AI for a Chief Communications Officer (CCO) at a $5M to $50M company covers 4 jobs: drafting press materials under a human editor, answering employee questions from approved documents, watching reputation, and setting rules for disclosing AI use. BCG’s 2026 survey found 68% of CCOs say their function lags on AI.
The tools are usually already in the building. What’s missing is the process: who approves what, which documents the AI may draw from, and what the company says in public about its own AI use.
So that’s what this page covers. It’s written for the comms lead at a mid-sized company who also owns PR, the intranet and the crisis plan.
Communications leaders use AI mostly for brainstorming, first drafts, editing and research. Muck Rack’s 2025 State of PR survey found 77% of PR professionals already use generative AI tools, yet BCG’s 2026 study found more than 70% of communications teams aren’t capturing task-level productivity gains.
So usage is high and payback is low. That gap usually comes from people using AI one prompt at a time, with no shared rules, no source library and no review step.
Muck Rack’s State of AI 2025 report broke usage down by task: brainstorming (82%), first drafts (72%), editing (70%) and research (59%). More than half of respondents said their company had no AI policies, standards or guidelines.
BCG puts the gain at 26% to 36% at the task level and 34% to 47% at the process level, once the groundwork is in place. The groundwork is the dull part: an approved tools list, a source folder and named approvers.
Here’s what that looks like for a 4-person communications team at a $25M company. The hours are planning assumptions from the kind of week I see at this size, so swap in your own baseline before you budget against them.
| Communications task | Hours/week now | With AI | Hours given back | Mechanism |
|---|---|---|---|---|
| Media monitoring and the daily coverage summary | 5 | 1.5 | 3.5 | Zapier or Make.com routes new mentions to Claude, which tags risk and writes the digest |
| First drafts of releases, statements and talking points | 6 | 3 | 3 | Claude or Microsoft Copilot drafts from a 5-line brief and the approved fact sheet |
| Reporter research before pitching | 5 | 3 | 2 | AI summarizes each reporter’s last 10 stories; a person writes the pitch |
| Internal newsletter, all-hands recaps and leader notes | 6 | 3 | 3 | AI drafts from meeting notes and a 3-line summary; the owner edits |
| Repeat employee questions (benefits, policies, dates) | 4 | 1.5 | 2.5 | An answer bot replies from an approved folder and cites the document |
| Coverage and measurement reports | 4 | 1.5 | 2.5 | AI drafts the commentary from exported numbers; a person checks every figure |
| Approvals, legal and executive review | 3 | 3 | 0 | No change: people sign off |
| Total | 33 | 16.5 | 16.5 |
16.5 hours a week is close to half of one full-time role given back to the team. Look at the last row, though: approvals stay exactly where they are, because that step protects the company’s name.
Demand generation, campaigns and SEO sit with marketing, and I cover them in AI for the Chief Marketing Officer.
Use AI for press releases and media relations as a researcher and first drafter, with a person owning every fact, quote and send. Cision’s 2025 State of the Media survey of more than 3,000 journalists found 72% worry about factual errors in AI-generated PR content, so every AI draft passes a fact check before it goes out.
Here’s the release workflow I set up.
The release itself still matters. Cision found 72% of journalists call press releases the most useful resource PR teams provide, so accuracy decides whether yours gets used.
The common advice is to let AI scale outreach: more reporters, more variations, more follow-ups. I recommend pointing it at research and sending fewer pitches to better-matched reporters, because Cision found 86% of journalists immediately reject a pitch that misses their beat or audience.
So give AI the reading. Have it summarize a reporter’s last 10 stories: what they cover, what they never cover, and which sources they quote.
Then a person writes the pitch. Muck Rack’s 2025 survey found 70% of PR pros usually or always tailor pitches already; AI’s job is to make that research fast enough that the team does it for every name on the list.
AI helps internal communications in 2 ways: an answer bot that responds to employee questions from approved documents, and AI-drafted summaries that shrink long updates into a few lines people read. Microsoft’s 2025 Work Trend Index found the average worker is interrupted every 2 minutes, about 275 times a day, by meetings, email and chat.
Every all-staff email you send joins that pile. Microsoft counted 117 emails and 153 Teams messages a day for the average worker, with most emails skimmed in under 60 seconds.
So the comms job is to send fewer messages, each one easier to act on.
Most comms teams answer the same questions every month: when open enrollment closes, where the expense policy lives, which days the office is closed. An answer bot built on Microsoft Copilot or Claude, limited to a folder of approved documents, can answer those in seconds and show the document it used.
Three rules keep it safe. It answers only from the approved folder, it cites a source for every answer, and it hands off to a named person when it can’t find one.
Someone owns the folder. When a policy changes, the old document comes out the same day, or the bot keeps quoting it.
Before any all-staff message goes out, AI drafts a 3-line version: what changed, what you need to do, and by when. If those lines cover it, send them and link the long version.
Here’s a scenario I use to plan this work. It’s illustrative, so treat the numbers as planning figures.
A 700-person services company gets about 200 routine employee questions a week on leave policies, benefits deadlines and onboarding steps. HR and communications set up an answer bot limited to approved, current policy documents, and every answer links to its source.
Anything involving personal circumstances, conflicting information or a policy exception goes to HR. Communications reviews the questions that keep coming back and rewrites the unclear guidance at the source.
Over a comparable 4-week period, routine questions needing a human reply fall from 200 to 140 a week, a 30% drop. The count covers email, service tickets and bot escalations, so a shift between channels can’t pass for a reduction.
Employees keep a direct route to HR. And the comms team gets something it rarely has: a running list of what employees don’t understand.
I’d judge the drop next to 3 other numbers: answer accuracy, repeat contacts and employee feedback. Fewer questions with more wrong answers is a failure.
For a reality check, Coca-Cola Andina built an HR assistant on Microsoft Copilot Studio, and Microsoft’s case study reports nearly 5,000 conversations and a 4.1 out of 5 satisfaction score from 976 employee surveys. That shows employees will use self-service HR answers; the 30% figure above is still an assumption.
AI monitors reputation by reading news, social and review mentions as they arrive, scoring each one against a written risk rubric, and alerting a named person when a mention scores high. The payoff is time: a team that hears about a problem within 15 minutes has options that a team hearing about it the next morning has already lost.
Muck Rack’s 2025 survey found 53% of PR professionals don’t use social listening tools at all. At $5M to $50M that’s common, and it’s fixable with tools you already have.
The rubric is the document that matters. Write it in plain words: a safety claim, a lawsuit, an executive named negatively, or a story from a top-20 outlet on your list is high risk.
In a crisis, AI can draft a holding statement in a minute. My rule is that it goes nowhere without 2 named approvers, usually the CEO and legal counsel, against a 60-minute target from alert to approved statement.
Prepare 5 to 8 pre-approved templates for your likeliest incidents: a data breach, a product recall, a workplace injury, an executive departure, a service outage. AI fills in the facts, and people make the call.
AI also misses tone. It doesn’t know the injured employee’s family is reading, so a person reads every line before it goes out.
Here’s a second planning scenario, also illustrative. A business services company runs a drill simulating a customer-facing service outage.
On the first run, the team takes 75 minutes from the first simulated signal to an approved holding statement. The time goes to escalating the issue, gathering facts and drafting from a blank page.
Before the next comparable drill, the team sets up monitoring alerts, a named approval chain and a pre-approved statement structure. An AI assistant groups incoming reports, links the evidence and drafts from confirmed facts only, with unverified details marked as open.
| Step to an approved holding statement | First drill | Second drill |
|---|---|---|
| Spot and escalate the signal | 20 minutes | 5 minutes |
| Gather confirmed facts and draft | 30 minutes | 10 minutes |
| Review and approve | 25 minutes | 15 minutes |
| Total | 75 minutes | 30 minutes |
The 45 minutes saved come from 4 changes working together: monitoring, preparation, clearer ownership and AI drafting. I wouldn’t credit AI alone for it.
The incident lead verifies the operational facts, and communications approves the statement before the simulated release.
Yes, by written rule: decide when your company labels AI use before a reporter, employee or regulator asks. PRSA’s updated AI ethics guidelines, released in fall 2025, call for human accountability and disclosure, and the EU AI Act’s Article 50 labeling duties began applying on 2 August 2026 for in-scope content.
Few companies have written that rule down. Muck Rack’s State of AI 2025 report found only 20% of agency PR professionals disclose all of their AI use.
Trust is the reason to settle it early. Edelman’s 2026 Trust Barometer found 37% of people name the growing use of generative AI platforms among the events most affecting trust over the past 5 years.
This is the rule I write with communications teams. It fits on one page.
| Tier | What AI did | Disclosure | Example |
|---|---|---|---|
| 1. Assist | Research, editing, or a first draft a person rewrote and approved | No public label; log it internally | A press release drafted from a brief, fact-checked and approved |
| 2. Author | Produced most of the published text, with light human edits | Label it, for example “Drafted with AI, reviewed by [name]” | An AI-written FAQ page or report summary |
| 3. Synthetic | Generated or altered images, audio or video, or imitated a real person | Always label; imitate a real person only with written consent | An AI image of a facility, a cloned executive voice |
Two rules sit on top of the tiers. Quotes attributed to a real person are approved word for word by that person, every time. And if a reader would feel misled on learning how something was made, label it.
If you publish to EU audiences, Article 50 can require labels on AI-generated text published to inform the public on matters of public interest, with an exception where a person reviews the text and someone holds editorial responsibility. I’m not a lawyer, so check your exposure with counsel.
Writing the full policy and approval map is the job of our AI policies and compliance work.
Measure AI ROI in communications against a baseline taken before the change: hours per task, minutes from alert to approved statement, and repeat employee questions per week. Value the hours at a loaded rate, such as the BLS full-time private-industry figure of $54.00 per hour in June 2026, then subtract licenses, setup and training.
Measurement is already a big part of the job. Muck Rack’s 2025 survey found 44% of PR professionals say measurement and reporting take at least 25% of their job, and 67% say measurable results are the best way to raise PR’s standing.
BCG found leading communications teams are 1.7 times more likely to show AI’s value to leadership. A baseline is how you join them.
Here’s the worked version, using the 16.5 hours from the table above. The first-year cost is an assumption, so put your own quote in its place.
The math on a 4-person comms team
Capacity turns into money only when you decide where the hours go. In comms, I’d send them to executive visibility, relationships with the 20 reporters who matter most, and manager toolkits for big internal changes.
BCG also found half of CCOs plan to redeploy or reduce headcount because of AI. I build for redeployment only: the hours go back to your team.
Track 4 numbers each month: hours given back, alert-to-statement time, repeat questions per week, and coverage in your top-20 outlets. You can run your own version in the AI ROI Calculator before you spend anything.
Ordered people first, then process, then tech. Each step names who does it and what it produces.
Days 1 to 30: people
Days 31 to 60: process
Days 61 to 90: tech
I’m Lee Pomerantz, Certified Chief AI Officer, founder of eMediaAI, Fort Wayne, Indiana. I ran a digital agency for 24 years before rebuilding the company around AI implementation, and every engagement runs in one order: people first, then process, then tech.
For a communications team, the AI Opportunity Blueprint™ is where that starts. It’s a 10-day, $5,000 forensic audit: your team records the tasks, I map where the hours and approvals go, and you get a prioritized workflow map, an ROI projection, an implementation roadmap and governance recommendations, including the disclosure rule.
If a task returns less than 20% of its time, the report says cut it. If the numbers don’t hold, it says don’t build.
AI is taking over parts of PR work, mainly first drafts, research and media monitoring, while judgment, relationships and approvals stay with people. USC Annenberg’s 2025 Global Communication Report found 68% of agency executives strongly agree people will remain essential to PR. In eMediaAI’s planning model, a 4-person communications team gets about 16.5 hours a week back, and those hours are redeployed to higher-judgment work.
PR professionals mostly use general-purpose AI assistants, such as Claude and Microsoft Copilot, for brainstorming, first drafts, editing and research. Muck Rack’s State of AI 2025 report found PR professionals use AI most for brainstorming (82%) and first drafts (72%). At a $5M to $50M company, licenses already in place usually cover the first workflow, with routing tools such as Zapier or Make.com added for media monitoring.
Journalists judge a press release on accuracy and fit, and they watch closely for AI errors. Cision’s 2025 State of the Media survey of more than 3,000 journalists found 72% are concerned about factual errors in AI-generated PR content, while 72% still rate press releases as the most useful resource PR teams provide. A human fact check on every number and quote answers that concern.
AI for corporate communications can start with licenses a company already pays for, such as Microsoft Copilot, plus training and a few hours of setup per workflow. BCG’s 2026 survey found more than 60% of communications leaders plan to spend less than 10% of their functional budget on AI. eMediaAI’s AI Opportunity Blueprint™ is a fixed $5,000 for a 10-day audit that prices each workflow first.
A first AI workflow in communications usually takes a few hours per person over 60 days. Each person screen-records their 3 most time-consuming tasks, joins a 2-hour training session, and reviews AI drafts during the first month. Writing the disclosure rule and crisis templates takes the comms lead about 2 working days, and the first workflow can go live within 90 days.
The biggest risks of AI in PR are invented facts or quotes published under the company’s name, confidential information pasted into unapproved tools, and undisclosed AI content that surfaces later. Three controls cover most of that exposure: a fact check on every number and quote, an approved tools list stating which data goes where, and a written disclosure rule. Muck Rack found more than half of PR professionals work without an AI policy.
A Chief Communications Officer’s first AI move is a one-page note telling the team what AI is for and where the returned hours will go. Next, name one owner for communications AI, write the approval map and disclosure rule, and baseline hours per task. Only then pick 1 workflow that gives back at least 20% of its time, such as the daily media monitoring summary.
The EU AI Act’s Article 50 requires labels on certain AI-generated content from 2 August 2026, including deepfakes and AI-generated text published to inform the public on matters of public interest. The text duty carries an exception where a person reviews the content and someone holds editorial responsibility. A US company publishing to EU audiences should confirm its exposure with legal counsel before relying on that exception.
Train a communications team on AI with a 2-hour session on tools the company already licenses, using the team’s own releases, newsletters and crisis templates as practice material. Cover 3 things: which data may go into which tool, how to brief AI from an approved fact sheet, and who signs off each kind of content. BCG’s 2026 study found 71% of leading teams had upskilled at least a quarter of their staff.
AI in communications can pay back in under 6 months on a first workflow when hours are baselined before the change. A 4-person team getting 16.5 hours a week back at the BLS full-time rate of $54.00 per hour over 48 weeks gains $42,768 a year in capacity, so a $12,000 first-year cost pays back in about 3.4 months.
Book a free 20-minute call to see if the AI Opportunity Blueprint™ is right for you.
Book Your Call
© 2026 eMediaAI.com. All rights reserved. Terms and Conditions | Privacy Policy
