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.

Voice traits, audience, and banned phrases the AI writes against
Colours and fonts a generated design has to obey
Logo, product shots, and backgrounds tagged from the media library
One brand per workspace. Switch the workspace and everything follows.
The cost of writing without one

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.

Is this for you?

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.
What's inside

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.

How it works

From blank profile to on-brand output

01

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.

02

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.

03

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.

04

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.

vs the alternatives

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

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

Frequently asked questions

What is the difference between a brand and a workspace?

A workspace is the container: its own connected accounts, calendar, media, team, and API keys. Its brand is the identity the AI writes and designs in. Each workspace has exactly one, so a second brand is a second workspace.

Does the AI need a brand to work?

No. Every AI surface works without one and produces reasonable, generic output. A brand is what makes the output sound like you specifically. It is the difference between a draft you post and a draft you rewrite.

Do colours and fonts affect the writing?

No. Colours and fonts only reach image and video generation, where they are folded into the prompt along with your tagged logo and product shots. The written side reads the voice fields.

Can I have more than one brand?

Yes, one per workspace. Create a workspace for each brand or client, and everything in it, from connected accounts to AI drafts, follows that brand without you picking it each time.

Why does tagging a person need consent?

Because generating imagery from someone's likeness without their permission is not a grey area. Tagging an asset with the person role requires an explicit attestation that you have that permission, and the request is refused without it. Responses only ever expose whether consent exists, never who attested or when.

What happens to a brand when I delete it?

It is soft-deleted, so posts and generated assets that referenced it keep their history. Assets tagged to it stay in the media library untouched. If it was the workspace default, pick a new one.

Can the agent use a brand?

Yes. Ask the FoPost Agent for a campaign in a named brand and it writes and designs against that profile, using the same tools the dashboard does under the same permissions.