Email is where AI shines — structured enough to automate, nuanced enough that generic output fails. These prompts produce emails that sound like you, respect the reader's time, and actually get replies.
Replace [bracketed placeholders] with your specifics before sending to ChatGPT
After the first output, iterate with 'Make this 30% shorter' or 'Adjust for a more casual tone'
Always add your own voice — AI-generated emails sound generic without a personal touch
Save email variants that work — build your own library of winning prompts
For cold outreach, personalize with 1-2 specific references to the recipient beyond what the prompt generates
Subject lines matter more than body copy — ask ChatGPT to generate 5 subject lines and A/B test the top 2
The best emails are under 100 words — resist the urge to explain everything
Cold emails should pass the 'would a stranger reply?' test, not the 'is this professional?' test
Read the email aloud before sending — if it sounds formal/stiff, it needs a rewrite
Use Claude for emails requiring nuance; ChatGPT for volume and standard formats
Only if you leave it generic. Good AI emails use your voice, specific context, and personal touches that no AI can fully supply. The fingerprints of AI email are: overly formal language, vague platitudes, awkward transitions, and 'I hope this email finds you well' openings. Remove those and your AI-assisted emails are indistinguishable from fully human ones.
Yes, but be careful. Tools like Instantly, Smartlead, or Apollo let you bulk-send AI-personalized emails, but regulators and inbox providers are cracking down on spammy patterns. Personalization at scale works best when you vary 2-3 elements per email (subject line, opener, CTA) and warm up sending domains properly.
Claude produces the most natural email voice and is my default for important messages. ChatGPT is faster for high-volume or structured emails. For sales outreach at scale, tools like Apollo and Clay generate emails via AI with better personalization than raw ChatGPT because they pull in real recipient context.
Provide samples of your actual emails in the prompt, or build a Custom GPT with your email examples preloaded. Specify tone traits in the prompt: 'conversational, direct, uses contractions, occasional humor, starts with questions not niceties.' The more you train the AI on your voice, the less you'll edit.