• by Ray Schultz , Columnist, Yesterday

Here’s some advice for anyone thinking of using AI tools to write emails: Don’t do it.  

Generic AI output are inauthentic and lack credibility, "Word.now" found in an analysis of behavior based on AI-drafted emails.  

The analysis, drawn from user feedback and behavioral observations gathered during the platform’s development, identified a pattern the company describes as AI voice homogenization: the tendency of mainstream AI email tools to produce replies that read identically regardless of sender, industry or relationship context. 

This uniformity is a result of the fact that the same language models are producing similar phrasing, sentence structure and cadence for millions of users.

The study focused on professional correspondence, but can also be applied to marketing and sales emails.

 “The AI email generator market has grown quickly, but most tools optimize for plausibility, not authenticity,” said Marcus Chen, co-founder and CEO of Word.now. “When every professional using AI sounds the same, the differentiation that signals genuine engagement disappears entirely. Recipients notice — and they respond accordingly.” 

This analysis is supported by a plethora of other studies. For instance Validity recently found that 38% of consumers are not likely to trust marketing emails that they know were written by AI. 

“This is a problem,” Validity writes. “When AI enters the mix , traditional levers like messaging, positioning, and even channel optimization can be filtered, reframed, or bypassed entirely. Marketers aren’t just competing for attention anymore—they ’re competing for interpretation by AI systems they don’t control.”

As for Word.now, it is not researching these issues simply out of curiosity. The company offers a free AI email reply generator that senders can use without creating an account or handing over their email address, the company says.  

This tool provides an optional identity layer that Word.now says allows senders to submit writing samples that trains a personalized reply model to produce content in their own voice. 

These style-matched replies use the sender’s vocabulary, sentence length and tonal register, the company adds. 

One concern that deters adoption of inbox-connected tools is privacy. Tools that require inbox access and/or produce uniform replies are facing growing resistance. 

As for now, mailers appreciate the help from email writing tools like Word.now’s.

“I noticed my response rates dropping after I started relying on a generic AI tool,” said Jordan Alvarez, a consultant at a mid-sized financial advisory firm who participated in Word.now’s early access program. “Once I switched to something that actually learned how I write, the conversations felt natural again. Clients stopped asking if I was using a bot.”