How do you build a LinkedIn outreach sequence using AI?

What we're building

The prompt generates a 3-message LinkedIn outreach sequence (Introduction, Conversation Starter, Soft Breakup) for leads who just accepted your connection request. Each message is personalised using the recipient's headline and summary, so it reads like it was hand-written.

The AI LinkedIn Profile Enricher reads the recipient's full profile directly, including headline, summary, current role, and past experience, which is what enables natural personalisation.

Recommended workflow

The engagement flow this prompt sits inside.

Step 1. Generate the messages (this prompt)

Run the prompt through the AI LinkedIn Profile Enricher, row by row. Test on a small batch (typically 10 leads) before deploying at scale.

Step 2. Push the leads into an outreach sequence

Connect with leads and schedule the three messages on a Day 0 / Day 3 to 6 / Day 10 to 13 cadence using the LinkedIn Outreach chained phantoms.

Recommended. Warm the leads first

Before the connection request goes out, run the leads through the LinkedIn Profile Visitor and Auto Follow Phantoms. Visiting and following puts your name in their notifications, which lifts connection acceptance and message reply rates once the sequence kicks in.

Prompt. Template (any user)

Fill in the YOUR CONTEXT block only. AI instructions stay the same. Set processing to row by row.

--- YOUR CONTEXT (fill in before running) ---
Your first name: [YOUR FIRST NAME]
Your role or team: [YOUR ROLE or TEAM]
Your company: [YOUR COMPANY NAME]
What you help with: [one short sentence, e.g. building lead gen systems for B2B teams]
Your expertise area: [e.g. lead generation, recruiting, sales automation, growth marketing, partnerships]
Content you share: [e.g. webinars, playbooks, podcasts, case studies, newsletters]
--- END ---

You are writing a 3-message LinkedIn outreach sequence for someone who just accepted my connection request. No connection note was sent, so this is the first real contact they will have.

Use the recipient's headline and summary as the main personalisation source for Message 2. Never invent details. Only reference what is visible in their profile.

Strip legal suffixes (Inc, LLC, Ltd, GmbH, SAS, BV, plc, Corp, Co., LLP, K.K., AB, etc.), non-Latin script translations shown in parentheses, and geographic or division suffixes from the recipient's company name before using it.

Universal rules
- Each message starts with a different warm greeting that includes the recipient's first name. Never repeat the same opener twice in the sequence
- Maximum 300 characters per message, hard limit
- Every message references the recipient's company name (cleaned) AND their job title or field
- Tone: friendly, conversational, peer to peer. Reads like a quick LinkedIn DM, not a formal email
- Use contractions. Not salesy. Frame [YOUR COMPANY NAME] as a place these topics get explored, not as a product
- Use commas at natural pauses and full stops at the end of every sentence. A message without proper punctuation is invalid
- Only commas and full stops allowed. No em dashes, ellipses, exclamation marks, emojis, brackets, parentheses, quotation marks, or question marks. Phrase questions as statements ending in a full stop
- Never use "I noticed", "Following up", "Just bumping", "Circling back", "Hope this finds you well"

Message 1. Introduction
Thank them for connecting. Introduce yourself as: "My name is [YOUR FIRST NAME], I'm part of [YOUR ROLE OR TEAM] at [YOUR COMPANY NAME]". Mention what you help with in one short sentence. Close with a warm line about being glad to connect with people in their field at their company. 2 to 3 sentences max. No question, no pitch, no ask.

NEVER use "I am [Title] at [Company]". That reads like a job description.

Message 2. Conversation Starter
Read the recipient's headline and summary together. Identify the most specific, interesting thing they do or focus on. Reference it naturally to show you have actually looked at their profile. Bridge into a topic in [YOUR EXPERTISE AREA]. Mention you go deep on these topics at [YOUR COMPANY NAME]. Offer to look at their current setup or share what you see working. End with an invitation phrased as a statement ending in a full stop.

If headline and summary are sparse or generic, fall back to their job title or field. Never invent specifics.

Message 3. Soft Breakup
Mention you regularly share [CONTENT YOU SHARE] about [YOUR EXPERTISE AREA]. Position it as useful for their field. Close with a warm two-part line: first invite them to connect and discuss if they are open to it, then follow with a sincere well-wish for their work or projects at their company. The well-wish must feel genuine, never dismissive. The phrases "no problem" and "no worries" are banned.

Output format
Return ONLY the three messages, formatted exactly:

MESSAGE 1:
[message text]

MESSAGE 2:
[message text]

MESSAGE 3:
[message text]

No explanations, headers, or commentary before or after. Each message under 300 characters and properly punctuated. Any other response is invalid.

Output

The Enricher appends a single column per row containing all three messages, formatted as a labeled block:

MESSAGE 1:
Hi Lukas, thanks for connecting. My name is Oumniya, I'm part of the Client Growth team at Phantombuster. I help companies build their lead gen systems. Always nice to meet sales leaders working on similar challenges.

MESSAGE 2:
Hey Lukas, your focus on enterprise sales motion at Link11 caught my eye. Curious how you are approaching outbound right now. We dive into this kind of work at Phantombuster, and I'd be glad to take a look at your stack with you.

MESSAGE 3:
Lukas, quick one. We share webinars and playbooks on building lead gen systems that produce ready-to-buy leads. Useful stuff for sales leaders. Happy to connect and dig in if you are open to it. Otherwise, wishing you all the best with what you are building at Link11.

The output is then split into three columns and queued in the LinkedIn Message Sender on a Day 0 / Day 3 to 6 / Day 10 to 13 cadence.

Input fields (from LinkedIn Profile Scraper output)