Find warm introductions on LinkedIn with AI

The prompts run on your LinkedIn data archive, which you request under Me, Settings & Privacy, Data privacy, Get a copy of your data. They sit alongside the other AI prompts for your LinkedIn network. Background on live connections is in what a LinkedIn MCP is.

Where the path to someone shows up

LinkedIn defines your 2nd-degree connections as people connected to your 1st-degree connections, and the people in between are your possible introducers. Your data archive cannot list them, because LinkedIn's download page says it provides your own personal data and not that of other members.

LinkedIn's help page on viewing a connection's connections says to open the connection's profile and click the number of connections below their photo. Some people hide that list, and then the option is missing. For that case the page says you can still view shared connections from the People tab of the search results page. For one target person:

1. Type their name in the search bar at the top of LinkedIn.
2. Open the People tab of the results and look for shared connections, where LinkedIn shows them.
3. Copy those names for prompt 3.

1. Who you know at the company

Connections.csv in your archive has a Company and a Position column. It opens with a few lines of notes above the header. Some rows carry a connection date and nothing else, with no profile URL, so they cannot be matched to anyone.

Here is my LinkedIn Connections.csv. Skip the notes lines above the header.
List every connection whose Company is [company], or a close variant of the name.
Show name, position, profile URL and the date we connected, newest first.
Tell me how many rows have no company at all.

People change jobs, so open each profile before you ask anyone for an introduction. With LinkedAsk, the assistant searches the headlines of the connections LinkedAsk has seen instead, which name the employer in most cases.

2. Rank them by your last conversation

Prompt 2 needs the date of the last message between you and each person, sent by either side. messages.csv has the date, sender, content and profile URLs of each message, so it can match people from prompt 1 by URL.

For the people in that list, use messages.csv to find the date of the last message between us, sent by either of us.
Match by profile URL, never by name.
Rank them by that date, most recent first. Put anyone I have never messaged at the end, newest connection first.

Example answer (fictional people): "1. Priya Raman, last message in August. 2. Jonas Weber, last message in March. 3. Alice Moreau, no messages, connected in 2021."

3. Rank the shared connections you copied

Neither your archive nor LinkedAsk holds anyone else's connection list, so prompt 3 starts from the names you copied off LinkedIn and ranks them the same way as prompt 2.

These are my shared connections with [target name], copied from LinkedIn: [paste names].
Find each of them in my Connections.csv. If two connections share a name, ask me which one I mean.
Use their profile URL to find the last message between us in messages.csv, sent by either of us, and rank them by that date, most recent first.
Based on company and position, say how each one probably knows [target name].

4. Draft the request to your connection

Your archive has the full text of your messages. LinkedAsk holds message previews of about 50 characters, for people you saved, plus any note you keep on the person.

Draft a short message to [connection] asking whether they would introduce me to [target name].
Mention what we last talked about and say in one sentence why I want to meet [target name].
Make it easy to say no, and keep it under 80 words.

5. Write the note they can forward

This one needs nothing beyond your own words. A forwardable note saves your connection from writing the introduction for you.

Write a forwardable note of three or four sentences that [connection] can pass to [target name].
Say who I am and what I want to ask, and why it is worth fifteen minutes of their time.
No flattery and no attachments.

Two variations work well: a casual version for a former colleague, and a version for someone you have not talked to in years that opens with a quick catch-up.

Rank introducers on your live LinkedIn activity

Messages are one signal. LinkedAsk records more of them as you use LinkedIn, including reactions and comments on your posts, your reactions to their posts, your visits to their profile and the invitations you sent. You can connect it to your own AI, such as Claude or ChatGPT, and ask for the ranking from prompt 2 with those signals added.

The assistant can search the people LinkedAsk has seen by name or headline, so "people whose headline mentions Northwind" works without a file. For one person it returns a timeline of what happened between you, newest first, with the date of the last contact. You add the connection with the server address and a LinkedAsk sign-in. The guides for Claude and ChatGPT show where.

Pick by situation

• One target person: copy the shared connections from LinkedIn, then run prompts 3 to 5.
• A whole company: prompts 1 and 2 on your archive.
• You talk to people more through comments than messages: LinkedAsk, which adds comments and reactions on your posts to the ranking.
• Nobody you know is close to them: start a conversation of your own and keep it going with the follow-up prompts.