The Ai KhanLet's talk
← All case files

Case file 005LinkedIn outreach

The problem

“Our sales team spends the whole day on LinkedIn, and most of it isn’t selling.”

Solved

For
Teams running outbound on LinkedIn
Built
12Pilot, a live SaaS · I’m CTO & Cofounder
Stack
Web app · mobile app (Expo) · AI outreach · Stripe · MCP server
From
12Monday Technologies
Before: dusk to midnight, and the messages keep coming3D pixel film · made by me

01 · The pain

Outreach ate the day before selling could start.

Left for actual conversations: 9 h

And tomorrow, the same list again.

An illustration of a typical day, not a measurement.

  1. Find people worth messaging
  2. Check each profile
  3. Write connection notes
  4. Follow up on last week
  5. Answer replies across accounts
  6. Stay under LinkedIn’s limits
  7. Update the sheet

02 · The fix

The AI does the legwork. You keep the send button.

A model of 12Pilot’s AI Outreach with sample prospects. Open a draft, approve or skip it, and watch the safety queue spread approved messages across days.

AI Outreach · 8 drafts waiting

  • Note Hi Sana, saw the news about the new lending product. I help teams like yours run outreach without burning accounts. Worth connecting?

    Day 3 If they accept: a short thank-you and one question about the new lending product.

    Day 7 If no reply: one gentle follow-up, then stop.

0

messages sent without a human saying yes

Safety queue · sample cap 3 a day

  1. Today0/3

    ———

  2. Tomorrow0/3

    ———

  3. Day 30/3

    ———

Approve a few. Past the daily cap they roll to tomorrow instead of going out in a burst.

03 · How it works

Three jobs, one workspace, and one rule.

  1. Find the people

    Prospect Extractor pulls people from LinkedIn search and Sales Navigator; Google Map Extractor adds local businesses into the same lead database.

  2. Let the AI plan, not send

    AI Outreach drafts the sequence for each person. A human approves it. Campaigns then run through safety queues with daily caps.

  3. One inbox for the whole team

    The multi-seat Unibox puts every reply from every account in one place, with the lead’s context next to the thread.

  4. And in your pocket

    I built the mobile app too: the Unibox with AI reply suggestions, campaigns you can pause and resume, and every seat a tap away, on the same backend as the web app.

The hard part

An AI that is allowed to plan, but not to send.

Fully automatic outreach is easy to build and easy to regret: one bad message to a thousand people, or an account restricted for moving too fast. So the brake is in the product, not in a policy document.

The AI can draft every step. It cannot send a single one until a person approves, and even then the queue paces it.

AI drafts → every message, every follow-up

Human approves → then the safety queue sends, within daily caps

04 · Where it is now

Not a prototype. A product people pay for.

Status
Live
Plans from
$33/mo
My role
CTO

I also built the 12pilot.net landing page and the mobile app. Also in the product: Content Hub · API & webhooks · MCP server. See 12pilot.net