Generic AI output is a brand problem, not a model problem.
A brand profile holds what you sound like and what you look like in one place: voice traits, audience, phrases you would never use, your colours, your heading and body fonts, and the logo and product shots a design should reach for. Every draft and every generated image reads it before it writes a word.
Every draft arrives 80% right and needs the same 20% fixed.
The draft says "leverage" and "unlock" and "in today's fast-paced world". You delete those three phrases in every single post, forever, because nothing remembers you hate them.
The generated image is fine and completely off-brand. Wrong blue, wrong typeface, no logo. You regenerate it three times and then open a design tool anyway.
You run four clients. Each one has a voice, a palette, and a set of rules living in your head or in a doc nobody opens. The wrong tone reaches the wrong account on the day you are tired.
Honest fit check
For you if
- You edit the same three phrases out of every AI draft.
- You run more than one client or brand from one login.
- You generate images and want them in your palette, with your logo.
- Your tone is specific enough that generic output is worse than nothing.
Not for you if
- You want a full design system with component tokens and export pipelines. This is a palette, two fonts, and reference imagery, not a design tool.
- You want the AI to write in a voice you have never described. It reads what you give it. An empty profile produces generic output, correctly.
Say it once. Everything downstream obeys.
A brand is not a text field the AI glances at. It is the profile that shapes rewriting, captions, replies, recycled variants, and generated imagery, on every surface that writes for you.
Voice the AI actually writes against
Tone traits, what you sell, who you are talking to, examples of posts that landed, and the phrases you would never use. The rewriter pulls from all of it, and gets closer the more you correct it.
Colours and fonts a design has to obey
Save your palette and your heading and body typefaces. A generated image folds them into the prompt, so the output comes back in your colours rather than a model default.
Assets tagged, not re-uploaded
Tag a library item as logo, product shot, background, or person and it becomes reference imagery for anything generated for that brand. No separate asset store to keep in sync.
One brand per workspace
Agencies and multi-brand founders keep a workspace per client, and each workspace carries its brand. Switch the workspace and the accounts, calendar, and brand switch together.
Reaches every AI surface
Composer rewrites, caption generation, inbox reply suggestions, recycled variants, and image generation all read the same profile. You do not configure tone five times.
Consent is required for people
Tagging an asset as a person needs an explicit attestation that you have that person's permission. It is refused without one, on upload, on import, and on the edit that would promote an existing file.
From blank profile to on-brand output
Describe the brand
What it does, who it talks to, the industries it sits in, the tone traits you want, and the phrases you never want to see. Two minutes of typing, reused forever.
Add the visual identity
Drop in hex values for the palette and name a heading and a body font. Both are optional. Both make a generated design stop guessing.
Tag your assets
From the media library, tag a logo, your product shots, and background plates to this brand. A person shot needs a consent attestation before it saves.
Save it and forget it
The workspace has one brand. Everything you write or generate in it uses that profile, so being on-brand becomes the path of least effort.
Why a profile beats a prompt you paste every time.
Three ways teams try to keep AI output on-brand. Only one survives contact with a second client and a tired Friday.
A brand doc nobody opens
What most teams have
- Lives in a doc the AI cannot read.
- Enforced by whoever remembers it that day.
- Goes stale within a quarter.
Pasting a style prompt
What careful users do by hand
- Retyped or re-pasted on every single run.
- Drifts, because each paste is slightly different.
- Covers writing only. Generated images stay off-brand.
- Falls apart the moment you run a second brand.
FoPost brands
What you actually want
- One profile, read automatically by every AI surface.
- Covers writing and visuals: voice, palette, fonts, assets.
- One per workspace, so switching the client switches the brand.
- Improves as you correct it, instead of decaying.
Publishes to all major networks