AI can help you write personalised cold emails faster, but it can just as easily produce messages that feel generic, awkward or obviously machine-written. The difference usually comes down to workflow: what data you start from, which research signals you feed in, how you prompt, and how carefully a human reviews the result. This guide walks through a practical process you can use with any AI writing tool.
Short answer
To personalise cold outreach with AI without sounding robotic, start from verified contact data, give the AI one or two specific and relevant research signals per prospect, use tight prompt patterns that ask for short, plain first lines, and have a person review every message before it is sent. Use AI to draft and speed up research, not to replace judgement. A contact finder such as PeakAI can handle the data step by helping you find verified emails and phone numbers from LinkedIn before any writing begins.
Step 1: Start from verified contact data
Personalisation is wasted if the email never reaches the right person. Before you write anything, make sure your list is built on accurate, up-to-date contact details.
- Confirm the person is still in the role. Job changes are common, and writing to someone about a role they left looks careless.
- Use verified email addresses. Sending to invalid addresses hurts your sender reputation and can push future emails into spam.
- Keep source context. Note where each contact came from, such as a LinkedIn profile, so you can refer back to it during research.
This is where a contact finding tool fits. PeakAI, for example, works alongside LinkedIn to help you find verified emails and phone numbers for the people you want to reach, so your AI workflow starts with a clean list rather than guesses.
Step 2: Research signals worth using
AI writes better first lines when it has something real to work with. Not every signal is equally useful, though. Focus on information that is recent, specific and relevant to the problem you solve.
Signals that usually work well
- Role and responsibilities: what the person is likely accountable for in their job.
- Recent company news: a product launch, new office, funding announcement or expansion into a new market.
- Hiring activity: open roles that hint at priorities, such as building out a sales or operations team.
- Content they have shared: a post, article or talk on a topic connected to your offer.
- Tools or processes they mention publicly: clues about how they work today.
Signals to avoid or use carefully
- Personal details unrelated to work, such as family or hobbies, which can feel intrusive.
- Very old news that is no longer relevant.
- Compliments with no connection to your reason for writing.
A good rule: pick one strong signal per prospect and connect it clearly to why you are reaching out.
Step 3: Prompt patterns for first lines
Vague prompts produce vague emails. Clear constraints help the AI stay short, specific and human. Here are a few patterns you can adapt.
The signal-to-relevance pattern
"Write one sentence, under 25 words, that references [signal] for [name], [role] at [company], and links it to [problem we help with]. Plain language, no flattery, no exclamation marks."
The observation pattern
"Based on this note: [research note], write a short, neutral observation a peer might make. Do not pitch. Do not use the words 'impressive', 'amazing' or 'I came across'."
The question pattern
"Write a single, specific question for [role] at [company] about how they handle [process], based on [signal]. Keep it conversational."
Prompting tips
- Give the AI your tone guide and a few examples of emails you like.
- Ask for several options and pick the best, rather than accepting the first draft.
- Ban phrases that make outreach sound automated, such as "I hope this email finds you well".
- Keep the AI focused on the first line and a short body; you can write the call to action yourself.
Step 4: Human review checklist
Every AI-drafted message should be read by a person before it goes out. Use a quick checklist:
- Is it accurate? Check names, company details and any fact the AI referenced. AI tools can invent details.
- Is it relevant? Does the opening connect to a real reason for reaching out?
- Does it sound like you? Read it aloud. If it sounds stiff or overly polished, simplify it.
- Is it short? Cold emails are usually easier to read when they fit on a phone screen without scrolling.
- Is there one clear ask? A single, low-effort next step tends to work better than several options.
- Would you be happy to receive it? If not, rewrite it.
Step 5: Deliverability basics
Even the best-written email needs to land in the inbox. A few basics matter regardless of which tools you use:
- Authenticate your domain with SPF, DKIM and DMARC records.
- Warm up new sending domains or mailboxes gradually rather than sending high volumes on day one.
- Keep lists clean by removing invalid addresses and people who ask not to be contacted.
- Avoid heavy formatting, many links and large images in first-touch emails.
- Vary your messages. Real personalisation naturally makes each email different, which also helps avoid looking like bulk mail.
- Follow the rules that apply where you and your prospects are based, and always include a simple way to opt out.
What not to automate
AI is useful for speed, but some parts of outreach are better left to people:
- Final send decisions. A human should approve who gets contacted and what they receive.
- Replies to interested prospects. Real conversations deserve real attention.
- Sensitive or complex accounts. Key accounts and senior executives usually warrant fully hand-written outreach.
- Claims about results. Never let AI invent case studies, figures or promises.
- Targeting strategy. Deciding which companies and roles to pursue is a judgement call based on your market.
A simple workflow to put it together
- Define your ideal customer profile and target roles.
- Build your prospect list and find verified contact details with a tool such as PeakAI.
- Collect one relevant research signal per prospect.
- Use a tested prompt pattern to draft first lines and short bodies.
- Review every message with the checklist above.
- Send in small, steady batches from an authenticated domain.
- Track replies, learn which signals and angles resonate, and refine your prompts.
FAQ
Can AI write my whole cold email?
It can draft one, but you will usually get better results by using AI for research summaries and first-line options, then editing the full message yourself.
How do I stop AI emails sounding generic?
Feed the AI a specific, relevant signal for each prospect, set clear limits on length and tone, ban stock phrases, and edit the output before sending.
Why does contact data matter for AI personalisation?
Personalised messages only help if they reach the right person. Verified emails and phone numbers reduce bounces, protect your sender reputation and make sure your effort is not wasted.
Where does PeakAI fit in an AI outreach workflow?
PeakAI handles the data step. It helps you find verified emails and phone numbers for prospects on LinkedIn, so you can then research, write and review your outreach with confidence that it will reach real decision makers.
Is it okay to use AI for follow-ups?
Yes, AI can help draft follow-ups, but keep them short, add something new in each one, and review them just as carefully as the first email.
