Feature · Voice profile

Replies that sound like you wrote them at 2am.

The voice profile is the single thing that turns a generic AI reply into something your audience can't tell from your own typing. Persona, formality, emoji, length, samples, signature moves — every knob is tunable, per account.

The voice knobs

Six controls. Per account.

Most AI tools give you a 'tone' dropdown. We give you the actual mechanism — the parts of voice that make followers say 'wait, that's not a bot?'

V:01
P

Persona

Pick a starting point: brand voice, founder, expert, creator, mascot, pet POV, or fully custom. Each persona ships with a sensible default.

Starting frame
V:02

Formality slider

1–5, very casual to formal. The slider tunes vocabulary, contractions, and sentence rhythm without changing the underlying tone.

Per-account
V:03

Emoji habits

Single emoji at the end? Three-emoji stack? None at all? Set the rule once. The AI follows it consistently across every reply.

On-brand
V:04

Length target

Average word count for replies. Short and punchy for high-volume creators, longer for service businesses where detail matters.

Volume-aware
V:05

Sample replies

Paste 5–10 of your real past replies. The AI anchors to your sentence shape, not a generic 'casual but professional' prompt.

Voice anchor
V:06

Signature moves + sign-offs

Phrases or rituals you always use — 'sending a DM with the link 💌' or '— jess'. They appear at the right moments, never forced.

Brand consistency
Knowledge layer

The facts behind your voice.

Voice without facts produces vibey nonsense. Facts without voice sound like a help-centre. The knowledge layer keeps both anchored.

K:01

Business context

Free-form notes about your offer, products, pricing rules, location, hours — everything the AI needs to answer the recurring 'where, how much, when' questions.

Account-wide
K:02

Industry preset

Drop in a starting context for retail, coaching, real estate, healthcare, travel, or education. Edit from there, don't start from blank.

Head-start
K:03

Hard don'ts

Phrases or claims the AI must never make. Pricing speculation, medical promises, refund commitments — anything outside your authority.

Brand safety
A practical recipe

What to actually put in.

Persona

Pick the closest match. If you're the face of the brand, use Founder. If it's a brand voice that's not really you, use Brand. Custom is for edge cases — most accounts are fine on a preset.

Samples

Find 5–10 past replies you actually wrote. Pick a mix — a price answer, a thanks reply, a longer explanation, a sassy clapback if that's you. Avoid replies that were super specific to one event.

Signature moves

What do you always say? '🤍' at the end? 'check the link in bio'? 'gonna DM you'? Put them in. They become part of the voice without feeling forced.

Hard don'ts

What can't the AI promise? Be specific. 'Never quote shipping times. Never say a product is in stock. Never confirm refund eligibility.' These run as filters, not vibes.

FAQ

Common questions about voice.

How much do voice samples actually matter? +
A lot. The persona slider gets you to 'recognisably on-brand'. The samples get you to 'someone who knows you can't tell which replies are AI'. Plan to drop in 5–10 of your most representative past replies during onboarding — it's the single highest-leverage thing you can do.
Does the voice profile improve over time? +
Yes. Every reply you approve in the queue, and every edit you make to a draft, becomes part of the voice corpus the AI references. Three weeks in, the queue feels like reading your own DMs.
Can I have different voices per account? +
Absolutely — and you should. Voice profiles are per-account by design. Your founder voice on one brand and a playful pet-POV voice on another don't bleed into each other.
Can I override voice for a single post? +
Per-post tone overrides are on the roadmap. For now, voice is per-account. The good news: per-post context already shifts the AI's mood — a serious announcement post gets a serious response style without any extra config.

Get the AI to sound like you.