How to Use AI to Write Cold Emails That Get Replies

The problem with AI cold emails

Most people use ChatGPT like this:

"Write me a cold email to a restaurant owner about web design"

The result is a generic, over-formal email that reads like every other AI-generated pitch. Business owners delete it immediately.

The key is giving AI enough specific context that it can write something personalised and relevant.

What makes a cold email convert

Before using AI, understand the formula:

1. Specific observation — about their business specifically (not a generic compliment)

2. Relevant gap — what they're missing that matters to them

3. Proof — one result that shows you can fix it

4. Single CTA — one clear next step

Without all four, even AI can't save your email.

How to prompt AI for cold emails

Instead of a vague prompt, give AI all the data:

Bad prompt:

Write a cold email for a restaurant about web design

Good prompt:

Write a short cold email (under 80 words) for a web designer reaching out to "[Business Name]", an Italian restaurant in Shoreditch, London. They have 4.2 stars, 200+ reviews, and no website. The email should open with a specific observation, mention that their competitors have websites, reference one result (a restaurant in Hackney that went from 0 to 20 online bookings/week), and end with one question asking if a 15-minute call this week makes sense. Tone: direct, conversational, not salesy.

The difference: Specific business data, specific result, specific tone, specific length.

The Leadomos approach

Leadomos automates this at scale. For each lead found on Google Maps, it generates a personalised AI email using:

• Business name, type and location
• Their specific gap (no website, no social, low ratings)
• The service type you're offering

You don't need to write a prompt for each lead — the AI email comes pre-written with each contact. Just review, personalise slightly, and send.

Prompts for different niches

Web designer (no website):

Write a 70-word cold email from a web designer to [name], owner of [business name], a [business type] in [city]. They have no website. The email should note this specifically, mention that [competitor type] in their area all have websites, ask if they've thought about having one, and end with a soft CTA. Tone: friendly, concise.

Social media manager (no Instagram):

Write a 70-word cold email from a social media manager to [name] at [business name], a [business type] in [city]. They have no Instagram. The email should note this, mention that their type of business generates strong engagement on Instagram (with an example), and ask if a 15-minute call this week makes sense. Tone: direct, not pushy.

Photographer (hospitality):

Write a 70-word cold email from a photographer to [name] at [business name], a hotel in [city]. The email should note that their Google Maps photos look low-quality compared to similar hotels, mention that better photos improve booking conversion rates, and end with one question asking if they're open to a no-commitment quote. Tone: professional, direct.

After AI writes the draft

Always review and personalise:

1. Check for generic openers — delete anything like "I hope you're well"

2. Add one more specific detail — mention something you noticed about their actual listing

3. Read it aloud — does it sound like a human wrote it?

4. Cut it shorter — remove every unnecessary word

The goal is an email that feels like you spent 5 minutes researching them specifically — even if AI wrote the first draft in 5 seconds.

Follow-up emails with AI

Use AI for follow-ups too:

Write a short 3-sentence follow-up email to [business name] who didn't reply to my first email about web design. The follow-up should acknowledge they're busy, add one new data point (e.g. "67% of consumers say they've searched for a business online before visiting"), and end with the same CTA. Tone: light, not pushy.

Follow-ups consistently outperform first emails in reply rate. Don't skip them.