The honest case against AI email tools

If you're going to use an AI email tool, you should know the honest case against using one. We sell an AI email tool. We still think this is the right conversation to have publicly.

Case 1: Voice flattening

The most legitimate concern. Heavy AI use produces emails that converge on a mean. Even with brand voice extraction and careful editing, AI-assisted writers tend to drift toward AI patterns over time. The voice that distinguished your emails from competitors slowly homogenizes.

This is real. We see it in our own scoring data. Senders who use AI tools heavily for 6+ months show meaningfully more pattern-similarity to other AI-heavy senders than to their own pre-AI sends.

The mitigation: AI as draft starter, not draft finisher. Use the tool to generate options; edit heavily; keep the structural moves but rewrite the language. The senders who avoid voice flattening don't accept AI output without substantial revision.

The honest read: If your edge is voice (creator, founder, narrowly-positioned brand), AI assistance comes with real cost. Use sparingly, edit hard.

Case 2: Calibration drift

AI scoring is calibrated to a rubric. The rubric is good but it's not perfect. Heavy reliance on AI scoring can train writers to write to the rubric instead of to the audience. The result: emails that score well and convert poorly, because the rubric and the audience don't perfectly overlap.

This is a smaller problem than voice flattening but it's real. Our scoring rubric is empirically calibrated, but it's still a rubric. Your specific list might respond to patterns the rubric doesn't capture (or vice versa).

The mitigation: Score with the rubric; verify against your actual conversion data. Where they disagree, trust your data. If a low-rubric-scoring email converts well on your list, the rubric isn't fully capturing your audience's preferences.

The honest read: AI scoring is a useful shortcut, not a substitute for measuring your own conversions.

Case 3: Time saved on the wrong thing

AI tools save time on writing. They don't save time on thinking. Most teams that struggle with email aren't struggling because the writing is hard; they're struggling because they haven't thought clearly about who the email is for, what action they want, and why this is the moment.

If AI tools shortcut the writing without shortcutting the thinking, the result is more bad emails faster. The volume goes up; the quality stays flat or drops.

The mitigation: Use AI for drafting after you've done the thinking. If you can't articulate (in one sentence) what action you want and why this is the moment, don't draft yet — think more first.

The honest read: Tools amplify whatever process they're embedded in. If your email process is sloppy, AI makes the sloppiness faster.

Where AI tools still help

After all that, the cases where we genuinely think AI tools earn their place:

Volume scaling. Teams that need to ship 10+ emails per week (lifecycle programs, multi-product DTC, agencies running many clients) face a writing-capacity wall. AI tools push that wall further out, which lets the team ship sequences that would otherwise be deferred.

Variant generation. Even strong writers struggle to generate 5 meaningfully-different versions of the same email. AI tools generate variants quickly; the human picks the best one and edits. The pick-and-edit workflow lifts quality without homogenizing voice.

Scoring as a feedback loop. Local heuristics (spam words, all-caps ratio, readability) plus AI passes (clarity, CTA strength, persuasion structure) give a writer feedback that no individual editor would have time to provide on every draft. Used carefully, this lifts quality. Used as a substitute for human judgment, it drifts.

Sequence coherence. Holding 4-7 emails in your head while checking they reference each other coherently is hard. AI tools do this trivially. Sequence coherence checking is one of the cases where AI strictly dominates human-only workflows.

A/B winner prediction for small lists. Lists too small for statistical significance still benefit from calibrated rubric-based winner prediction. Our predicted A/B winner tool is one of the cases where small-list senders get value they couldn't otherwise.

The honest summary

If your edge is voice or thoughtfulness, use AI sparingly and edit hard. If your edge is volume, scale, or systematic quality, AI tools amplify your work and earn their place.

The wrong way to use AI tools: as a substitute for thinking. The right way: as a way to ship the thinking faster, with feedback you wouldn't otherwise get.

We sell the tool. We still think this conversation is worth having out loud.

For the full rubric we use to score emails, our scoring guide walks through it. And if you want to see what scoring tells you about your own sends, try our editor — anonymous, no signup. The detector also flags AI-flavored writing in your own sends, including human-written ones.