Skip to content
Imagine
Book a demo

Find the buyers hiding in your connection requests

Once a week, have the agent sort your pending connection requests against your ICP and your CRM. Accept the buyers and send each a short hello that asks what made them reach out, and leave the recruiters and vendors for later. People who ask to connect with you are the warmest list you have. The Imagine AI agent sorts the invites and drafts each message for approval.

Last updated
Oct 20, 2026
Written by
Sky Yang, Co-founder and CEO
How we wrote this
Written from how the Imagine AI agent runs this play on LinkedIn, with approval.
Book a DemoSee the free benchmark
When it runs

Connection requests are piling up in your inbox.

One prompt. The agent runs the play.

The agent has its own computer and browser, so it finds the people and the news itself. It drafts every comment, connection note and message, and you approve what goes out. Use it when:

  • You have dozens of pending invites you never get to.
  • Your posts bring in connection requests from strangers.
  • You can't tell buyers from recruiters at a glance.
Pending invites
Fig. 01: Inbound invites triage, run by the agent. You approve before anything goes out.

The skills behind this play.

The agent has its own computer and browser, so it does the legwork itself. Every skill is on for every play; these are the ones this play leans on.

[01]Used in this play

Writes in each person's voice

Drafts for the company page, founders, executives and employees from each person's best posts. Imagine AI started as the operating system for ghostwriting teams, so sounding like you is the core of the product.

[02]Used in this play

Engages for you

Comments, connection requests and messages on LinkedIn, aimed at the people your posts pull in who match your ICP.

[03]Used in this play

Finds the buyers in your audience

Matches everyone who engages to your ICP and hands sales a worked list every week.

[04]Used in this play

Maps content to revenue

Connects to HubSpot or Salesforce and traces each closed deal back to its first content touch: post, person, deal stage moves.

[05]On

Designs the graphics

Multi-image posts, infographics and visuals from your brand kit, made in its own sandbox, with the right people and companies tagged.

[06]On

Has its own computer and browser

It does the legwork itself: finds the people, checks job changes, launches and funding news, reads the sources and fact-checks claims before anything ships.

[07]On

Runs one calendar for the team

Plan launches, sequence posts, schedule comments and edit together. Approvals sit with each person or route through one owner.

[08]On

Works from your content filesystem

Positioning, brand kit, decks, campaigns and compliance rules in one place it reads before every draft, remembered across threads.

[09]On

Turns what you have into posts

Events, podcast episodes, videos, decks and call notes become a week of posts across the team.

[10]On

Knows what works

Reads the public benchmark of 1,000+ B2B companies and tracks your competitors, so post types and formats follow your industry's data.

[11]On

Plugs into your stack

Engager data over API and webhooks, and the benchmark over MCP for your own agents.

Step by step

The playbook lives as a file the agent reads. Edit it once and every run follows it.

What you need
  • Your ICP.
  • HubSpot or Salesforce, to flag open deals and customers.
  • One line on what you'd want to know from a buyer who reached out.
Files/Imagine AISend to agent

inbound-invites-triage.md

Save

Step 1 · Sort the invites

  • Pending requests matched against your ICP and CRM: buyers, open deals, customers, and everyone else.

Step 2 · Accept the buyers

  • You accept the ones worth a conversation from the sorted list.

Step 3 · Say hello with a question

  • A short first message asking what made them reach out, or offering something specific to them.

Step 4 · Route the deals

  • Anyone on an open deal goes to the owner instead of getting a new thread from you.

Fig. 02: The playbook as a file in your workspace.

Anatomy of the hello

The outline, what to avoid and how to measure it, in the file the agent drafts from.

Files/Imagine AISend to agent

inbound-hello.md

Save

Outline

  • 1Thanks for connecting.
  • 2One line that shows you looked at who they are.
  • 3One question: what made them reach out?

Example from the run above

  • Thanks for connecting, Sarah! Saw you've been reading the benchmark posts. What made you reach out?
  • Hey Marcus, good to connect. Are you looking at LinkedIn for the growth team or more for yourself?
  • Thanks Priya! You're at one of the companies we follow in the benchmark. Happy to share your numbers if useful.

Avoid

  • Auto-pitching everyone who connects.
  • Leaving buyers in pending for weeks.
  • Missing that the person is already on a deal.

Measure

  • ICP invites per week.
  • Replies to the hello.
  • Invites that turn into a call.

Fig. 03: The outline, part by part.

Join the high-growth companies already building with Imagine AI

Questions teams ask first.