Product · Write

AI that writes like you, not like AI.

Everyone has the same models now. The difference is what you point them at. Most tools prompt a general-purpose model with “write a LinkedIn post about X”, which is why the feed reads the way it does. Each account here gets a voice profile built from what that person has already published, and it gets sharper every time you edit a draft.

A voice profile per accountLearns from your editsFlags its own AI tells

Used by leaders from

★★★★★4.8/5on G2LinkedIn® Verified Partner
GoogleAppleMetaNVIDIACiscoSAPAccentureGoldman SachsAmerican ExpressHSBC
The problem with AI drafts

Everyone can smell it now

Two years ago an AI-written post passed. Now readers spot it in the first line and discount everything after it, which makes a bad draft worse than no post at all.

The same rhythm every time

Three-beat sentences, a rhetorical question, a tidy lesson. It is not wrong, it is just identical to the four hundred other posts published that morning.

Vocabulary that gives it away

Delve, tapestry, testament, navigate the landscape. Words almost nobody used in 2022 and everybody’s drafts use now.

No specifics, because it has none

A model can produce the shape of a story but not the number, the date or the thing someone actually said. Those are the parts people remember, and they have to come from you.

How the voice profile works

Trained on one person, not on LinkedIn

01

It reads what they published

Sentence length, how they open, whether they use lists, how formal they are, the words they reach for. Built per account, from real posts rather than a questionnaire.

02

You give it the substance

A voice note, a rough paragraph, a link, a bullet. The model shapes what you give it; it does not invent the facts, and it will not pretend to.

03

Your edits teach it

Every change you make is a signal. If you always cut the rhetorical question, it stops writing them. The first fortnight is the worst it will ever be.

04

It marks its own tells

The same checks as our free post roaster run inside the editor, so the phrases that read as machine-written get flagged before anyone else sees them.

Ten people, ten voices

Why one voice profile is not enough

This is where most tools quietly fall over, and it is the whole reason we built ours the way we did.

One shared voice

Everything sounds like the tool

A single tone setting across every account means the engineer, the founder and the recruiter all post in the same register. Anyone following two of your clients notices, and it is the fastest way for a ghostwriter to get found out.

A profile per account

The drafts do not match each other

Each connected account carries its own profile, trained separately. Give the same brief to two accounts and you get two genuinely different posts, which is the point.

What this does not do: make you interesting. It reproduces how you say things, not what you have to say. If the input is a generic thought, the output is a generic post that sounds like you wrote it, and that is arguably worse. The raw material still has to come from a person who noticed something.

Questions

About AI-written LinkedIn posts

Will people be able to tell it was AI-assisted?+
Not if you edit it, and you should. The profile handles rhythm and vocabulary, which is most of what gives a draft away, but the specifics are yours. A post with a real number and a real name in it does not read as generated regardless of how it started.
Does LinkedIn penalise AI-written posts?+
There is no evidence of a detection penalty and LinkedIn has never claimed one. The penalty is from readers, not the algorithm: a post that sounds generated gets scrolled past, and that is a reach problem by a different route.
How many posts does it need to learn a voice?+
Around ten gives it something useful. Fewer than that and it falls back on safer, blander patterns, which is exactly what you do not want. If someone has never posted, start by writing three yourself and let it learn from those.
Can I turn the AI off entirely?+
Yes. Plenty of people use the calendar, approvals and reporting and write every word themselves. Nothing degrades if you never open the drafting side.
Is my content used to train a shared model?+
No. A voice profile belongs to its account and is not pooled across customers. Your client’s writing does not end up shaping a stranger’s drafts.

Give it ten of your posts and judge it

Fourteen days free, no card. Connect one account, let it read what you have written, and see whether the first draft sounds like you.

From $19 a month for one account. Teammates and approvers are free.