How do you let AI run your LinkedIn account?
For the past month I haven't been running my own LinkedIn — AI writes the content, publishes it, and I just approve. I do it with Postuby, which is software we built ourselves. There are other tools that do this and they work on a similar logic; I'm walking you through ours because that's what I use. Follow the steps below and setup takes 15 minutes — your posts start arriving the very next morning.
What you'll need
- A LinkedIn account (personal profile or company page)
- A Postuby account — sign up at postuby.com
- 3-5 of your own best past posts (optional, but the single biggest quality lever)
- A few concrete numbers/facts about your work (so the AI doesn't invent any)
Step by step
1. Sign in to Postuby
Go to postuby.com, create your account and open the dashboard. I'd do all of this on a computer — you'll be copying and pasting text.
2. Connect your LinkedIn account
In the dashboard go to Settings → Channels and connect your LinkedIn account. LinkedIn's own permission screen opens, you approve it, and your account appears in the channel list. Skip this and the wizard won't let you continue.
Company pages work the same way — just watch which profile/page you authorise while connecting.
3. Open the LinkedIn wizard
Open the "LinkedIn Autonomous Content" wizard from the dashboard. You'll get a 4-step setup: mode, channel, your voice, and strategy. We'll fill them in one by one.
4. Step 1 — Pick the mode and posting times
There are two modes. Autonomous: every morning it scans the news, writes fresh posts for the day and drops them in for your approval — a continuous flow, and the one I use. One-off plan: it generates 7 or 15 days of content in one go. In autonomous mode you choose how many posts a day you want (I use 3), how many of those should react to live news, and the publishing times — up to 4 slots.
Set the times to when your audience is actually on LinkedIn. The defaults — 08:30 / 12:30 / 17:30 — work well.
5. Step 2 — Choose the channel
Select the LinkedIn channel you just connected. The brand field is optional; if you pick one, that brand's sector, tone and identity feed into the content too.
6. Step 3 — Teach it your voice (the most important step)
Three fields. "About you": a few sentences on what you do and what you stand for — this is where the system's voice comes from, and it's required. "Your best 3-5 posts": paste real posts you wrote, separated by a line with three dashes (---); the AI copies your sentence patterns from these. "Concrete data / facts": write your real numbers (how many clients, how many years, which product) — this field is what stops it from inventing things.
The 5 minutes you spend here decide the quality of the next month. Leave it empty and the output turns into generic AI-flavoured filler.
7. Step 4 — Set the strategy
Pick your objective (personal brand / B2B leads / hiring / awareness), choose the format (plain text, image post or PDF carousel — you can mix), and list your content pillars separated by commas (e.g. AI, entrepreneurship, social media). Then choose which topics get scanned for news each morning (ready-made ones like AI, Social Platforms, Marketing, Tech & Startup; you can add your own keywords). Finally set the tone (professional / warm / witty / bold / energetic), hashtag preference, language, and account voice: first-person "I" for a personal profile, "we" for a company account.
8. Generate the plan
The moment you hit generate, the system starts working and drops you into the content editor. Posts are written one by one; if you chose image or carousel formats the visuals get generated too, so give it a few minutes.
9. Approve — the rest is automatic
Every generated post lands in your approval queue first. You read it, edit if needed, and hit "Approve". Approved posts go out to LinkedIn on their own at the times you picked — the system checks every minute and publishes whatever is due. Nothing you haven't approved ever gets posted.
In autonomous mode a new batch is ready the next morning; your daily job is just reading and approving. That's literally all I've done for a month.
Tips
- Don't skip pasting your real posts — it's the single biggest quality lever.
- Give it your real numbers. Without data, the model fills the gap with vague sentences.
- Read and edit carefully the first week; the system settles into your language over time.
- Keep news topics narrow. Too broad and you get posts reacting to irrelevant headlines.
- Three posts a day is a good start: one reacting to the news, two in your own voice.
- Running a company account? Set the voice to "Corporate / Brand", otherwise it writes as "I".
Frequently asked
Is it fully automatic — do I never touch it?
Writing and publishing are automatic; approval stays with you. Posts land in your approval queue first and nothing goes to LinkedIn without your approval. I spend a couple of minutes a day reading and approving — that's it.
Will the AI post made-up information?
The "Concrete data / facts" field in the wizard exists exactly for this — once you put your own numbers there, the model stays inside them. Still, don't approve blindly in the first weeks.
Is it LinkedIn only?
No, Postuby handles other platforms too. But this guide covers the LinkedIn autonomous flow, which is the most refined one.
Does it write in Turkish?
You pick the language in the wizard's strategy step — Turkish, English, German, French and Italian are available.
Other tools do this too — why Postuby?
True, other companies do this. Postuby is software we built ourselves and it's what I run on my own account — so what I'm describing here is the exact setup I've had running for a month.
What changed in a month?
The biggest difference is consistency: I used to go weeks without posting when work got busy, now I'm out there every day. I share the numbers in the video on Instagram.