Everyone can tell when they have received a template. "I'd love to add you to my professional network" is not a message, it is wallpaper, and people treat it accordingly.
Linkedroid can write a different connection note for every single profile, based on what is actually on that person's profile. This guide covers how to set it up, what the presets do, and the one setting most people get wrong. Written against version 3.7.0.
Where it lives
AI personalization is not on the campaign screen. It is in Preferences, under the Personalization tab. Open it from the gear icon in the top right, or from Settings in the sidebar.
This catches people out, because it means the setting is global rather than per campaign. Change it here and every campaign that sends connection requests uses the new prompt.
Turn on Personalized Messages
The first toggle is Personalized Messages, described as "Add personalized message to connection requests". With it off, Linkedroid sends bare connection requests with no note at all.
Bare requests are not a bad thing, incidentally. They perform fine, and they never sound wrong. If you are not going to write something worth reading, sending nothing beats sending a template.
With the toggle on, two buttons appear: Regular and AI-Powered.
Regular is a fixed message with variables. Same text for everyone, with their name dropped in.
AI-Powered generates a fresh note per profile from that person's actual data.
The four presets
Pick a Preset fills in a prompt and a sensible set of profile fields. There are four, and they are meaningfully different, not just tone variations.
Sales Outreach
Prompt: write a short LinkedIn connection request note, connecting because you have a product or service that could help them based on their role, under 300 characters, casual and professional. Uses first name, job title and company.
Honest about intent. Use it when you actually are selling, because the alternative is opening with a friendly note and pitching two messages later, which people find worse.
Networking
Prompt: write a friendly LinkedIn connection note expressing genuine interest in their work and expertise, warm, brief, under 300 characters. Uses first name and job title.
The default, and the right one for most people. Notice it deliberately uses fewer fields.
Recruiting
For reaching candidates. Uses first name, job title and company.
General
Prompt: write a short, friendly connection note mentioning something relevant about their profile and expressing interest in connecting, under 300 characters. Uses first name and job title.
The safe fallback when you are connecting for no single reason.
Writing your own prompt
The AI Prompt box is fully editable. You are writing instructions for a writer who has the person's profile in front of them, so the useful things to specify are constraints and things to avoid.
What works:
- Give it a length. Every preset says "under 300 characters" for a reason. LinkedIn caps connection notes, and a note that gets truncated mid-sentence reads worse than a short one.
- Tell it what not to do. "Do not compliment their profile" and "do not use the word excited" remove the tells that make AI writing obvious.
- Give it a reason to be connecting. The model cannot invent your actual motive. If you say you run a logistics consultancy and want to talk to operations leaders, the notes get specific.
What does not work: asking for something the profile does not contain. "Mention a recent post of theirs" will produce a vague sentence, because the prompt only receives the profile fields you select below.
Profile Fields to Include, and why fewer is better
Six checkboxes: First Name, Last Name, Job Title, Company, Location, Full Name. By default the first five are on.
The instinct is to tick everything. Resist it.
Every field you add is another detail the AI feels obliged to work in. Tick Location and you start getting notes that mention the weather in Atlanta. Tick Company and Job Title and Last Name and you get a sentence that recites the person's business card back at them, which reads exactly as automated as the template you were trying to escape.
The Networking preset uses two fields: first name and job title. That is enough for a note that sounds like a person wrote it after glancing at a profile, which is the honest target.
Start with the preset defaults. Only add a field if you can say what you want the AI to do with it.
What it actually produces
Here is what lands, seen from LinkedIn's own Manage Invitations screen after a campaign run.
The shape is: greeting by first name, a specific reference to their role, then a reason to connect. Different for every recipient, because it is written from their profile rather than filled into a slot.
Check this screen after your first AI campaign. It is the only place you see exactly what went out.
The quota nobody mentions
LinkedIn limits how many connection requests with a note you can send. Free accounts get a monthly allowance. Once it runs out, LinkedIn will not attach your note.
Linkedroid handles this rather than failing: when the personalized-invite quota hits zero, it sends the request without a note instead of stopping the campaign. Premium accounts reporting unlimited notes are handled too.
So if you turn on AI personalization and later find plain requests going out, you have most likely used up the month's allowance. Nothing is broken.
Review before you scale
The failure mode with AI notes is not that they are bad. It is that they are subtly wrong in the same way a hundred times, and you do not notice until someone tells you.
So: run a small campaign, 20 or 30 profiles. Open Sent invitations. Read every note. You are looking for a phrase that repeats, a tone that misses, a field being used oddly.
Adjust the prompt, run another small batch, read those. Once two batches in a row read well, scale up.
What to read next
AI notes also work inside drip campaigns, where the Send Message node uses the same preset and prompt system. That is the next guide, and it is where personalization gets genuinely useful, because a follow-up that references the first message is far harder to template.
Questions to help@linkedroid.com.
Comments
Post a Comment