The old approach to professional communication was slow and inconsistent. You wrote the email yourself, it took longer than it should, the quality varied depending on how tired you were, and a significant portion of what got sent was mediocre. AI tools fixed all of that. In 2026, anyone can produce a grammatically clean, appropriately toned, professionally structured piece of communication in about forty seconds.

Which is why everyone’s inbox now reads the same.

Consider what this looks like in practice. A founder pitches a Series A investor with a cold email that opens with three polished sentences about market opportunity and ends with a call to action so smooth it could have come from a template library, because it basically did. A university student submits a job application cover letter that is technically flawless and completely indistinguishable from the forty other applications in the pile. A corporate communications team launches a new campaign with copy that has been run through three AI tools and one human approval and arrives in the market sounding like it was written by nobody in particular for nobody specific.

None of these people made a bad decision. They made the rational one, given what they were taught. Get the thing done. Make it professional. Move on.

The problem is that professional communication was never just about producing something adequate. It was about signalling something real: that there is an actual person here, with an actual perspective, who actually thought about you specifically before sending this.

The trust gap the research is confirming

A 2025 global study by the University of Melbourne and KPMG, surveying over 48,000 people across 47 countries, found that while 66% of people now use AI regularly, fewer than half actually trust it. That number has moved in the wrong direction as adoption has increased. The more people encounter AI generated communication, the more sensitive they become to it, and the less they trust what they are reading.

A 2025 study by Peter Cardon and Anthony Coman, reported by the University of Florida, put a number on it. Shown the same congratulatory email from a supervisor, 83% of 1,100 professionals rated the sender sincere when AI had helped lightly, and only 40 to 52% when AI had written most of it. Readers forgive the tool. They do not forgive the absence of the person.

The inbox your AI written email arrives in is the problem. Microsoft’s Work Trend Index 2025 measured 117 emails per worker per day, most skimmed in under 60 seconds, plus 153 Teams messages a weekday. An email that reads like every other email is skimmed with the rest.

Researchers at the California Management Review, in a December 2025 analysis of 4,951 publications on authenticity, identified what they called the Layer Coherence Triad: credibility, transparency, and reputation working together as the conditions under which professional communication actually builds trust. Their analysis found this triad appears in fewer than 9% of cases, but when it does, positive authenticity outcomes follow 82% of the time.

AI is very good at producing the surface layer of that triad. Clean, credible sounding, appropriately transparent. What it cannot produce is the third element, reputation, because reputation is the accumulated evidence of consistent, specific, human behaviour over time. No draft tool can fake that. Readers feel its absence even when they cannot name what is missing.

What the smarter operators are actually doing

The entrepreneurs I watch who communicate well in 2026 are not the ones who have stopped using AI. They are the ones who have a clear and deliberate policy about where it belongs in their process.

Writing the policy down matters more than the details. Most teams I work with have an unspoken one, and unspoken policies drift toward whatever is fastest on a busy Thursday. A table like this takes ten minutes to agree on, and it changes the question from whether someone used AI to whether the parts that carry the relationship were written by a person. That is the question readers are actually asking, even when they never say it out loud.

A founder I know uses AI to produce a rough first draft of any pitch email, then rewrites the opening two sentences completely from scratch, adds one specific detail about the person she is writing to, and cuts anything that sounds like it came from a template. Total additional time: about four minutes. The result reads like her. The AI did the structural scaffolding. She built the room.

A communications team at a midsize consumer goods company stopped running campaign copy through AI entirely after a competitor launched a near identical campaign in the same month. Same rhythm, same structure, same hedged benefit language. Both had used the same tools. Neither could explain why their work felt generic, because neither had anyone in the room asking the harder question: what do we actually believe about this product that nobody else would say?

A student in one of my courses rewrote her LinkedIn profile summary three times with AI assistance, each version more polished than the last, each one making her sound less like herself. The version that got a response from a recruiter was the fourth one, written without AI, in the room, under pressure, in about twenty minutes. It had two sentences that were slightly rough. It also had a specific detail about a project failure that no algorithm would have suggested she include.

Research published in the Journal of Marketing and Social Research in 2025 found that AI generated content significantly reduces perceived authenticity and brand trust compared to content made by humans, and that explicitly disclosing AI authorship makes the trust penalty worse, not better. The instinct to be transparent about AI use, while ethically correct, does not solve the underlying problem. The underlying problem is that the communication did not carry a human point of view.

A Worked Example: The Delay Email

Here is the table in use. Park Hyejin is an account manager at a cosmetics manufacturer in Seongnam that makes products for overseas brands. A problem on the filling line has pushed a shipment for a US retail client back by two weeks. She has to tell the client’s buyer, someone she has worked with for three years, and she asked an AI tool for a first draft. The draft was polite, fluent, and could have been sent by any supplier in the world.

The AI draft followed every rule of a professional apology and broke the only one that matters: it did not tell the buyer what happens next. Hyejin’s version is shorter and a little blunter, and it says “That is on us,” which no template will volunteer. It also contains an offer she had to get approved by her own manager before she could write it. That is the part a tool cannot draft, because it did not exist until a person went and made it happen. The tool was still useful. It caught a tense error and suggested a cleaner subject line. It just was not allowed near the apology.

This matters more now than it did a year ago, because readers are getting more suspicious, not less. In the 2025 Intelligent Information Society User Panel Survey, released by the Korea Media and Communications Commission with KISDI, 38.9% of respondents had used generative AI, and concern about the spread of false information rose 9.6 percentage points on the previous year, with concern about content that is hard to tell from the real thing up 8.9 points (방송미디어통신위원회, 28 May 2026). In October 2026 the same commission began a public campaign asking people to check what generative AI produces before they trust it (방송미디어통신위원회, 1 October 2026). Korean readers are being told by their own government to check. Overseas buyers are learning the same habit on their own. Write the email that survives the check.

Building your own version takes one meeting. Pull five recent emails from the team that mattered: a price change, a complaint, a delay, a thank you, a no. For each one, mark the sentences a reader would remember a week later. Those sentences are the ones that stay human, and everything around them is fair game for the tool. Then write the result on one page and pin it where people draft. In my experience the hardest row to agree on is the apology, because it is the one where AI saves the most time and costs the most trust. Teams that settle that row first tend to settle the rest quickly.

Before you send anything that carries a relationship, decide which parts you are handing off with The Offload Line. The opening line is where most AI drafts give themselves away, which is why Stop Starting Emails with I Hope This Finds You Well is worth a reread, and the idea of keeping your word that sits under every good apology is in 체면 (Chemyeon) in Korean Professional Culture. The thinking behind all of it is in The Prompt Is Not the Skill.

The Korean concept that frames this better than most Western thinking does

In Korean professional culture, there is a concept called 진정성 (jinjeongseong): authenticity, but specifically the kind that is earned through consistent behaviour over time rather than through self presentation alone. It is not about being unfiltered or raw. It is about communicating in a way that reflects real values and real experience in a way that has been demonstrated repeatedly, not performed once.

진정성 is what distinguishes the professional who has used AI to become a more efficient version of themselves from the one who has used AI to replace the version of themselves that was doing the work. The first person’s communication, AI assisted or not, still carries their actual perspective. The second person’s communication is technically competent and professionally weightless.

The old thinking was that communication was a support function, a means of transmitting information, and that efficiency was the highest virtue. The more progressive position, and the one that the research increasingly supports, is that communication is where professional identity lives. It is where trust is built or eroded. It is the evidence, accumulated over years, of who someone actually is.

AI can draft. AI can edit. AI can help you move faster. What it cannot do is build that evidence for you.

→ This is one of the topics I speak on for corporate teams and at industry events: the intersection of AI tools and authentic professional communication. If you are looking for a keynote or workshop that meets your teams exactly where they are right now, the Work With Us page has more on speaking engagements.

If you want to check whether your written communication is as clear as you think it is, the Copy Checker is a quick diagnostic.

---