This doc explains my concerns about AI generative tools for creating content. In early August, I received an email update from the CEO of an async work platform. I don't use the platform, but as I SaaS-watch for fun — similar to bird-watching — I'm subscribed to their updates. What I read genuinely shocked me.

My Shock

The email proudly announced that the company had fully integrated AI into their content creation workflow. Every blog post, every help doc, every customer email — all generated by AI. The tone was celebratory, as if this were an unqualified win. I saw it differently: a business outsourcing its voice to a tool that doesn't understand its audience.

The Problem

AI-generated text often reads flat. It lacks the rhythm, nuance, and intentional structure that make human writing engaging. For SaaS companies — whose messaging must cut through information overload — this flatness is dangerous. When every sentence feels the same, readers disengage. They skim. They leave.

My Experiment

I ran a simple experiment: I took a piece of SaaS product copy and generated an AI version of the same message. Then I compared readability scores, sentence variety, and — most importantly — how each version felt to read. The AI version scored adequately on formal metrics but lacked variation in sentence length, used repetitive transition phrases, and avoided concrete examples. It was technically correct but emotionally flat.

Why Worry?

If businesses rely on AI-generated content without human oversight, they risk:

• Losing their distinct voice — AI converges toward the average, not the distinctive.
• Reducing reader trust — audiences can sense when text lacks genuine insight.
• Worsening information overload — more words, less meaning, same fatigue.

Solutions

AI tools can be useful, but they need a human in the loop. Here's what I recommend:

• Use AI for drafts and outlines, never for final copy without heavy editing.
• Vary sentence length intentionally — short sentences create impact; longer ones build context.
• Read everything aloud. If it sounds unnatural, rewrite it.
• Inject specific examples, data points, and stories that only a human who knows the product can provide.