Draft a LinkedIn connect note plus two follow-up messages, with no fake flattery

What this does

You give the AI one real fact from the person’s profile and your honest reason to connect. It drafts a connect note and two follow-ups. You copy, edit, and send them yourself from LinkedIn.

Who it’s for

SDRs and AEs doing social outreach one person at a time. Not for automated outreach, bulk connecting, or scraping profiles.

What you need

• Person’s role and one fact you actually read on their public profile or post
• Why you’re reaching out, in one honest sentence
• What you’d ask for in the follow-ups
• Tool: Any AI chat tool (for example ChatGPT, Claude, Gemini, or Grok), LinkedIn (manual use only)
• Cost: $0 with a free AI chat tier (limits change)
• Setup: 3 minutes (estimate)

Steps

1. Open the person’s profile yourself. Write down one fact you really read (a post, a role change, a shared group).
2. Fill the [VARIABLES] in the prompt and run it.
3. Check the connect note length. LinkedIn’s note limit has changed over time and differs by account type; check the box in your own account before you paste.
4. Edit each draft until it sounds like you. Delete anything you didn’t actually read or don’t mean.
5. Send the connect note by hand. Wait for acceptance. Send the follow-ups by hand, spaced a few days apart, only if they connected and haven’t asked you to stop.

The prompt

Write a LinkedIn connect note and two follow-up messages for a salesperson. Use ONLY the facts I give you. Do not invent shared interests, flattery, or mutual connections.

NO-ADDITIONS RULE: Do not add facts, tips, statistics, scarcity or urgency, promises, references or testimonials, or the gender of any person that I did not supply.

Person's role and industry: [ROLE, INDUSTRY]
One thing I actually read on their profile or in a post: [REAL FACT, or "nothing specific"]
Why I'm reaching out, honestly: [ONE SENTENCE]
What I sell, one sentence: [PRODUCT]
What I'd ask for in the follow-ups: [e.g. a 15-minute call, or an opinion on a question]

Output:
1. CONNECT NOTE: 280 characters or fewer, no pitch, no link. If I said "nothing specific," don't pretend to know anything about them.
2. FOLLOW-UP 1 (after they accept): under 60 words, gives one useful point or question, no pitch.
3. FOLLOW-UP 2 (about a week later, no reply): under 60 words, makes the ask, easy to say no to.
4. NOTE: list any phrase that might sound like flattery or implies I know more than I do.

Plain words. No emoji. No "I'd love to pick your brain."

Sample input (fake data)

All names and companies are fictional. This is the input for run 4 (the output below), on the current prompt. Runs 5 and 6 used different inputs; see the test log.

Role/industry: Facilities Director at a regional hospital group
Real fact: Commented on a post last week saying their team just finished moving to a new work-order system
Reason: I sell preventive maintenance scheduling software and want to compare how teams plan maintenance after a system change
Product: preventive maintenance scheduling software
Ask: a 15-minute call

Sample output

Real output from run 4 on 2026-10-14, using the current prompt, from an unspecified AI assistant (Grok Bot agent session; model name not shown). Copied as produced. Nothing was fixed after the fact; the problems we found in it are listed under “How it went”.

1. CONNECT NOTE
Hi, I saw your comment about moving to a new work-order system. I sell preventive maintenance scheduling software and would like to connect.

2. FOLLOW-UP 1 (after they accept)
Thanks for connecting. How is planning preventive maintenance going now that the new work-order system is in? Any part you'd do differently?

3. FOLLOW-UP 2 (about a week later)
I sell preventive maintenance scheduling software and I'm interested in how teams plan maintenance after a system change. Would you be open to a 15-minute call? If the timing's wrong, no problem.

4. NOTE
- "I saw your comment" assumes you actually read it; send only if you did.
- Follow-up 1 assumes the maintenance planning changed with the new system. You didn't say it did.
- No other claims about you, other teams, or results are in these drafts.

Check before you send

• Every fact in the note is something you actually saw.
• Connect note fits the limit shown in your account.
• Nothing implies a relationship or conversation that doesn’t exist.
• Follow-ups only go to people who accepted, and stop if they say stop.

How it went

• Test date, tool, model: 2026-10-14. Unspecified AI assistant (Grok Bot agent session; the model name is not shown to the agent). One tool only.
• Runs logged: 6, on 6 different fake inputs (see the test log). Runs 1-3 used the original prompt. Runs 4-6 are a second round on new fake inputs after the prompt was fixed, as listed below. Every run was done by the same AI assistant session that wrote the fake input, and in round 2 the same session also edited the prompt, so these are not blind or independent runs. A script counted words and characters against the prompt’s limits. No second person re-ran it from the written steps.
• Why the label is “Tried in an AI tool,” not “tested”: we reserve “tested” for plays with at least 3 logged runs in at least 2 different AI tools, and we ran one tool. We also have no independent reproduction of the steps and no timed manual baseline. The runs are real, but this is one tool’s behavior, and 3 runs per round is too few to say a problem is gone.
• Time saved: Not measured. We did not time a manual baseline or an assisted run.
• What we changed after round 1: Added: no-additions rule (no facts, tips, stats, urgency, promises, references, or gender not supplied). No other change.
• Round 2 (current prompt, runs 4-6). What worked:
  • No mutual connections, shared interests, invented counts (‘a few owners’), or ‘one thing I hear’ lines appeared this round. The first round had at least three of those.
  • Lengths: connect notes 140-142 characters (limit 280), follow-ups 16-34 words (limit 60), counted by script.
  • The NOTE section caught the two soft claims the model still wrote (runs 5 and 6).
• Round 2: what failed or needed fixing:
  • Run 5 connect note still implies other conversations (‘I’m talking with finance leaders in food distribution’). The rule is general and the model slipped on this specific phrasing; the NOTE flagged it but it is still in the draft.
  • Run 6 ‘I work with schools’ stretches the input. Also flagged, not removed.
  • Follow-up 1 in runs 4 and 6 is a question built on a mild assumption (planning changed; open-house follow-up is a pain). Flagged in run 4, not run 6.
• Checked against “Check before you send”: ‘Every fact is something you actually saw’ needed small edits in runs 5 and 6 and none in run 4; lengths fit; no implied relationship.
• Edge cases: ‘Nothing specific’ input (run 5) stayed honest about the person but not fully about the sender. We still did not check LinkedIn’s current note limit or use LinkedIn; nothing was sent.
• Round 1 (original prompt, runs 1-3): what failed, kept for reference:
  • Run 1 follow-up 1 asserts ‘the hard part of seasonal hiring is often the first two weeks, when new people don’t know the routes yet’ as something the sender hears. The sender gave no such fact. The NOTE flagged the phrase ‘one thing I hear’ but the claim itself would still read as the sender’s experience.
  • Run 2 connect note: ‘follow your work in this space’ implies the sender has seen their work. Flagged in NOTE, but the draft still contained it.
  • Run 1 ‘I’ve talked with a few landscaping owners’: the count is the model’s wording.
  • Follow-up 1 in runs 2 and 3 is a question built on an assumption (field-to-office data gap; compliance training) that the input didn’t state.
• Not yet observed: how a different model or tool handles the same prompt (a second-tool run is still to do).

Watch-outs

LinkedIn’s terms prohibit scraping and unapproved automation tools (bots, auto-connect, auto-message). This play only drafts text. You type or paste and send each message by hand. Don’t paste scraped profile data. Not legal advice; read LinkedIn’s current User Agreement.

See the Safe Use page: AI drafts, you decide, you send.

Variations

• Owner version: connecting with local business owners or referral partners; ask for a coffee or a referral rule of thumb.
• Manager version: ask for a version for hiring or partnering, not selling.
• Event version: add ‘we both attended [EVENT]’ only if true.

Sources

None. Original wording. (Add any public framework or post used, with link and checked-on date.)

Retest

Re-check by: 2026-12-13 (60 days after the 2026-10-14 runs), or sooner when run in a second tool.

Test log

Changelog

• 2026-10-14: Draft written. Not run yet.
• 2026-10-14: Ran 3 times in one AI assistant on fake data. Filled in sample input and output, how it went, and the test log. Front matter set to runs_logged 3, last_tested 2026-10-14, retest_by 2026-12-13. The 2-tool bar for the word “tested” is not met, so the play is not labeled tested. The prompt text was not changed after the runs.
• 2026-10-14: Prompt revised after round 1. Added: no-additions rule (no facts, tips, stats, urgency, promises, references, or gender not supplied). No other change. Re-ran 3 fresh runs (4-6) on new fake inputs in the same AI assistant. Updated the sample input and output to run 4, the “How it went” section, and the test log (original runs 1-3 kept, marked). Front matter set to runs_logged 6, last_tested 2026-10-14, retest_by 2026-12-13. The 2-tool bar for the word “tested” is not met, so the play is not labeled tested.
• 2026-10-15: Published with the evidence label “Tried in an AI tool” (6 runs, 1 tool, fictional inputs). Earlier wording that implied more evidence than one tool’s runs was replaced with the label; the runs, results, and limits above are unchanged.