Almost every marketer has already asked ChatGPT to write an email. The result is structured but generic — it gets edited, sent, and forgotten.
AI is useful in email marketing, but for specific tasks: taking the routine work off your plate, speeding up a draft, generating subject line options. It can't replace a copywriter.
What AI does well
Writing a first draft. Writing that first paragraph is often the hardest part, and AI can do it in seconds. The draft will need work, but editing existing text is faster than starting from a blank page.
Subject line options. Ask it to generate 10 options — a good subject line is often found through trial and error rather than on the first try.
Simplifying complex text. Promotion terms, technical changes, legal details — AI can translate them into plain language. Upload the source text and ask for a simpler rewrite.
Adapting for segments. A finished email for new customers is easy to adapt for returning ones: the structure stays the same, while the tone and emphasis change.
Editing. AI catches repetition, awkward phrasing, and unnecessary words. It works fast, but leave the final review to yourself.
What AI does poorly
Brand voice. AI doesn't know how your company talks, so it writes generically: polite, structured, without character. Everything that makes an email recognizable is added by a human.
Specifics. AI doesn't know your numbers, case studies, or real customer stories. Without them, an email ends up generic and says nothing at all.
Campaign context. What was in the previous email, what funnel stage this is, what matters to this segment — all of this needs to be explained in the prompt. The less context you give, the worse the result.
Tone in long texts. On short tasks, AI holds the tone reasonably well. In long emails and sequences, it starts repeating itself and slipping into templated phrasing.
Trust AI with | Leave to a human |
|---|---|
Structural draft | Brand voice and tone |
Subject line options | Specific numbers and case studies |
Simplifying complex text | Final editing |
Adapting for segments | The emotional hook at the start |
Finding repetition and filler | Sequence strategy and logic |
How to write a prompt for an email
A bad prompt: "Write an email about our promotion."
A good prompt gives AI four things: context, audience, task, and constraints.
Context. What the product is, what the promotion is, what the situation is. The more detail, the more precise the result.
Audience. Who the email is for: new subscribers, returning customers, a cold list. This changes the tone and the arguments used.
Task. What the reader should do: buy, sign up, visit the website.
Constraints. Length, tone, and anything that shouldn't be mentioned.
Example: "Write an email for returning customers of an online clothing store. Promotion: 20% off the new collection, three days only. Goal: drive traffic to the website and a purchase. Tone: friendly, no corporate stiffness. Under 150 words. Don't use the word 'unique.'"
Workflow
Step 1. Brief. Write down: audience, goal, key message, CTA. Five minutes now saves an hour of rework later.
Step 2. Draft. Give the brief to AI and get a rough draft. Don't judge it as a finished piece of copy.
Step 3. Edit. Add your brand voice and specifics, and remove templated phrasing.
Step 4. Subject line. Ask AI to generate 7–10 options based on the finished text. Pick the best one or use it as a starting point.
Step 5. Review. Ask it to find repetition, clunky phrasing, and filler. The final call is yours.
Common mistakes
Sending the first draft without editing. Unedited AI text reads like AI text, and your audience can tell.
Expecting results from a single prompt. Good copy is an iterative process. The first prompt sets the direction; the following ones refine it.
Not fact-checking. AI states inaccuracies with total confidence. Check numbers, dates, and product claims by hand.
Using AI only for the copy. It can also help with subject lines, sequence structure, and adapting content for segments.
Letteros helps not just with email markup but with copy too. The editor has a built-in typography tool that inserts non-breaking spaces, fixes quotation marks, and removes stray characters. AI generation works separately: you can generate block text or a heading right inside the editor without switching to third-party services.