LinkedIn icebreakers from an audience in plain English

LinkedIn icebreakers from an audience in plain English

An n8n workflow for LinkedIn connection campaigns. You describe the people you want to reach in a sentence on a
password-protected form. OpenAI turns that into an Apollo people search, an Apify actor pulls the matching leads, and
OpenAI writes a short icebreaker for each one. The leads land in a Google Sheet, and the workflow then starts your
PhantomBuster agent, which sends the connection requests.

What is inside

• workflow.json: the workflow as n8n exports it, eight nodes and no sticky notes. It holds no credential and no key.
• README.md: these notes.

What we changed from the original copy

• Run Apify Actor & Get Results carried its author's live Apify token. It now reads Bearer <YOUR_APIFY_TOKEN>.
• Trigger PhantomBuster Agent carried its author's PhantomBuster key and agent. They are now YOUR_PHANTOMBUSTER_KEY and
  YOUR_AGENT_ID.
• Add to Google Sheet pointed at its author's sheet. It now asks for YOUR_SHEET_ID and writes to the first tab.
• The worked example in Personalize Outreach used what looked like a real person's LinkedIn profile. It now uses a
  made-up founder, Maya Torres, with the same shape.
• Generate Search URL named gpt-4.5-preview. OpenAI's deprecations page, read on 2 November 2026, lists it as shut
  down on 14 July 2025 with gpt-4.1 as the replacement, so the node now uses gpt-4.1.

Before you start

• An n8n instance, on n8n Cloud or your own server.
• An OpenAI API key.
• An Apify account and its API token. The workflow runs the public actor code_crafter/apollo-io-scraper, which bills per
  lead (Good to know).
• A Google account and a sheet for the leads.
• A PhantomBuster account with an agent that sends LinkedIn connection requests, and its API key.
• A user name and password for the form.

Set it up

1. In n8n, create a workflow, open its menu and choose Import from File. Pick workflow.json.
2. Open On form submission. Its authentication is Basic Auth, so add a Basic Auth credential with a user name and a
   strong password, and select it.
3. Add an OpenAI credential and select it in Generate Search URL and Personalize Outreach.
4. In Run Apify Actor & Get Results, replace <YOUR_APIFY_TOKEN> in the Authorization header with your token. The
   brackets go too.
5. Make a sheet whose first tab has these headers: id, first_name, last_name, name, linkedin_url, title,
   email_status, photo_url, icebreaker. Add a Google Sheets credential in Add to Google Sheet and replace
   YOUR_SHEET_ID with the ID from the sheet's address, the part between /d/ and /edit.
6. In PhantomBuster, set up the agent that sends the requests and point it at the same sheet, with the icebreaker
   column as its message. Copy the agent's ID from its address and your API key from the workspace settings.
7. In Trigger PhantomBuster Agent, replace YOUR_PHANTOMBUSTER_KEY in the header and YOUR_AGENT_ID in the body.
8. Save, open the form from its test address, send a description, then publish the workflow.

How a run goes

1. On form submission asks one question: describe your audience in plain English. It is required.
2. Generate Search URL has gpt-4.1 rewrite the answer as an Apollo people search address, built on the example URL
   in its prompt, and return it as searchUrl.
3. Run Apify Actor & Get Results runs the actor on that address and waits for its results. The body asks for up to
   500 records, with personal and work emails.
4. Limit keeps the first 3 leads.
5. Personalize Outreach gives gpt-4o each lead's name, city, current title and company, and two earlier jobs. It
   answers with one icebreaker that follows the template in its prompt.
6. Add to Google Sheet adds each lead, or updates it when its id is already there.
7. Aggregate waits for every lead, and Trigger PhantomBuster Agent launches your agent once.

Change the look

• How many leads go through: maxItems in Limit, 3 in the file. Raise it once the first runs look right.
• How many the actor fetches: totalRecords in the body of Run Apify Actor & Get Results.
• The icebreaker's voice: the template and rules in Personalize Outreach, and its worked example.
• What the search may change: the prompt of Generate Search URL lists the fields it may edit.
• What the sheet keeps: the column mapping of Add to Google Sheet.

Good to know

• Billing: the actor's Apify listing, read on 2 November 2026, charges $0.0015 for each lead. The body asks for 500
  while Limit uses 3, so a run can pay for many leads it throws away. Lower totalRecords to match.
• Emails: the actor is asked for personal and work emails, but the sheet has no column for them, only email_status.
  Add an email column to the mapping, or set both options to false.
• Short careers fail: the icebreaker prompt reads three entries of employment_history. A lead with fewer stops the run
  at Personalize Outreach.
• PhantomBuster gets only the agent's ID. The leads reach it only if the agent reads the sheet itself (step 6).
• The search prompt names organizationLocations[] as a field to change, while its example uses personLocations[], so
  a location can land on either the company or the person.
• Models: OpenAI's page lists gpt-4o and gpt-4.1 for shutdown on 14 January 2027, with gpt-6-luna as the
  replacement. Change both nodes before then.
• LinkedIn's User Agreement bans bots that add contacts or send messages, so automated requests can get an account
  restricted. Keep the daily volume in PhantomBuster low.
• Your data: the audience text goes to OpenAI, the search runs on Apify, and each lead's profile goes to OpenAI. Check the
  rules for contacting people where you and they are before you send anything.
• Your keys: the Apify token and the PhantomBuster key sit in headers, so take them out before you share the workflow.