One voice per account
Each connected account has its own profile. Your personal LinkedIn and your company page are different accounts and they sound different.
Voice profiles
One voice per connected account. Learned automatically.
Every social account you connect gets its own voice profile. Antwork reads the existing posts on that specific account — not from "the platform" in the abstract — and models your real voice there. Your personal LinkedIn and your company page are different. Your two X accounts are different. The AI drafts using the right profile in each case.
How it works
How your voice gets built.
- Step 01
Connect an account
As soon as you authorize an account via OAuth, Antwork reads the existing posts on that specific account and runs them through Vertex AI to extract patterns. Every account you connect triggers its own analysis.
- Step 02
A profile is built per account
Tone, sentence length, typical hooks, words you avoid, emoji or not, capitalization, in-jokes. One independent profile per account — your personal LinkedIn doesn't bleed into your company page.
- Step 03
Every draft uses the right profile
When your AI asks to post to an account, Antwork injects the voice profile for that specific account. Voice changes with the destination account, not just the platform.
What you get
Why one profile per platform changes the output.
Bias toward what works
Weights your top posts — your highest-impression hooks count more than your average post. The AI learns from your best self.
Auto-refreshes
Periodic re-fetch as you publish more. Your voice evolves; the profile evolves with it.
Visible and editable
You can read the generated profile, pin rules, ban specific words, or paste your own samples as reference.
Filters AI tells
Excessive em-dashes, "delve", "in summary", corporate filler: all filtered out unless they're in your real voice.
Multi-account
Multiple accounts on the same platform? Each account has its own profile. What one learns doesn't leak into another.
Real workflows
What you'll notice in practice.
Your LinkedIn stops sounding like your cofounder's
If both of you ask the same AI for a post about the same launch, you get different drafts: yours with your hooks, theirs with theirs.
Before
"We are thrilled to announce..."
After
"Three months of no sleep and we can finally show it."
Your tweets stop reading like compressed LinkedIn
The X profile uses your actual X cadence — short hooks, contrarian first line, no corporate.
Threads actually sounds casual
If your Threads is lower-case and conversational, drafts match. None of that "🚀 Excited to share..."
Cross-posting with native voice
Ask "post this to LinkedIn, X, and Threads". Antwork adapts the same message into three different voices of yours — not one.
FAQ
What people ask.
What if the account I connect has no posts yet?
With no existing posts on that account, there's no voice to extract. Two options: paste 3–5 reference posts manually, or (if you provided a website) Antwork falls back to your Brand DNA as the starting voice. Once you start publishing, the account's real profile takes over.
Do I need to provide a website?
No, it's optional. The website is only used to seed an initial Brand DNA, which acts as a default voice when a freshly connected account has no history yet. Skip it and voice profiles still work fine on accounts that already have posts.
Can I tweak it manually?
Yes — every profile is editable. You can add rules ("never use emoji"), paste 3–5 posts as reference, or shift tone globally.
Does my voice mix with other users' voices?
No. Each account has its own profile. The model doesn't train on your posts; it just reads them at generation time and discards the context after.
How long does it take to build?
Between 30 seconds and a couple of minutes per account, depending on the volume of existing posts. It's per account — connecting 3 accounts triggers 3 parallel analyses.