linkedbost

LinkedIn lead generation, run by your agent

LinkedIn lead generation through an MCP server means the agent searches your own seat, imports your existing lists as suppression, researches each person, drafts messages you approve, and paces sending inside enforced caps. You review; it executes. No shadow database and no bought lists.

Who this is for: Founders and small sales teams who need pipeline from LinkedIn without hiring an SDR or risking the account.

Why this is worth automating

How to run it, step by step

You type the sentence. The agent picks the tools.

  1. 1

    “Import this CSV as a lead group and tell me who I have already spoken to.”

    Your existing exports become suppression memory on day one, so nobody gets a second first-touch.

  2. 2

    “Search my LinkedIn for second-degree heads of growth at Series A fintechs in the US.”

    Search runs through your own seat and entitlement. Results stay a preview until you save them into a group.

  3. 3

    “Add a column for funding stage and what each person last posted about.”

    Researched facts become persistent columns rather than something you remember once.

  4. 4

    “Draft a three-step sequence from my seat and show me every draft before anything sends.”

    Creating a sequence sends nothing. Enrollment queues drafts behind your approval.

  5. 5

    “Who replied this week, and which sequence is underperforming?”

    Reporting becomes a question rather than a weekly ritual.

The numbers

Invitations per day20Enforced server-side on a new seat
Invitations per week80Our enforced ceiling. LinkedIn publishes no number
Our own acceptance rate41%Across 861 conversations over three years
Targeting spread we measured32% vs 2.6%Two audiences, comparable copy

What goes wrong

Buying a list

Bought lists collapse acceptance rate, and acceptance rate is the signal that protects the account. Import your own history instead.

Scaling volume before targeting

A tight list at 20 a day beats a broad list at 40, because the broad one spends acceptance on people who were never going to reply.

Treating pending as rejected

An invitation sent yesterday is not a no. Hold recent sends separately or your acceptance number lies to you.

Questions people actually ask

Can an AI agent generate LinkedIn leads for me?

Yes. Connected through an MCP server it can search your seat, import and dedupe lists, research each person, and draft outreach. Every send still passes your approval, and pacing is enforced by the server rather than suggested.

Do I need Sales Navigator for LinkedIn lead generation?

No. Search runs on whatever entitlement your seat already has. Sales Navigator adds filters and InMail credits but does not raise the invitation limit.

Where do the leads come from?

Your own LinkedIn seat and your own CSV imports. There is no shadow database and no purchased data.

How many leads can I contact per day safely?

LinkedIn publishes no invitation figure at all, daily or weekly. We enforce 20 a day and 80 a week on a new seat, and treat acceptance rate as the real constraint.

Run this from Claude

Connect the LinkedBoost MCP server and say the sentences above. Every send waits for your yes, and the caps are held by the server rather than typed into a settings field.

Related use cases