Manager cascade kits. After a big announcement, an agent turns the core message into a short talking-points brief, an FAQ, and a team-meeting script for each manager. Then it follows up to see who actually ran the conversation. That goes straight at the 87% vs. 21% gap.
• Audience versions. Draft frontline, desk-based, and regional versions of the same update, matched to the channels each group actually uses, instead of one email for everyone.
• Change playbooks on demand. Keep a "change comms in a box" kit (narrative, FAQ, timeline, checklist) and have an agent fill in a first version whenever a new change lands. Gallagher recommends building this kit itself.
• Listening and measurement. Summarize pulse-survey comments, intranet questions, and help-desk themes into a weekly "what employees are asking" report. Gallagher's own advice for mature teams is to "move from drafting to insights."
• Volume control. Track what's been sent to each audience this week and flag when a new message would pile onto an overloaded group.

Each of these keeps the human judgment (tone, timing, what leadership should say) with the comms team. The agent handles the versioning, chasing, and counting that otherwise gets dropped.

Put the guardrails in first

Given Gallagher's governance findings, the order matters. Before an agent touches employee communications, decide three things. Which sources it can draw from (approved messaging, not random Slack threads). Who signs off before anything is sent. And what it should never do, such as message employees directly about pay, performance, or layoffs. Teams that write this down are the ones Gallagher found pulling ahead.

What this means for comms teams going forward

Internal comms doesn't need AI to write faster. It already does. What it needs is help with the work that never fits in the day: equipping managers, tailoring messages, and showing leadership what landed. Gallagher's data suggests the teams that get there won't be the ones with the most AI tools. They'll be the ones with a clear plan for how AI fits, and agents doing the follow-through that people used to skip.